High‐Frequency
Passive Components:
a critical challenge for power electronics
Prof. Charles R. Sullivan
[email protected]
Dartmouth Magnetics and Power
E l e c t r o n i c s Re s e a r c h G r o u p
http://power.engineering.dartmouth.edu 1
Power Electronics Research at
Dartmouth: 2 groups
Prof. Jason Stauth: Power integrated circuits.
Resonant switched capacitor integrated converters.
Applications in PV, battery systems, RF communications, power
delivery for digital systems.
Prof. Charles Sullivan: Magnetics, circuits and systems
Modeling and optimization of “macro” magnetics:
50 W to 250 kW.
Fabrication, materials, design and modeling of microfabricated
magnetics: 1 W to 25 W, on‐chip or co‐packaged. Ref [1]
power.thayer.dartmouth.edu 2
Magnetics in power electronics
Increasingly critical bottleneck
Responsible for much of the
Size (volume and weight)
Power loss
Solantro 350 W PV microinverter:
Cost Miniaturized control chips are
great but passives are still huge.
Difficulty in design
(long development cycles)
power.thayer.dartmouth.edu 3
Two goals for magnetics research
Models:
Accurate, capturing effect that are usually ignored.
Fast, for use in optimization.
Simple and easy, for engineers who usually don’t bother.
Innovative designs and technologies for higher
performance.
Identify limitations of present technology and overcome
them.
Start from fundamental goals and explore ways to
accomplish them.
power.thayer.dartmouth.edu 4
Winding models vs. Core models
Linear, well known Nonlinear material
material properties. properties, known only
Behavior is a solution to through measurements.
Maxwell’s equations. Models are behavioral,
Numerical, analytical, or based on measurements.
mixed solutions. Physics‐based micromagnetic
models exist, but can’t
Often complex address ferrite loss yet.
geometries
Usually simple geometries.
power.thayer.dartmouth.edu 5
High‐frequency winding loss models
Current
waveforms
Physical Design
Geometry & Materials ? Loss
Winding ac resistances?
power.thayer.dartmouth.edu 6
Loss calculated from
Primary Secondary
currents
Conventional, incorrect, model for
transformer winding loss
(assume sine waves for now).
2 2
Pwinding = I1 R1 + I2 R2
0°
Problem: Loss varies drastically
depending on relative 180°
phase/polarity.
Factor of 4 error in this case.
Correct model options:
R1 and R2 that are only for
specific phase relationship.
Resistance matrix.
power.thayer.dartmouth.edu 7
Winding models
Current
waveforms
Physical Design
Geometry & Materials R(f) Loss
Winding ac resistances?
Frequency‐dependent resistance matrix R(f).
Captures interactions between windings.
Ref:[32, 2, 27]
power.thayer.dartmouth.edu 8
Winding models
Current
waveforms
Physical Design
Geometry & Materials R(f) Loss
Remove
core
effect
Electrical See [30] Foo, Stein and Sullivan, APEC
Measurements 2017, for the measurement approach
power.thayer.dartmouth.edu 9
Predictions from physical structure
1‐D fields 2‐D or 3‐D fields
Rectangular Analytical Numerical
conductors (Finite Element, PEEC, etc.)
(e.g. foil and PCB)
Round‐wire Simulation‐tuned Simulation‐tuned physical
conductors physical model model + dc field simulation
(including litz):
State of the art
before work at
G2Elab.
10
Winding models:
1D, rectangular conductors
Physical Design Dowell,
Spreen [32] Loss
Geometry & Materials
M2SPICE
(MIT) [31]
Circuit model
for simulation
power.thayer.dartmouth.edu 11
Round conductor:
Textbook problem
Cylinder subjected to uniform field
Dowell’s model is a crude approximation.
power.thayer.dartmouth.edu 12
Textbook solution
Exact solution,
described by Bessel
functions.
Use for winding loss
analysis pioneered by
Ferreira.
power.thayer.dartmouth.edu 13
Actual problem
Array of cylinders
subjected to uniform field
Several solution
approaches …
But first, does it matter?
power.thayer.dartmouth.edu 14
Using the Bessel solution
for the real problem
Not a valid solution!
power.thayer.dartmouth.edu Real Solution (FEA) 15
Simulation Results
100 Bessel
Proximity loss factor
10
Dowell
1 Real behavior is between Dowell and
Bessel.
0.1
Sometimes closer to Dowell.
Identical in low‐frequency range with
simple correction.
0.01
1 10
d/
power.thayer.dartmouth.edu 16
Xi Nan’s model [ref 26]
100
Bessel function method
Weighted average of Dowell method
Dowell‐like and 80
Experimental Data
Our model
Bessel‐like behavior:
“Simulation tuned 60
Rac/Rdc
physical model”
Fits experimental 40
results better than
Dowell or Bessel. 20
Can be applied to 2D 0
or 3D field 0 2 4 6 8 10 12
d/
configurations …
power.thayer.dartmouth.edu 17
Full winding loss model: 2‐D, full
frequency range, multi‐winding interactions
Hybridized Nan’s method Homogenization with
([2] Zimmanck, 2010) complex permeability
(Nan 2009, Meeker, 2012 [28],
etc.)
Available in
FEMM
power.thayer.dartmouth.edu 18
Winding models
Round wire/2D: Current
“Hybridized Nan’s method” waveforms
Physical Design
Geometry & Materials R(f) Loss
Remove
core
effect
Electrical References:
Measurements [30] for measurement approach
[2] for the “Hybridized Nan’s Method”
power.thayer.dartmouth.edu 19
Advanced litz wire models including
1.4
construction details 1.2
FastLitz
New model
Maxwell 2-D
skin-effect resistance, m
Important for large number of 1
strands and/or small number of turns. 0.8
0.6
Our recent research results:
0.4
Basic guidelines [10]
0.2
Detailed model [11]
0
125 5x25 25x5 5x5x5
Research needs: 2.5
Optimization, verification and
Loss per unit length (W/m)
2
economics.
1.5
Terminations that preserve litz
behavior. 1
0.5
0
0 10 20 30 40 50 60 70
1050 strands of AWG 44, Top level twisting pitch (mm)
constructed as 5x5x42 20
1D, 2D and 3D modeling approaches
1D: can use analytical models.
For Xfrmers and good (quasi‐) distributed gap Ls.
Dowell isn’t precise but we know how to do better.
2D: Fast, easy, low‐cost simulations.
Naïve sections for E‐cores can be misleading.
Mimic return path for to reduce error 5X [Ref 25]
3D: Use for
verification,
not design.
power.thayer.dartmouth.edu 21
Core models
Flux
waveforms
Physics
Loss
Loss
calculation
Loss
model
Electrical Dynamic model
Circuit simulation
Measurements
power.thayer.dartmouth.edu 22
Core Loss Calculation Models
Steinmetz equation:
Sinusoidal waveforms only
Various types of modified/generalized/etc. Steinmetz equations.
Extend to non‐sinusoidal waveforms.
Most common:
improved Generalized Steinmetz Equation (iGSE) [4]
Loss Map/Composite Waveform Method [5]
power.thayer.dartmouth.edu 23
Waveform effect on core loss:
Concepts, rather than how‐to
Initial hope in GSE model: instantaneous loss
depends on B and dB/dt: p(t) = p(B(t), dB/dt)
If this worked, you could add up loss for
incremental time segments:
Eloss = E1 + E2 + …
B(t)
or better, an integral…
It doesn’t work:
flawed concept
power.thayer.dartmouth.edu 24
Improvement that enabled iGSE [4]
Loss depends on segment dB/dt
and on overall ΔB
Still Eloss = E1 + E2 + …, but E1 depends on a global
parameter as well as a local parameter.
B(t) dB/dt
ΔB
power.thayer.dartmouth.edu 25
Composite waveform method [4]
Same concept as GSE: add up independent loss for each segment.
B(t)
= +
Eloss = E1 + E2
Unlike the GSE, this works pretty well in simple cases:
Waveforms where ΔB is the same for the segment and the
whole waveform!
It reduces to the same assumptions as the iGSE.
power.thayer.dartmouth.edu 26
What we know how to do for non‐
sinusoidal waveforms:
For simple waveforms, add up the loss in each segment.
= +
For waveforms with varying slope, add up the loss for each
segment, considering overall ΔB and segment δB.
ΔB = + + δB
See iGSE paper for how those factor in [4].
For waveforms with minor loops, separate loops before
calculating loss (see iGSE paper [4]).
= +
power.thayer.dartmouth.edu 27
Loss models for each segment
iGSE derives them from a Steinmetz model
Limitation: Steinmetz model holds over a limited
frequency range.
Loss map model uses square‐wave data directly for a wide
frequency range.
Clearly better if you have the data.
Can also map with different dc bias levels.
Sobhi Barg ([29] Trans. Pow. Electr., March 2017) shows
that the iGSE gets much more accurate if you use different
Steinmetz parameters for each time segment in a triangle
wave.
power.thayer.dartmouth.edu 28
Limitation for all of the above:
open research question.
“Relaxation effect” 1 cycle
Simple theory says
loss for one cycle
should be the
same for both
flux waveforms.
In practice, it’s different.
i2GSE (Jonas Mühlethaler
and J. Kolar) captures this
but is cumbersome and 1 cycle
requires extensive data.
power.thayer.dartmouth.edu 29
Modeling Conclusions
Winding loss:
Complex but feasible to model accurately.
For 2 or more windings, need resistance matrix.
1D rectangular conductors: analytical solutions.
2D rectangular conductors: numerical simulations.
1D or 2D round wire: Simulation‐tuned physical
models are better than Dowell or Bessel.
Core loss
Nonlinear and can only be found experimentally.
Open questions on data needed and models.
power.thayer.dartmouth.edu 30
Design
Models (often) predict poor performance
What can we do better?
Optimization and design innovation for kHz
frequencies.
MHz frequency challenges and solutions.
Reconsideration of passive components.
power.thayer.dartmouth.edu 31
High‐frequency winding design
Most critical is proximity effect: interaction of field and
conductors.
Rac/Rdc
Not just diameter < skin depth: = 27.7
need d << δ in a multilayer winding. with
How much improvement is possible d = 2δ
with many thin layers vs. a single layer?
With a number of layers, , can improve by 1/
With a minimum thickness, , can improve by
For 10X improvement: 100 layers, t ~= δ/7 Ref:[6,7,8]
Need right combination—optimization is essential.
power.thayer.dartmouth.edu 32
Litz wire
Strands d << δ.
Invention: 1888,
Sebastian de Ferranti.
Image: Noah Technologies
Analysis 1917 Howe; 1926 Butterworth.
Conventional design options:
Papers with lots of complex math.
Catalog guidelines … but these can lead to higher
loss than with solid wire at much higher cost.
power.thayer.dartmouth.edu 33
Litz‐wire design options
10
32
34
36
Loss
Normalized Loss
38
Full-bobbin design—
40 double the loss of cheaper
design.
42
1
Cheaper and 44
Good design
lower loss 46
requires optimization
48
50
0.1 1 10 100
Cost….~ amount of Cu
Normalized Cost
power.thayer.dartmouth.edu
Litz wire
Strands d << δ.
Invention: 1888,
Sebastian de Ferranti.
Image: Noah Technologies
Analysis 1917 Howe; 1926 Butterworth.
Conventional design options:
Papers with lots of complex math.
Catalog guidelines … but these can lead to higher
loss than with solid wire at much higher cost.
One solution: single‐formula design.
2
b
[Ref 10], http://bit.do/simplitz n k
e
NS
power.thayer.dartmouth.edu 35
Inductors
Fringing field
near gaps x/2
complicate
design. x
s
Options to
change:
Winding shape.
Gap
configuration
[Ref 15]
power.thayer.dartmouth.edu 36
Winding shape optimization
Shape winding configuration to
work with curved gap field.
Applies to round wire and litz wire,
not foil.
Can actually work better than a
distributed gap!
Ad‐hoc approach common, but full
optimization is available [Ref 16].
power.thayer.dartmouth.edu 37
Costs: Cu vs. Al (as of 8 March 2016)
Mass:
$5.00/kg vs. $1.6/kg (wrong metric)
Volume basis:
4.42 ¢/cm3 vs. 0.43 ¢/cm3 10X
Resistance basis:
$Ω $Ω
7.67 vs.
m 2 1.22 m2 7.3X
>7X more cost effective … dc or low frequency.
What about high frequency?
Experiments and analysis show that the performance
gap between Al and Cu is smaller at high frequency!
power.thayer.dartmouth.edu
Real comparison of Al and Cu
Fair comparison of good designs: Compare
a design optimized to use Al well, vs.
a design optimized to use Cu well
140%
Cu Al 1 Al 2 Result: [Ref 22]
120%
100%
where to use Al:
80% Most situations!
60% Where to use Cu:
40%
Where compact size is more
20%
0%
important than efficiency, cost,
Volume Loss Cost Weight temperature or weight.
If termination cost difference
exceeds wire cost difference.
power.thayer.dartmouth.edu 39
Miniaturization with MHz frequencies?
We have good materials and design methods for 20 kHz to 300 kHz .
New semiconductors emerging
GaN and SiC power devices:
now commercially available,
> 10X switching speed vs. Si.
Theoretically allows smaller,
more efficient magnetics.
But can this be realized
in practice?
Windings?
Core materials?
IEEE power electronics magazine,
Credit to Jelena Popovic and Dragan March 2015
Maksimovic for the ball and chain analogy 40
Windings at MHz frequencies … Litz?
100
Litz benefits drop off rapidly
80
in the MHz range [36]
% loss reduction
Barely better than a 60
solid‐wire winding.
40
Huge room for 20
improvement in theory:
A single‐layer winding 0
100 kHz 1 MHz 10 MHz
only has current in one skin depth: At 10 MHz, 21 µm.
0.2% of a 1 cm winding window (0.23% with litz).
→ 400X improvement theore cally available.
41
Foil: < 20 µm at low cost
Easy to get thickness << skin depth.
Freestanding foil down to ~ 6 µm.
On plastic‐film substrates for ease of handling from 35
µm to << 1 μm.
Thin layers have high dc resistance—
need many in parallel.
Challenges:
Achieving uniform
current density—laterally and among layers.
High capacitance between layers.
Terminations
42
One concept for MHz foil windings:
capacitive ballasting
Overlapping insulated layers create series capacitance for each
layer.
Cartoon: real structures have many more layers
Port 1
Capacitive ballasting forces equal current sharing.
Can create integrated LC structure, a concept with
a long history.
In addition to integration, solves MHz winding
loss challenges.
43
Resonant structure for wireless
power [34]
Many stacked layers with no vias and no
terminations.
Current sharing between many thin layers
enforced by same capacitance used for
resonance.
44
Operation principle – single section
Each section:
Equivalent circuit model
Side view
Inductive current loop
Capacitive connection between foil layers through
dielectric
power.thayer.dartmouth.edu 45
Operation principle – many sections
Strong mutual coupling between all layers.
Each section capacitance is coupled to form a Section
parallel LC resonator.
Coupled section capacitance forces equal
current sharing in each layer.
Integrated capacitance Section
eliminates high current
terminations.
Experimental Q = 1180
with 66 mm diameter.
> 6X improvement over state of the art.
Improves range and efficiency of WPT (improves
from ~50% to ~90% at d = D)
power.thayer.dartmouth.edu 46
Materials for Performance factor: MIT measurements, 2015-2016
MHz range loss-limited power capability
of a magnetic material
MnZn and NiZn
ferrites for MHz
power.
Significant
improvements in last
1‐2 years.
More in
development.
Thin‐film
materials prove
this is possible.
Winding approaches
that overcome skin
and proximity effect Modified perf. factor:
allow using top derated for winding loss
graph.
2017.06.12 47
47 Measurements and performance factor comparisons …AJ Hanson, et al, IEEE Trans. on Pow. Electr. 31 (11), 2016
Reconsideration of passive
components [35]
Start from fundamental function of passives and
consider possible technologies.
Identify alternatives, and/or
Confirm value of standard
approaches.
These functions are:
Energy storage
Transformation (voltage/current ratio)
Isolation
48
Passive functions
Energy storage
Magnetic (inductor)
Electrostatic (capacitor)
Kinetic energy (moving mass)
Elastic energy (spring)
Others considered but rejected (e.g., pneumatic).
Transformation (voltage/current ratio)
Isolation
49
Energy storage density limits
MK Mass
Spring (K)
Cap
Inductor (L) LC
Often want resonant pairs: LC or MK
MK looks attractive, but requires transduction.
Electromagnetic: limitations similar to L.
Piezoelectric: candidate for further exploration.
50
High‐level analysis of potential
Optimistic assumptions to examine future potential.
Resonant switched capacitor (ReSC) circuit
(aka switched tank converter, STC)
Limited but expanding application scope.
Performance limited by
Dissipation and temperature rise.
Mechanical and electrical breakdown.
51
Approximate capability in 1 mm3
At this level of analysis:
All options offer extremely high
performance.
With advanced materials, potential
performance of
piezo and LC resonators is broadly
similar.
Commercial materials look worse, but
still impressive.
52
Experimental results in ~1 cm3
0.5 mΩ ESR in a 250 V dc
rated component.
Experiments match theory.
Without considering any
limitations of today’s
power switches, over 10
kW would be possible at
over 99% efficiency.
53
Conclusions from Fundamental
Examination of Passives
Piezo resonators: excellent potential, but much work needed to
reach full capability.
LC resonant structures: easier technology, experiments
demonstrate good performance already.
Scaling of piezo to small sizes is excellent whereas
magnetics performance degrades.
54
Conclusions on Magnetics
Proximity effect is the primary winding design consideration.
Established winding loss reduction techniques include litz wire,
interleaving, distributed gaps, quasi‐distributed gaps, shaped windings, and
parallel windings. Few designs use these to their maximum potential.
Full models of twisting effects in litz are now available.
Aluminum wire can achieve lower loss than copper wire in cost‐limited
designs. This is an under‐utilized opportunity.
For MHz frequencies, litz strands are too big. Ways of using thin foil
effectively are under development, e.g. resonant designs, including WPT.
Winding loss analysis methods are available if not always applied well; core
loss modeling state of the art is less solid and new models are needed.
New core materials are valuable if the have low enough loss to offer
competitive performance factor at any frequency in the kHz or MHz range.
power.thayer.dartmouth.edu 55
Description of key references
Key references in high‐frequency power magnetics with an emphasis on publications from our group and a focus on discrete
components rather than chip‐scale microfabricated components; for our perspective on the latter see [1].
For windings, Zimmanck’s method can efficiently generate frequency dependent winding loss matrices for any geometry, 1D, 2D,
or 3D, and use them to predict loss for different nonsinusoidal waveforms in any number of [2]. This method applies very
generally, including to coupled inductors, wireless power transfer coils, etc. References cited in [2] provide more detailed
background, including [26,27]. See also [28]. A systematic approach to generating full models for loss and simulation for 1D
geometry is provided in [3]. To use 2D models effectively for 3D geometries such as E‐cores, the strategy in [25] can reduce the
error involved by a factor of 5.
Although the Dowell model is reasonably accurate, see the appendix of [9] for a simple correction that can enhance the accuracy.
Also useful in the appendix of [9] is a simple effective frequency approach to address winding loss with non‐sinusoidal windings.
Strategies to reduce proximity effect loss, using multiple thin layers or avoiding multiple layers, are compared in [6, 7, 8],
considering different types of optimization constraints. An overview of the most common implementation of thin layers to reduce
proximity effect loss, litz wire, is provided in [9]. A practical guide to using it is provided in [10], and the most complete model
including effects of details of twisting construction, is in [11]. Approaches for using thin foil layers beyond frequencies where litz is
practical are discussed in [12]. An implementation of these concepts for a resonant coil for applications such as wireless power
transfer is described in [13]. For other applications, thin foil layers can have capacitance issues; circuits designs that reduce the
voltage swing on the windings (e.g., [14]) can help reduce the impact of the capacitance.
The impacts of gap fringing and the quasi‐distributed gap technique for reducing these problems are discussed in [15]. This
reference includes data showing that a small gap is not effective for reducing the impact of fringing. With round‐wire or litz‐wire
windings, shaping the winding can allow excellent performance with a standard gap [16].
In inductors with substantial dc resistance, two windings in parallel can be a good choice for good dc and ac resistance[17]. It is
possible to extend this approach to applications in which the inductor carries a combination of line frequency ac current and high‐
frequency switching ripple, using, if needed, a capacitor to prevent low‐frequency current from flowing through the high‐
frequency winding [18]. A foil winding with a semi‐circular cutout region near the gap [19, 20, 21] can also be used to achieve a
favorable ac/dc resistance combination.
Although copper windings are most common, aluminum can offer advantages if cost or weight are important [22, 23].
Performance factor for magnetic materials is described and extended in [24], and data on performance factor is provided for many
materials in the MHz range. For coreloss with non‐sinusoidal waveforms, the iGSE model remains the standard method [4],
although some of its limitations are now known, as discussed in [5].
56
References, p. 1 of 3
[1] C. R. Sullivan, D. Harburg, J. Qiu, C. G. Levey, and D. Yao, “Integrating magneticsfor on‐chip power: A persepctive,”
IEEE Trans. on Pow. Electr., 2013.
[2] D. R. Zimmanck and C. R. Sullivan, “Efficient calculation of winding loss resistance matrices for magnetic components,”
in IEEE Workshop on Control and Modeling for Pow. Electr., 2010.
[3] M. Chen, M. Araghchini, K. K. Afridi, J. H. Lang, C. R. Sullivan, and D. J. Perreault, “A systematic approach to modeling
impedances and current distribution in planar magnetics,” IEEE Trans. on Pow. Electr.,, 31(1), pp. 560–580, Jan 2016.
[4] K. Venkatachalam, C. R. Sullivan, T. Abdallah, and H. Tacca, “Accurate prediction of ferrite core loss with nonsinusoidal
waveforms using only Steinmetz parameters,” in IEEE Workshop on Computers in Pow. Electr., 2002.
[5] C. R. Sullivan, J. H. Harris, and E. Herbert, “Core loss predictions for general PWM waveforms from a simplified set of
measured data,” in IEEE Applied Power Electronics Conference and Exposition (APEC), Feb. 2010, pp. 1048–1055.
[6] M. E. Dale and C. R. Sullivan, “General comparison of power loss in single‐layer and multi‐layer windings,” in IEEE Pow.
Electr. Specialists’ Conf., 2005.
[7] M. E. Dale and C. R. Sullivan, “Comparison of single‐layer and multi‐layer windings with physical constraints or strong
harmonics,” in IEEE International Symposium on Industrial Electronics, 2006.
[8] M. E. Dale and C. R. Sullivan, “Comparison of loss in single‐layer and multi‐layer windings with a dc component,” in
IEEE Ind. App. Soc. Ann. Mtg., 2006.
[9] C. R. Sullivan, “Optimal choice for number of strands in a litz‐wire transformer winding,” IEEE Trans. on Pow. Electr.,
vol. 14, no. 2, pp. 283–291, 1999.
[10] C. R. Sullivan and R. Y. Zhang, “Simplified design method for litz wire,” in IEEE App. Pow. Electr. Conf. (APEC), 2014,
pp. 2667–2674.
[11] C. R. Sullivan and R. Y. Zhang, “Analytical model for effects of twisting on litz‐wire losses,” in IEEE Workshop on
Control and Modeling for Pow. Electr. (COMPEL), 2014.
[12] C. R. Sullivan, “Layered foil as an alternative to litz wire: Multiple methods for equal current sharing among layers,”
in IEEE Workshop on Control and Modeling for Pow. Electr. (COMPEL), 2014.
[13] C. R. Sullivan and L. L. Beghou, “Design methodology for a high‐Q self‐resonant coil for medical and wireless‐power
applications,” in IEEE Workshop on Control and Modeling for Pow. Electr. (COMPEL), 2013, pp. 1–8.
57
References, p. 2 of 3
[14] M. Chen, K. Afridi, S. Chakraborty, and D. Perreault. “A high‐power‐density wide‐input‐voltage‐range isolated dc‐dc converter
having a multitrack architecture,” in IEEE Energy Conversion Congress and Exposition (ECCE), 2015.
[15] J. Hu and C. R. Sullivan, “AC resistance of planar power inductors and the quasidistributed gap technique,” IEEE Trans. on Pow.
Electr., vol. 16, no. 4, pp. 558–567, 2001.
[16] ] J. Hu and C. R. Sullivan, “Analytical method for generalization of numerically optimized inductor winding shapes,” in IEEE
Pow. Electr. Spec. Conf., 1999.
[17] A. van den Bossche and V. Valchev, Inductors and Transformers for Power Electronics. Taylor and Francis, 2005.
[18] C. Schaef and C. R. Sullivan, “Inductor design for low loss with complex waveforms,” in IEEE App. Pow. Electr. Conf., 2012.
[19] J. D. Pollock and C. R. Sullivan, “Gapped‐inductor foil windings with low ac and dc resistance,” in IEEE Ind. App. Soc. Ann. Mtg.,
2004, pp. 557–663.
[20] J. D. Pollock and C. R. Sullivan, “Modelling foil winding configurations with low ac and dc resistance,” in IEEE Pow. Electr.
Specialists’ Conf., 2005.
[21] W. Lundquist, V. Yang, and C. Castro, “Low ac resistance foil cut inductor,” in IEEE Energy Conv. Cong. and Exp., 2014, pp.
2182–2186.
[22] C. R. Sullivan, “Aluminum windings and other strategies for high‐frequency magnetics design in an era of high copper and
energy costs,” IEEE Trans. on Pow. Electr., vol. 23, no. 4, pp. 2044–2051, 2008.
[23] “Aluminum: The material of choice for transformers,” 2014, Siemens Industry, Inc.
[24] A. J. Hanson, C. R. Sullivan, and D. J. Perreault, “Measurements and performance factor comparisons of magnetic materials at
MHz frequencies,” in IEEE Energy Conv. Cong. and Exp., 2015; also early access in IEEE Trans. Pow. Electr.
[25] A. F. Hoke and C. R. Sullivan, "An Improved Two‐Dimensional Numerical Modeling Method for E‐Core Transformers", in IEEE
App. Pow. Electr. Conf., 2002.
[26] Xi Nan and C. R. Sullivan, “Simplified high‐accuracy calculation of eddy‐current loss in round‐wire windings,” in IEEE Pow.
Electr. Spec. Conf, 2004.
[27] C. R. Sullivan, “Computationally efficient winding loss calculation with multiple windings, arbitrary waveforms, and two‐ or
three‐dimensional field geometry,” IEEE Trans. on Pow. Electr., vol. 16, no. 1, pp. 142–50, 2001.
[28] D. C. Meeker, “An improved continuum skin and proximity effect model for hexagonally packed wires,” Journal of
Computational and App. Mathematics, vol. 236, no. 18, pp. 4635–4644, 2012.
58
References, 3 of 3
[29] Sobhi Barg, K. Ammous, H. Mejbri, and A. Ammous, “An Improved Empirical Formulation for
Magnetic Core Losses Estimation Under Nonsinusoidal Induction,” IEEE Trans. Pow. Electr. 32(3),
March 2017
[30] Benedict Foo, A. Stein, C. Sullivan, “A Step‐by‐Step Guide to Extracting Winding Resistance from an
Impedance Measurement”, APEC 2017, Poster session D09, paper 1925
[31] M. Chen, M. Araghchini, K. K. Afridi, J. H. Lang, C. R. Sullivan, and D. J. Perreault, “A systematic
approach to modeling impedances and current distribution in planar magnetics,” IEEE Trans. on
Pow. Electr.,, 31(1), pp. 560–580, Jan 2016.
[32] Spreen, J.H.; , "Electrical terminal representation of conductor loss in transformers," Power
Electronics, IEEE Transactions on , vol.5, no.4, pp.424‐429, Oct 1990. doi: 10.1109/63.60685.
[33] Sullivan, C. R., & Muetze, A. (2010). Simulation model of common‐mode chokes for high‐power
applications. IEEE Transactions on Industry Applications, 46(2), 884‐891.
[34] A. L. F. Stein, P. A. Kyaw and C. R. Sullivan, "High‐Q self‐resonant structure for wireless power
transfer," IEEE Applied Power Electronics Conference and Exposition (APEC), Tampa, FL, 2017, pp.
3723‐3729.
[35]P. A. Kyaw and C. R. Sullivan, "Fundamental examination of multiple potential passive component
technologies for future power electronics," IEEE 16th Workshop on Control and Modeling for Power
Electronics (COMPEL), Vancouver, BC, 2015, pp. 1‐9.
[36] C.R. Sullivan, “Prospects for advances in power magnetics” CIPS 2016 ‐ 9th International
Conference on Integrated Power Electronics Systems
59