This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Analysis and Design Considerations of a

Contactless Magnetic Plug for Charging Electric
Vehicles Directly from the Medium Voltage DC
Grid with Arc Flash Mitigation
Richard B. Beddingfield, Member, Suvendu Samanta, Member, Mark S. Nations, Student Member,
Isaac Wong, Student Member, Paul R. Ohodnici Jr. Member, Subhashish Bhattacharya, Member

Abstract—Electric vehicle charging has shifted to higher as CHAdeMO 2.0 reports a charging power level of 400kW
voltages to achieve higher power for more rapid charging which can recharge a depleted battery bank up to 80% in 15
capabilities. This paper provides a contactless magnetic minutes or less, depending on the specific battery capacity
plug solution that enables medium voltage grid connec-
tions for electric vehicle charging to achieve 3.5 kVDC [6], [7]. Clearly, commercial charging stations with several
to 400 VDC , 150 kW rapid charging capabilities. This fast charging or extreme fast charging ports need medium
novel magnetic plug improves upon existing electric ve- voltage (MV) supplies. The state-of-the-art dc fast charging
hicle charging solutions by guaranteeing safe operation stations are commonly implemented with a charging plug
and connection through galvanic and physical separation attached to the charging station such as DELTA’s 150 kW
from the medium voltage side. It achieves this with a
gap and barrier in the transformer core. We introduce a Ultra Fast Charger or with the plugs and stations separated like
unique, asymmetry in the core to localize parasitic ca- ABB’s 175 kW Terra HP [8], [9]. In either implementation, a
pacitance, fully separating the medium and low voltage line frequency transformer is required for connecting the MV
regions. This approach eliminates arcing risk and allows to charging stations low voltage (LV) outputs. As a result,
rapid charging capabilities to be delivered to the general the system needs a larger space which is a serious concern,
public. This gapped core constitutes the plug action of
our proposed charging system. We present solutions for especially in urban areas where land costs are high. Also, other
the unique challenges of this solution through a detailed issues of this conventional system include higher cost and lack
analysis of the magnetic design. We confirm this analysis of modularity.
in finite element analysis and experimentation. The solution In [10]–[12], a transformer-less power converter solution is
is verified through a scaled laboratory prototype of 20 kW, provided, where several converters are connected in an input-
1 kVDC to 50 VDC that is representative of the proposed 150
kW design. We demonstrate safe, arc free, disconnection series output-parallel configuration. These systems have high
in included active content, a new solution for high power, modularity, and good high voltage and current sharing between
electric vehicle rapid charging. the modules. However, these approaches require complex
Index Terms—battery chargers, safety, transformers
controls for dynamic power, voltage, and current sharing and a
significant number of magnetic and switch components which
increases size, weight, cost and unreliability.
I. I NTRODUCTION Others present solutions with front-end converter devices
using MV SiC semiconductor devices [13], [14]. In [13], a 6
T RANSPORTATION electrification is a potential solution
for both the growing environmental concerns due to
greenhouse gas emission from internal combustion engines,
kV SiC MOSFET front end converts 4.16 kVAC to 8 kVDC .
Similarly, [14] uses the Wolfspeed 10 kV SiC devices to
convert power from 12.47 kVAC to 850 VDC . These proposals
and for the limited stock of fossil fuel. Electric and Plug-
eliminate the line frequency transformer, while utilizing stan-
in Hybrid Electric Vehicles (EVs/PHEVs) will represent a
dard control techniques for maintaining input power factor and
significant component of this electrification [1]. To meet
for various loading conditions.
demand, EV charging stations have continued to grow both in
Recently, contactless power transfer technologies have be-
power and output voltage levels. Fast charging levels must be
come popular with the availability of wide-bandgap (WBG)
greater than 20 kW at a 400 VDC , while extreme fast chargers
devices [15]. While [10]–[14] propose MV EV charging
can be greater than 300 kW with 800 VDC outputs [2]–[4].
solutions focusing on circuit topologies and control schemes,
For example, Tesla supercharger takes about 32 minutes to
the safety concern of bringing MV to EV stations is yet to
charge the battery of Tesla Model S with a charging power of
be addressed. The objective of this work is to present an
120 kW [5]. Furthermore, ultra or extreme fast chargers, such
extension of contactless power transfer technology to develop
Richard is funded by the Oak Ridge Institute of Science Education. an intrinsically safe MV to LV magnetic plug. This proposed
This research is funded by ARPA-E under the program CIRCUITS. system provides a solid barrier insulating all MV from the LV

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

domains thus addressing all MV safety concerns. This barrier Resonant Barrier
Iin Io
Tank
enables a contactless magnetic plug that increases reliability C1
S1 S3 S5
with arc free disconnects. Electric arc incidents are prevented Ip Cr Ll RP
IS
in the disconnected process by the physics of separating the Co Vo
VDC Vp Lm Rm Ro
magnetic core which collapses the magnetic and electric fields Vinv Vs
on the low voltage side while the high potentials remain behind S2 C2 S4 S6

a barrier. This enables uninterrupted operation as the plug may Half Bridge Rectifier and Load
safely disconnect from the energized MV bus to allow no- Inverter Split-core Transformer (Modelled at R L)
power maintenance on the low voltage side without the need Fig. 1: Proposed circuit for contactless MVDC Rapid charger.
for MV switch gear.
This paper presents a novel magnetic plug to achieve safe
MV connected LV charging of EV and PHEVs. The details t0 t1 t 2 t3 t5 t6 t 7 t9
of the symmetric inverter synchronous rectifier are beyond
S1 S2
scope but discussed in other work such as, [16]. We present t
solution for a rapid charger that is 150 kW with 7-8 kV vinv S2
DC input, rectified 4.16 kVAC , and a 400 VDC output. A
ipri S1
half-bridge inverter, II-A, provides a 1/2 MV step down and 0.5vin

medium frequency (MF) excitation for the high power medium iS1 iS1=iin
vS1 0.5vin
frequency transformer (HPMFT). A series resonant tank, II-B,
is used on the MV side to compensate the reactive power due
to the increased leakage inductance from the large physical
space between the MV and LV windings. The output of the Fig. 2: Waveforms for contactless MVDC Rapid charger. Gate
HPMFT transformer is rectified with a full bridge synchronous (top), Inverter (middle), Switch (bottom).
rectifier, II-C. Although, the operating principle of our pro-
posed HPMFT is similar to conventional transformers, there A. Inverter
are unique features that require careful design and consider- Selection of the inverter topology should provide soft-
ation. As such, analytical models are presented in detail in switching with low conduction loss in the load range. It should
III. We propose a geometrically asymmetric magnetic core also be highly reliable and controllable. Finally, if the inverter
for the HPMFT design to realize both contactless MV to LV has a low voltage gain, the transformer design is simplified
conversion and mechanical plug / unplug action. We use the by reducing the required voltage ratio. A symmetrical half-
presented models to perform parametric optimization for our bridge converter is a good solution to meet these requirements
HPMFT plug with COMSOL FEA verification. We explore and provide a low voltage gain. It provides a maximum
our solution with a 1/8 scale prototype of 20 kW, 1 kVDC to gain of 0.5, and has high reliability with only two switching
50 VDC DC-DC converter IV. The HPMFT performance is devices with low current requirements TABLE I. The primary
discussed in V. current, Ip is sensitive to the transformer design and therefore
II. C ONVERTER TOPOLOGY derived further in the paper in (13). While needing two more
Either non-resonant converters or resonant converters con- capacitors than an asymmetrical half-bridge, the symmetrical
verters can safely realize the power conversion directly from structure supplies balanced AC voltage to the transformer. This
7 - 8 kVDC MV to applicable 400 VDC EV charging at reduces magnetic flux saturation risk and reduces the resonant
the power levels desired, at or above 150 kW . The trans- tank voltage stress. An asymmetric inverter tank must block
former parameters can impact the selection of the converter. significant DC voltage. Fig. 2 shows the significant waveforms
A contactless transformer is chosen to provide good isolation of the half bridge inverter. During t0 −t1 , the inverter switch S1
between the high and low voltage side of the system with conducts and the active power transfer occurs from the input to
a physical barrier and is discussed in detail in III. Since output. At t1 the gate pulse of S1 is withdrawn transferring Ip
a contactless transformer has higher leakage impedance, a the S2 body diode. At this point, the inverter voltage polarity
resonant converter is chosen for the DC-DC topology, Fig. changes before the current due to the lagging power factor.
1. The following sections discuss the details of the inverter, Clearly, S1 experiences a hard turn-off. The S1 and S2 dead-
II-A, a resonant tank, II-B, and rectifier, II-C. If we neglect time occurs during t1 −t2 . S2 is triggered at t2 and experiences
the switch on resistance and primary winding resistance, we a zero voltage switching at turn-on because the body diode is
may may determine the power flow for the proposed circuit, conducting the current. Finally, at instant t3 the inverter current
Fig. 1, in (1). This is for frequency control ω = 2πfr with a polarity changes and is carried by S2 . This completes a half
50% duty cycle and a synchronous rectifier load of apparent switching time period of the converter, which repeats in every
resistance Rld . A detailed controls discussion is beyond the switching cycle.
scope of this paper but can be found elsewhere, [17], [18].
2
0.5VDC ω 4 Cr2 L2m Rld B. Resonant Tank
P = 2
2 (ω 2 C (L + L ) − 1) 2
ω 2 L2m (ω 2 Cr Ll − 1) + Rld r l m The resonant tank is realized with an additional series ca-
(1) pacitor, Cr , that is tuned to the transformer leakage inductance,

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

TABLE I: Switch Stress not optimized for the test conditions but is illustrative of the
150 kW design.
Switch IRM S Ipk VRM S Vpk
√
I VDC A. Core Material Selection
1-2 √p Ip VDC
2 2
V√ The first aspect for transformer design is understanding the
3-6 √πP πP out
Vout
2 2Vout 2Vout 2 magnetic material properties that are critical to the compo-
nent performance. These properties depend on dynamic and
excitation details [20]. As such, material properties from the
U.S. DOE Office of Energy material datasheet collection [21]
Ll , (2). The resonant tank capacitor has to be film type as it
are used to supply data for the material core loss Steinmetz
carries the high frequency resonant current. In general, there is
factors, (3), and the dynamic relative permeability, µr . These
some maximum switching frequency which can be chosen to
datasheets provide unique factors, kw the weight proportional-
be limited by converter losses [19], which provides a limit for
ity factor, α the frequency factor, and β the flux density factor
the resonant frequency, fr , to design the capacitor and leakage
for several excitation patterns. The volume of the core Vc is
inductance. Since a gapped transformer, and not traditional
derived from the designed geometry and density is denoted
wireless power coils, is used to provide contactless power
with D instead of ρ to avoid confusion with resistivity.
with insulation between high and voltage sides, there is a
β
relatively high coupling factor between the coils. As such, Pc = kw DVc f α Bpk (3)
a single resonant tank on the primary side is sufficient to
compensate the transformer leakage inductance. This resonant All designs must not saturate the core (4). Often, this saturation
tank networks does not require an extra inductor, because it is avoided by designs with higher frequency or higher turns.
is fully utilized in the tank. With the high coupling between However, high voltage inverters limit solutions to the low to
transmitter and receiver coils, i.e. primary and secondary medium frequency range [19] to avoid excessive switching
windings, compensation in the secondary side is not necessary. losses. Similarly, additional turns lead to excessive winding
losses through additional length, this is especially true in high
1 conversion ratio transformers as the low voltage current can
fr = √ (2) be very high. These constraints on the design drive solutions
2π Ll Cr
with to higher operating flux densities which can preclude the
use of some low saturation flux density ferrites. Therefore,
C. Rectifier
the core material and core geometry become a critical design
Although a current doubler rectifier could aid in the very combination.
low voltage gain that is needed, this rectifier requires two Z 4f1
r
additional inductors. A typical full bridge rectifier requires Vsec (t)dt ≤ Nsec Ac Bsat (4)
none. Because of the high current in the secondary side of 0
the transformer, TABLE I, the selection of rectifier devices
are very crucial which otherwise impact in the efficiency and B. Core Assembly
cost. The sinusoidal current at the rectifier input ensures zero The core geometry must lie above the core cross sectional
reverse recovery of all the rectifier devices. This allows the area requirements while minimizing the winding length. Op-
selection of traditional, low voltage silicon devices which are timally, a circular cross section would be within windings.
much economical than SiC devices. However, this is only available in custom ferrites. Barring
circular, a square cross section better than rectangular profiles.
III. T RANSFORMER D ESIGN Further, to contain stray and leakage flux within the HPMFT
There are many design points to consider in the transformer it is desired to have high permeability core material in as
optimization. These design points are also heavily interlinked much of the core perimeter as possible. Given the limitations
to system level choices e.g. voltage, maximum switching on custom ferrites for our required maximum dimensions, we
frequency, and transformer geometric decisions. By using the propose an assembly of four ’CC’ cores to build a dual axis
HPMFT as both the PHEV charging plug, and the primary ’EE’ core of five limbs, referred to as a ’Quindent’ geometry,
mechanism for reducing the medium voltage to charging Fig. 3. The asymmetric gap is exaggerated for clarity.
levels, there are additional constraints to the design, e.g. a core The ribbon width, w, and build thickness, b, and win-
gap, a winding gap, weight, and volume. This section provides dow length, lw , are illustrated. A ferrite design would have
analytical models of materials, III-A, the windings, III-C and equivalent w and b. The window length must be greater than
III-D, and geometric parameters, III-B needed for determining tp + ts + tg and include space for bobbins and insulation. The
a viable transformer design. These were used for a parametric window height, hw , is in and out of the page must encompass
design of a HPMFT for a 150 kW, 3.5 kVDC to 400 VDC rapid the designed windings and required insulation thicknesses.
charger. These models are verified FEA. A scaled (1/8) 20 kW, The corners of the core window have a bend radios rb .
1 kVDC to 50 VDC laboratory prototype is constructed and The winding thickness of the primary, tp , and secondary
scaled demonstrate key concepts and measure key performance ,ts , winding bundles and the space between the windings,
metrics. An important note is that the hardware prototype is gw , are also shown with simplified winding paths rectangular

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Fig. 3: Quindent HPMFT, midsection profile (left) and 3D Fig. 4: Figures of merit for aluminum and copper windings.
(right) Gray = cores, Red = primary, Green = secondary.
and electrical conductivity as our comparative material. Met-
around the core. It should be noted that a minimum gw is rics of merit are shown in Fig. 4. While C110 has superior
dictated by standard [22]. The mean length turn (MLT) is electrical and thermal conductivity, 1100 is easier to form,
shown as a winding centerline. In a practical build, some coil more machinable, and much less dense. Aluminum is both
bend radius results in a spline MLT profile. In this geometry significantly cheaper per unit weight and less volatile in cost
Ac = 4bwFf where Ff is the fill factor of the laminated core than copper. [23]. The primary consideration in minimizing
material, between 0.75 and 0.9 for tape wound and 1 for ferrite conduction loss in this winding is current crowding due to the
materials. skin effect and proximity effect. The current density due to
It is clear that this layout has several implications to the skin effect is given by (8) where JS is the surface current
other design parameters. For instance, the leakage inductance density, d is the depth below the surface of the conductor,
depends on the winding gap, (5). The MLT also depends on the and δ is the skin depth. For the frequencies of interest δ
resulting core area which will impact the winding resistances can be approximated as (9) where ρ is the bulk conductor
described in III-D and III-C. The core magnetic path, lm , resistivity, f is the frequency of the current, and µ is the
depends on the inner and outer lengths of this geometry(6). magnetic permeability of the conductor. While increasing the
The core volume, critical for core loss can also be derived from winding thickness initially reduces the winding resistance this
these parameters by taking a swept volume of the core area reduction is limited by the skin effect. Fig. 5 shows the impact
and core centerline (7). The box volume of the transformer at the upper frequency range of the resonant control, 50 kHz
component is trivial and left to the reader. for a single layer of foil. Common foil thickness no greater
Np2 µ0 (M LTs + M LTp )(
tp +ts
+ gw ) than 2.5δ are 1.016 mm (40 mil)for the aluminum and 0.813
3
Ll = (5) mm (32 mil) for copper. One trade off with aluminum is better
2lm
utilization meaning a larger effective cross sectional area for
higher resistivity. For example, the 1100 skin depth is 20.5%
(8 + 2π)b
lm =   (6) larger than C110 at 30kHz.
b(8+2π)
ln lh +lw +rb (π−4)+2lg +1 d
J = Js e δ (8)
Vc = 4bwFf (2(lh + lw ) + b(4 + π) + 2πrb ) (7) r
2ρ
δ≈ (9)
8π 2 f 10−7
C. Secondary Winding Design 2) Secondary layering: Another challenge in the secondary
The secondary winding design presents many challenges winding design is the impact of proximity losses [24]. While
due to the high current at medium frequency. With a well a single foil sheet may be adequate to address skin effect
designed synchronous rectifier of near unity power factor, issues, proximity loss effects may be dominant. This can
PFr , and rated power, P and secondary voltage, Vsec , Isec ≈ be ameliorated by multiple interwoven layers ns per turn
P/Vsec . While a large conductor area is needed to meet desired [25], [26]. A limit of 8 layers is practical for construction.
current density, Jsec , only a few turns are needed for the low This effect drives an optimization path towards secondary
voltage side. windings with significant height, hs (10). This makes the
1) Secondary Conductor Material: The use of a foil sec- secondary thickness, ts ,greater as given by (11). A first order
ondary enables flexibility in choosing copper or aluminum approximation of the winding DC resistance is shown in (12)
windings. We present a comparative case study to identify which can be used for parametric optimization. The higher
suitability for use as the HPMFT secondary winding. While order effects of interweaving with skin and proximity impacts
99.5% pure aluminum alloys, e.g. Al 1350, are typically used, for multiple turns must be explored in FEA. Fig. 6 shows the
we focus on 99.0% pure, e.g. 1100, due to better availability. 2D axisymmetric FEA model derived current density for solid
We chose 99.99% pure copper, C110, or its excellent thermal and layered turns of Al 1100 and C110. It is clear that layering

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

shown below in (14,15, 16). This accounts for a number of

primary layers, np , with insulation thickness tins−p . With the
coil and core geometry, the winding resistance is Rp , (17).

Np P Vp2 Pc
Ip = + + (13)
Ns Vs P Fr 2πf Lm Vp
s
Ip /(Jp np )
dp = 2 (14)
Ff π

tp = 2np (dp + 2tins−p ) (15)

Fig. 5: Calculated effective AC winding resistances due to skin hp = Np (dp + 2tins−p ) (16)
effect at 50kHz.
M LTp 4ρNp (b + w + tp )
Rp = ρ = (17)
Ap Ip /Jp

E. Electric Field Management

Two approaches are used to manage the high voltage of the
primary side and protect users on the low voltage side. First,
encapsulant is used to seal the primary wining, provide insula-
tion and enabling some thermal stability. This encapsulant will
not fill the primary side window thus allowing cooling airflow.
To further mitigate the high voltage hazard and to enable a
plug action, a gap is maintained in the magnetic core. This
Fig. 6: Current distribution in low voltage winding designs. gap is asymmetrically cut in the core such that only the half
of the core that is physically near the high voltage winding is
significantly reduces the current density distribution but at the present when the HPMFT is unplugged. The other half of the
expense of increasing the overall winding thickness. core and secondary winding are physically separated from the
high voltage side with a barrier when the plug is closed.
P 1) Encapsulating the Primary Winding: While the core gap
hs = (10) provides separation between medium and low voltage sides,
Vsec PFr Jsec ns ws
it is also important to ensure that the electric field between
the primary winding and the core is maintained below the
ts = 2ns Nsec (ws + tins−s ) (11) breakdown of air, 3 kV/mm [27]. While encapsulant material
can be used to fill the entire space between the winding
M LTs 4ρNsec (b + w + 2(tp + gw ) + ts )) and core, this leaves no room for cooling. Therefore, careful
Rs = ρ = (12)
As P/(Vsec Jsec PF r ) design of the spacing between the winding and core as well as
the the encapsulant thickness is needed to ensure appropriate
electric field limits are met. The required thickness of the
D. Primary Winding Design encapsulant, te , is estimated by using the boundary conditions
There are few significant design options for the primary for the electric field, Eair at the boundary between air and the
winding. Since many turns are required foil is not an attractive encapsulant. The encapsulant has a dielectric constant ke and
option due to assembly complexity. The winding should be the distance between the winding and the core is lp . The tuning
designed to support the primary current, Ip , which is the sum performed on the encapsulant must also consider the field
of the reflected secondary current and the magnetizing current stress that the barrier must support and may be asymmetric
and effective core loss current (13). Generally, litz wire is in the gap. A high detail 2D FEA that was derived from the
required to support higher currents in the primary winding. parametric optimization and 3D FEA was used explore this
However, litz has the penalty of low fill factors, Ff , between issue. Fig. 7 and Fig. 8 show that the electric field in the air
0.5 - .7. This means more physical area is needed to support a on the high voltage side is maintained below 2kV . However,
particular copper area. This can have a detrimental impact by because the encapsulant is sealed to the horizontal limb of the
forcing a longer mean length turn, M LTp One may be able to high voltage side core, the electric field in the gaps is different
use solid wire if a skin depth and current density condition is for positive or negative voltage. Therefore, the gap length and
Ip
met, π(4δ) 2 ≤ Jp . Then, Ff is simply 1. The winding geometry gap material must be designed for the worst case, negative
values, diameter, dp , coil thickness, tp , and coil height, hp , are voltage.

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Fig. 7: Electric field for gaps on the high side with 3.5kVAC .

Fig. 9: Similar losses between gapped and gap-less hysteresis.

and 25mm wide cores while a center gap would have a face
fringing flux percentage of nearly 17%. This is because the l
is only 5mm instead of 85mm.
In comparing the excitation losses of the core material
with both a gap-less and with the 0.5mm gap, no measurable
Fig. 8: Electric field for gaps on the high side with −3.5kVAC . difference in losses was found. This was done using the
approach in [20] with corresponding hysteresis loops in Fig.
9. The core loss is measured from the area of the loop. A
VAC−pk − Eair lp more careful study with a more precise loss measurement
te =   (18) technique, e.g. calorimetry, could shed light on the losses due
Eair ke1−1 to the gap. One possibility for a lack of difference is that
2) Transformer Gap: A gap is desired to both enable the loss increase due to cutting core and even the slightest
mechanical separation as well as for controlling electric fields misalignment of ribbon layers with no gap is near the increase
and parasitic coupling between the high and low voltage sides. due to fringing and some small gap. Potential evidence of this
An asymmetrically cut gap, Fig. 3, achieves both of these effect can be seen by investigating the relative permeability, the
goals. The edge of the gap enables a path for fringing flux. For ratio of flux density, B, to the magnetic field H. The measured
a core of ribbon width w, build thickness, b, and a gap length relative permeability, µrm of the gap-less core is expectantly
lg the magnetic path is a parallel combination of the lossless low at only 1800 compared to the nominal, uncut, core
gap and edge reluctance (19) and the anulus path reluctance of material, µr = 5000 (22). However, the gapped core effective
the core faces (20) where lf is the length of the face around the permeability, µre , meets the expected value of 320 from (23).
gap. The asymmetry of the gap cut aids in reducing the impact The magnetic length path, lm , is 390mm and a total gap length
of fringing flux on the winding as the gap is not in the center g of 1mm, two gaps of lg . As the effective permeability of the
of the winding. The fringing flux can be an additional source core is reduced by the gap, the magnetizing inductance, Lm ,
of loss in magnetic ribbon cores (21), where kf accounts for shown in (24), is also reduced which will impact the primary
the geometry, resistivity and layer effects [28]. The fringing current draw. The core area, Ac , depends on the core geometry
flux that enters the broad surface of the ribbon enters at a and is chosen to meet saturation constraints with the secondary
normal vector and induces additional eddy currents. winding design (4).

4lg 1.222 B
Rg = (19) µrm = (22)
µ0 (4bw1.222 + lg π(b + w))) µ0 H Hmax

π ur
Rf[wkb] = (20) µre = g (23)
lm µr + 1
  
2lf
µ0 [wkb] 1 + ln lg

Np2 µre µ0 Ac

Rg
2 Lm = (24)
Pf ringe = kf f 2 Bpk wb (21) lm
Rg + R2fw
There is an added benefit to the asymmetry of the gap IV. P ROTOTYPE AND E XPERIMENTAL R ESULTS
location. The usable area for fringing flux around the gap is A parametric approach is able to provide multiple design
reduced by the short core limb on the non-centered gap. In possibilities that incorporate the competing constraints. Fig. 10
this design this means that the flux into the broad ribbon face shows the method for determining a solution space of HPMFT
is less than 10% of the magnetizing flux for both of the 15mm designs and a final solution. The core geometry is restricted to

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Given: P, Vp , Vs , fr , Ns−max

Core area solution set:

Winding gap: gw [22]
Bpk ∈ 0.05T : Bsat (4)

Define min core gap, g: Pri coil solution set;

V
2V
max( Vbp , mtl. min.) Np = Vps Ns
Lm , Ip (6, 13)
dp , hp , tp (14, 16, 15)
Define coil prop.: np ∈ 1 : np−max
ρs , Js−max , ρp , Jp−max
ws−min , hs−max Design core window:
Fig. 11: Parametric fronts of losses and volume for quindent
III-B; exclude
designs of square ferrite and ft3l of different ribbon widths.
too big designs
Sec coil solution set;
ws ≤ 4δ, hs , ts (9-11)
Ns ∈ 1 : Ns−max Exclude lossy designs:
ns ∈ ns−min : ns−max Core loss: (3, 21)
Coil loss: (13, 17, 12)

Select from preliminary

HPMFT solutions
Increase
space (7) E>
Eb
Jp > Jpri−max
ns−min ++ FEA check
Fig. 12: Magnetizing (left), leakage (right) fields for HPMFT.
max
J sec− Used for validating and scaling lab prototype, TABLE II
ns−min ++ Js > All Pass
Finished design asymmetric geometries can have more of an impact. Where
high mesh details are needed, e.g. winding skin effect, the 2D
Fig. 10: Flow chart of HPMFT design method. model is used. The homogenization of the 2D axisymmetric
FEA are adjusted to reflect the behavior exhibited in the 3D
FEA, Fig. 12. With FEA results matching closely with the
the Quindent design and the primary coil geometry is vertically
analytically predicted values we began developing a laboratory
wound with wire, litz or solid, while the secondary coil is
prototype that matched the FEA predicted behavior, Fig. 13.
radially wound with foil. At each node, all combinations are
calculated until they are excluded for volume and losses. A
preliminary design is chosen to model in FEA and if it passes A. Hardware
all electric field and current density limit checks, the design Using the results of the parametric design and FEA ver-
is chosen as final and solutions are stored on the parametric ification, a low voltage hardware prototype was constructed
front. and tested with the build metrics in XX. This prototype was
Fig. 11 shows the preliminary HPMFT solutions results designed to help study the high voltage loss models but operate
of an optimization using the presented analytical models for at 1/8 voltage and power. The low voltage is a 1000 V to 50
a 150 kW, 3.5 kVAC to 400 VAC design. Only the power, V transformer that uses 20 turns on the primary and one, 7
voltage levels, and material parameters are needed to populate layer foil turn on the secondary. The difference between the
the solution space if the presented equations are used. One transformers is the number of turns and the winding gap, gw .
constraint administered was designs of only one type of ribbon The magnetic core and configuration is the same as the core
layer. However, investigation around the crossover points of specified for the high voltage design. The low voltage HPMFT
the FT3TL core material of ribbon widths of 15mm and 25mm prototype, the same size as the 150 kW design, is shown in
found a design that uses both ribbon widths and the same Fig. 13 next to a fast charging electrical contact type plug.
transformer build with minimized losses and volume. This Our contactless plug is very similar in size and volume to the
solution was verified in 3D and 2D axisymmetric FEA. The traditional design but twice the power. Fig 14 shows the plug
magnetic fields were explored primarily in 3D FEA where in the laboratory setup for scaled testing of the high power

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

TABLE III: Experimental Values

Value FEA Predicted Low Voltage Measured

Rp 1Ω 1.2 Ω
Rs 900 µΩ 850 µΩ
Lm 4.6 mH 0.37 mH
Ll 96.4 µ H 113 µ H
gw 4mm 11.25mm
KCoupling 0.98 0.77

Fig. 13: 150 kW HPMFT and 63 kW CHAdeMO 1.0 plugs.

TABLE II: Setup Parameters
Parameter Symbol Value
Primary turns Np 20
Secondary turns Ns 1
Primary type Pri 6 Awg Cu Litz
Secondary type Sec Layered Al foil Fig. 15: Max power, full resonant excitation; 2 kV VAC−pk
Primary layers np 1
Secondary layers ns 7 this time, the inverter and rectifier where designed only to
Secondary Height hs 6 in
meet the required excitation and loading requirements. One
Secondary layer thickness ws 5 mil
Core gap g 0.5 mm parameter that could not be matched was the magnetizing
Winding gap gw 11.25 mm inductance. Maintaining the same core and gap but with
Core material Mtl ft3TL significantly fewer turns reduces this magnetizing inductance,
Core width w 25 mm and 15mm resulting significantly higher magnetization current. This will
Core build b 11 mm
Window height hw 175 mm, cut at 170/5mm be explored in loss separation plots below.
Window width lw 25mm
B. Resonant Compensation Selection
The resonant frequency was chosen as 30 kHz. The selected
nanocrystalline core for the HPMFT has suitable operating
frequency between 10-50kHz where designs are more efficient
than comparable ferrites. Furthermore, low voltage secondary
circuitry needs to handle significant currents. Higher frequency
presents design challenges from skin and proximity effects. A
balanced design was chosen at 30 kHz. The leakage inductance
is an artifact of the spacing between the windings and is found
from (5). Then, the resonant capacitor is defined by (2).
Voltage regulation is a critical requirement for proper battery
charging. The load varies significantly from nearly no load
Fig. 14: Scaled power converter circuit for HPMFT validation. during a trickle charge to full load at the initial connection. The
resonant converter operation can adequately provide power
over this large range, (1). However, it is critical to investigate
medium frequency transformer plug. TABLE II enumerates key internal voltages to ensure that the terminal voltages are
build parameters. not overloaded. Fig. 15 and Fig. 16 show transformer input
In order to better understand the scale up issues, the low voltages at different compensation levels. The terminal voltage
voltage design uses the same core design and is excited to and therefore voltage regulation is dependant on the interaction
the same volt-seconds per turn (flux density) as the proposed of the resonant tank and the static transformer gain. Fig. 17
high voltage. This means that the low voltage prototype shows how this interaction can cause changes in the effective
magnetizing losses are the same as the high voltage design. turns ratio.
In order to understand the conduction losses, the transformer
is operated up to the rated output current. The windings are
designed with added resistance to match the FEA predicted C. Measured losses
winding resistance and are constructed in similar methods. The measured total losses, recorded with a Yokogawa
A comparison of high voltage design FEA predicted and WT3000, for various resonant compensation percentages are
measured low voltage parameters is shown in TABLE III. At shown in Fig. 18. A higher frequency, 50 kHz, excitation

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Fig. 16: Max power, 72% resonant excitation; 1.7 kV VAC−pk

Fig. 19: HPMFT partial resonant excitation loss separation.

a rated load efficiency of about 96.5%. In the final proposed

medium voltage system, 10 kV SiC mosfets would be used
for the inverter. Paralleled or high current, 600V class, mosfets
should be used to be used for the rectifier to achieve reasonable
system efficiencies.

V. D ISCUSSION AND C ONCLUSION

We presented a method of safely providing 150 kW for
rapid charging electric vehicles that couples medium voltage
Fig. 17: Compensation level impact on voltage ratio Vs /Vp grids to standard 400 VDC charging. Our novel design uses
a high step down transformer with deliberate gaps to provide
complete electrical and electric field isolation. We presented
the many interconnected design models that estimate operating
metrics necessary for parametric optimization. This gapped
transformer solution provides all of the typical transformer
performance capabilities to the DC-DC converter. It provides a
high voltage step down, galvanic isolation, and operates at low
losses. By introducing the gap, a plug action is now available
to untrained personnel for safe connection between electric
vehicles and medium voltage supplied charging stations. Fu-
ture work will study the viability of other converter topologies
Fig. 18: Total quindent losses of LV HPMFT prototype.
and control techniques. A low voltage prototype, based on the
FEA validated parameters of the proposed high voltage design,
was also tested in the low voltage prototype to explore what allowed exploration of various resonant compensation methods
potential future switches could enable. The 30 kHz, non- and high frequency operation. Our design shows great promise
resonant excitation was halted early due to poor voltage for safe, robust, and very high efficiency DC rapid charging
regulation. solutions.
We separate the losses of the HPMFT prototype to deter-
mine where the performance might be improved, Fig. 19. IGSE R EFERENCES
method is used to estimate the core loss considering the semi- [1] S. S. Williamson, A. K. Rathore, and F. Musavi, “Industrial elec-
resonant excitation [29]. While significant magnetizing current tronics for electric transportation: Current state-of-the-art and future
contributes to losses in the low voltage prototype, the total challenges,” IEEE Trans. Ind. Electron., vol. 62, no. 5, pp. 3021–3032,
May. 2015.
current levels of the primary and secondary windings at 20 kW [2] M. Yilmaz and P. T. Krein, “Review of battery charger topologies,
are near the rated current levels of the 150 kW design. With charging power levels, and infrastructure for plug-in electric and hybrid
similar winding resistances, these losses are representative vehicles,” IEEE Transactions on Power Electronics, vol. 28, no. 5, pp.
2151–2169, May. 2013.
of total losses in the medium voltage contactless magnetic [3] L. Rubinoa, C. Capassob, and O. Veneri, “Review on plug-in electric
plug. With the presented breakdown in losses, it is expected vehicle charging architectures integrated with distributed energy sources
that the high voltage HPMFT design will operate near 99.5% for sustainable mobility,” Applied Energy, vol. 207, pp. 438 – 464, 2017.
[4] V. M. Iyer, S. Gulur, G. Gohil, and S. Bhattacharya, “An approach
efficiency. However, dielectric losses are not currently included towards extreme fast charging station power delivery for electric vehicles
which will lower this predicted efficiency. with partial power processing,” IEEE Trans. Ind. Electron., pp. 1–1,
The inverter and rectifier were assembled only for testing the 2019.
[5] Y. Zhang, Z. Yan, T. Kan, and C. Mi, “Interoperability study of fast
HPMFT at scaled excitation levels. The inverter is built with wireless charging and normal wireless charging of electric vehicles with
the CREE 1.7kV SiC modules-CAS300M17BM2. The inverter a shared receiver,” IET Power Electronics, vol. 12, DOI 10.1049/iet-
is soft switched due to the resonant operation. The efficiency pel.2018.6080, no. 10, pp. 2551–2558, 2019.
[6] “Chademo releases the latest version of the protocol enabling up to
of the inverter at the rated load is around 99.5%. However, 400kw,” Jun. 2018. [Online]. Available: https://www.chademo.com/
a high current passive diode bridge is used rectification, with chademo-releases-the-latest-version-of-the-protocol-enabling-up-to-400kw/

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JESTIE.2020.2999589, IEEE Journal
of Emerging and Selected Topics in Industrial Electronics
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

[7] C. Suarez and W. Martinez, “Fast and ultra-fast charging for battery Richard B. Beddingfield (GS’13, M’18) re-
electric vehicles - a review,” in 2019 IEEE Energy Conversion Congress ceived his EE BS, MSPE, and PhD in 2014,
and Exposition (ECCE), DOI 10.1109/ECCE.2019.8912594, pp. 569– 2015 and 2018 at North Carolina State Uni-
575, Sep. 2019. veristy (NCSU). His focus is the intersection of
[8] D. E. N. B.V, New UFC charger brochure V1 EN 2018-04-13, 2018. high power converters and fundamental mag-
[9] ABB, Global product portfolio, 2019. netic designs and materials. As an ORISE post-
[10] M. Vasiladiotis and A. Rufer, “A modular multiport power electronic doctoral research fellowship with NETL he works
transformer with integrated split battery energy storage for versatile with high temperature magnetics, strain induced
ultrafast ev charging stations,” IEEE Trans. Ind. Electron., vol. 62, no. 5, anisotropy magnetic ribbon materials and power
pp. 3213–3222, May. 2015. and pulse transformers. He is also a member of
[11] P. Chaudhary, S. Samanta, and P. Sensarma, “Input-series-output- The Minerals Metals and Materials Society.
parallel-connected buck rectifiers for high-voltage applications,” IEEE
Trans. Ind. Electron., vol. 62, no. 1, pp. 193–202, Jan. 2015. Suvendu Samanta (M’16) received his Elec-
[12] L. Gill, T. Ikari, T. Kai, B. Li, K. Ngo, and D. Dong, “Medium voltage trical Engineering Ph.D. from Concordia Uni-
dual active bridge using 3.3 kv sic mosfets for ev charging application,” versity, Montreal, QC, Canada, in 2018 where
in 2019 IEEE Energy Convers. Congr. Expo., pp. 1237–1244, Sep. 2019. he won the Canadian Governor General’s Gold
[13] S. Sen, L. Zhang, T. Chen, J. Zhang, and A. Q. Huang, “Three-phase Medal. He was with Coal India Ltd. from 2009
medium voltage dc fast charger based on single-stage soft-switching to 2011 and a Research Engineer 2014 to 2016
topology,” in 2018 IEEE Transport. Electrific. Conf. Expo, pp. 1123– at National University of Singapore. He is a
1128, Jun. 2018. Postdoctoral Fellow at the FREEDM Research
[14] X. Liang, S. Srdic, J. Won, E. Aponte, K. Booth, and S. Lukic, “A Center, NCSU with interests in modeling and
12.47 kv medium voltage input 350 kw ev fast charger using 10 kv sic control of power converter topologies, wireless
mosfet,” in 2019 IEEE APEC, pp. 581–587, Mar. 2019. power transfer, and transportation electrification.
[15] S. Samanta and A. K. Rathore, “Small-signal modeling and closed- Mark Nations (S’18) received his EE BS in 2019
loop control of a parallel-series/series resonant converter for wireless from NCSU. He started and ran the specialty
inductive power transfer,” IEEE Trans. Ind. Electron., vol. 66, no. 1, pp. manufacturing firm Black Rock Precision LLC
172–182, Jan. 2019. from 2012-2018. He began his ORISE Gradu-
[16] M. M. Jovanovic and B. T. Irving, “On-the-fly topology-morphing ate Fellow in 2019 under Dr. Subhashish Bhat-
control-efficiency optimization method for llc resonant converters op- tacharya at the FREEDM Systems Center where
erating in wide input-and/or output-voltage range,” IEEE Transactions he develops advanced magnetics components
on Power Electronics, vol. 31, no. 3, pp. 2596–2608, 2016. to complement and promote the adoption of
[17] Q. Cao, Z. Li, and H. Wang, “Wide voltage gain range llc dc/dc emerging wide bandgap power semiconductor
topologies: State-of-the-art,” in 2018 International Power Electronics technologies.
Conference (IPEC-Niigata 2018 -ECCE Asia), pp. 100–107, 2018.
[18] K. Yoon, Y. Noh, S. Phum, S. Meas, S. Jang, and E. Kim, “Llc Isaac Wong received the B.Sc in electrical engi-
resonant converter with wide input voltage and load range at fixed neering from the University of Illinois at Urbana-
switching frequency,” in 2012 Twenty-Seventh Annual IEEE Applied Champaign, Urbana, IL in 2015. He is currently
Power Electronics Conference and Exposition (APEC), pp. 1338–1342, a graduate student in Bhattacharya’s Group at
2012. the FREEDM Systems Center at North Car-
[19] A. Anurag, S. Acharya, S. Bhattacharya, and T. Weatherford, “Thermal olina State University. His research focus is in
performance and reliability analysis of a medium voltage three-phase WBG powr converters and enabling magnetics
inverter considering the influence of high dv/dt on parasitic filter designs.
elements,” IEEE Journal of Emerging and Selected Topics in Power
Electronics, pp. 1–1, 2019.
[20] R. Beddingfield and S. Bhattacharya, “Multi-parameter magnetic ma-
terial characterization for high power medium frequency converters,” Paul R. Ohodnicki Jr. earned his M.S. (2006)
in The Materials, Metals, & Materials Society (TMS) Supplemental and Ph.D. (2008) in materials science and en-
Proceedings, pp. 693–708, 2017. gineering from Carnegie Mellon University. He
[21] S. R. Moon, P. Ohodnicki, K. Byerly, and R. Beddingfield, “Soft is currently an Associate Professor in the Me-
magnetic materials characterization for power electronics applications chanical Engineering and Materials Science de-
and advanced data sheets,” in 2019 IEEE Energy Conversion Congress partment at the University of Pittsburgh and the
and Exposition (ECCE), pp. 6628–6633, Sep. 2019. Associate Coordinator of the Engineering Sci-
[22] “International standard iec 60950-1,” 2005. ence program. He was recently the technical
[23] J. C. Olivares-Galvan, F. de Leon, P. S. Georgilakis, and R. Escarela- portfolio lead in the Functional Materials Team
Perez, “Selection of copper against aluminium windings for distribution of the Materials Engineering & Manufacturing
transformers,” IET Electric Power Applications; Stevenage, vol. 4, no. 6, Directorate of the National Energy Technology
pp. 474–485, Jul. 2010. Laboratory. Ohodnicki has published more than 130 technical publica-
[24] M. E. Dale and C. R. Sullivan, “General comparison of power loss tions and holds more than 10 patents, with more than 15 additional
in single-layer and multi-layer windings,” in 2005 IEEE 36th Power patents under review and an R&D 100 Award (2019) in sensing and
Electronics Specialists Conference, pp. 582–589, Jun. 2005. power electronics, advanced devices and enabling functional materials
[25] C. R. Sullivan, “Layered foil as an alternative to litz wire: Multiple for photonic and wireless sensing and power magnetics component and
methods for equal current sharing among layers,” in 2014 IEEE 15th materials design.
Workshop on Control and Modeling for Power Electronics (COMPEL), Subhashish Bhattacharya received his B.E.
pp. 1–7, Jun. 2014. (Hons), M.E. and PhD degrees in Electrical En-
[26] M. E. Dale and C. R. Sullivan, “Comparison of single-layer and multi- gineering from Indian Institute of Technology-
layer windings with physical constraints or strong harmonics,” in 2006 Roorkee, India in 1986, Indian Institute of Sci-
IEEE International Symposium on Industrial Electronics, vol. 2, pp. ence (IISc), Bangalore, India in 1988, and Uni-
1467–1473, Jul. 2006. versity of Wisconsin-Madison in 2003, respec-
[27] P. A. Tipler, College Physics. Worth, Jun. 1987. tively. As a Professor at NCSU, he as au-
[28] R. B. Beddingfield, S. Bhattacharya, and P. Ohodnicki, “Shielding of thored over 350 peer-reviewed technical articles,
leakage flux induced losses in high power, medium frequency trans- 2 book chapters, and has 5 issued patents in
formers,” in 2019 IEEE Energy Conversion Congress and Exposition Solid-State Transformers, MV power converters,
(ECCE), pp. 4154–4161, Sep. 2019. FACTS, Utility applications of power electronics
[29] K. Venkatachalam, C. R. Sullivan, T. Abdallah, and H. Tacca, “Accurate and power quality issues; high-frequency magnetics, active filters. He is
prediction of ferrite core loss with nonsinusoidal waveforms using only a founding faculty member and co-PI of NSF ERC FREEDM systems
steinmetz parameters,” in 2002 IEEE Workshop on Computers in Power center, Advanced Transportation Energy Center, and DOE initiative on
Electronics, 2002. Proceedings., pp. 36–41, Jun. 2002. WBG based Manufacturing Innovation Institute, PowerAmerica.

2687-9735 (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: CALIFORNIA INSTITUTE OF TECHNOLOGY. Downloaded on July 04,2020 at 07:09:35 UTC from IEEE Xplore. Restrictions apply.