SL Series XR Series TS Series MS Series MT Series

Vol. 5, No. 4 www.pels.org

December 2018

For your engineering success

Features

 Stateodular
24 M Multilevel Converters
of the art and future progress
Rainer Marquardt

 ultilevel Inverters for Grid-Connected

32 M
Photovoltaic Applications
Examining emerging trends
Xing Zhang, Tao Zhao, Wang Mao, Don Tan, and Liuchen Chang

42 Failure Analysis of the dc–dc Converter

On the cover
Modular multilevel converters have paved
the way for power electronics in very demanding
application areas.
image licensed by ingram publishing
MV ac drive image courtesy of Rockwell Automation
A comprehensive survey of faults and solutions
for improving reliability
Levy Ferreira Costa and Marco Liserre
52 A Critique of Little Box Challenge
Inverter Designs
Breaking from traditional design tradeoffs
Clint W. Halsted and Madhav D. Manjrekar

Departments
& Columns
66
4 From the Editor
8 President’s Message
13 Happenings
18 Entrepreneur Viewpoint
20 Passive Components
61 Patent Reviews
63 Expert View
66 Society News
84 In Memoriam
86 Book Review
88 Event Calendar
92 White Hot

Digital Object Identifier 10.1109/MPEL.2017.2788965

December 2018 z IEEE Power Electronics Magazine 1

IEEE Power Electronics Magazine
Editor-in-Chief IEEE Power Electronics 2019 Members-at-Large Advertising Sales
Ashok Bindra Society Officers Sibylle Dieckerhoff Walter Chalupa
12 Magnolia Ave. Alan Mantooth diecker@win.tu-berlin.de National Sales Manager
Nesconset, NY 11767 USA President Dan Kinzer Kerstin Chalupa
+1 631 471 5895 mantooth@uark.edu dan.kinzer@navitassemi.com International Sales Manager
bindra1@verizon.net Frede Blaabjerg Telephone: +1 973 835 7015
Marco Liserre chalupapels@aol.com
President-Elect/VP of Products liserre@gmail.com
Deputy Editors-in-Chief fbl@iet.aau.dk
Robert N. Guenther (Industry) Annette Mütze IEEE Power Electronics
VP-Product Development Braham Ferreira muetze@tugraz.at Society Staff
NWL Immediate Past President Mike Kelly
Nominations Committee Chair João O.P. Pinto Executive Director
Bordentown, NJ 08505 USA joaonofre@gmail.com
+1 609-298-7300 ext. 1245 j.a.ferreira@tudelft.nl m.p.kelly@ieee.org
RGuenther@nwl.com Mark Dehong Xu Donna Florek
Don F.D. Tan Technical Community
xdh@zju.edu.cn
John Shen (Academic) Senior Past President, PELS Program Specialist
Grainger Chair Professor Long-Range Planning 2020 Members-at-Large
d.florek@ieee.org
Department of Electrical Committee Chair Johan Enslin
jenslin@clemson.edu Becky Boresen
and Computer Engineering dong.tan@ngc.com Technical Community
Illinois Institute of Technology Mario Pacas Noriko Kawakami Program Specialist
10 West 35th Street, Suite 1600 Executive VP Global Operations kawakami.noriko@tmeic.co.jp b.boresen@ieee.org
Chicago, IL 60616 USA pacas@uni-siegen.de Brad Lehman Jane Celusak
+1 312 567-3352 lehman@ece.neu.edu Project Manager
zjohnshen@gmail.com Yan-Fei Liu
VP Technical Operations Helen Li j.celusak@ieee.org
Tom Keim (Contributing) yanfei.liu@queensu.ca hli@caps.fsu.edu Alicia Tomaszewski
+1 352 363 6688 Project Manager Transportation
tkeim@alum.mit.edu Luichen Chang Marco Rivera Electrification Community
VP Conferences marcoesteban@gmail.com a.tomaszewski@ieee.org
lchang@unb.ca Pradeep Shenoy Kellie Gilbert
Magazine Advisory Board
John Shen Peter Wilson pshenoy@ti.com Editorial Assistant/News Editor
Chairman Executive VP Standards kellie.gilbert@ieee.org
Illinois Institute of Technology P.R.Wilson@bath.ac.uk Technical Committee Chairs IEEE Periodicals
Don F.D. Tan Jian Sun Keiichi Hirose Magazines Department
Senior Past President Treasurer Communications Energy Systems 445 Hoes Lane, Piscataway, NJ
IEEE Power Electronics Society jsun@ecse.rpi.edu hirose36@gmail.com 08854 USA
Robert N. Guenther Jinjun Liu Yaow-Ming Chen Jessica Welsh
NWL Managing Editor
Executive VP Membership High Performance and Emerging
jjliu@mail.xjtu.edu.cn Technologies Geraldine Krolin-Taylor
Prasad Enjeti Senior Managing Editor
Texas A&M University Don Tan ymchen@cc.ee.ntu.edu.tw
Uday Deshpande Janet Dudar
William A. Peterson Division II Director Senior Art Director
E&M Power dong.tan@ngc.com Motor Drives and Actuators
udayd@ieee.org Gail A. Schnitzer
William Gerard Hurley Associate Art Director
National University of Ireland, 2018 Members-at-Large Rolando Burgos Theresa L. Smith
Galway Liuchen Chang Power and Control Core Production Coordinator
lchang@unb.ca Technologies
Mark Dehong Xu Mark David
William Gerard Hurley rburgos@ieee.org
Zhejiang University, China Sr. Manager Advertising and
gerard.hurley@nuigalway.ie Robert Pilawa-Podgurski Business Development
Ernie Parker Power Conversion Systems
Crane Aerospace and Electronics Katherine Kim Felicia Spagnoli
katakim@gmail.com and Components Advertising Production Manager
Annette Mutze pilawa@illinois.edu Peter M. Tuohy
Graz University of Technology, Ernie Parker
ernie.parker@crane-eg.com Sudip Mazumder Production Director
Graz, Austria Sustainable Energy Systems Kevin Lisankie
Dan Kinzer Grant Pitel mazumder@uic.edu Editorial Services Director
Navitas Semiconductor grantpitel@gmail.com
Alireza Khaligh Dawn M. Melley
Seung-Ki Sul Vehicle and Transportation Systems Staff Director,
sulsk@plaza.snu.ac.kr khaligh@ece.umd.edu Publishing Operations

IEEE prohibits discrimination, harassment, and bullying.

For more information, visit http://www.ieee.org/web/aboutus/whatis/policies/p9-26.html.

IEEE Power Electronics Magazine (ISSN 2329-9207) (IPEMDG) is published quarterly by the Institute of Electrical and
Electronics Engineers, Inc. Headquarters: 3 Park Avenue, 17th Floor, New York, NY 10016-5997 USA, Telephone: +1 212
419 7900. Responsibility for the content rests upon the authors and not upon the IEEE, the Society or its members. IEEE
Mission Statement: To educate,
Service Center (for orders, subscriptions, address changes): 445 Hoes Lane, Piscataway, NJ 08855-1331 USA. Telephone: inform, and entertain our community of
+1 732 981 0060. Individual copies: IEEE members US$20.00 (first copy only), nonmembers US$99.00 per copy. Subscrip- IEEE Power Electronics Society members
tion rates: Annual subscription rates included in IEEE Power Electronics Society member dues. Subscription rates available on technology, events, industry news, and
on request. Copyright and reprint permission: Abstracting is permitted with credit to the source. Libraries are permitted to
photocopy beyond the limits of U.S. Copyright law for the private use of patrons 1) those post-1977 articles that carry a code general topics relating to consumer elec-
at the bottom of the first page, provided the per-copy fee indicated in the code is paid through the Copyright Clearance Cen- tronics and to further serve and support
ter, 222 Rosewood Drive, Danvers, MA 01923 USA; 2) pre-1978 articles without a fee. For other copying, reprint, or republi- our Members in professional career devel-
cation permission, write Copyrights and Permissions Department, IEEE Service Center, 445 Hoes Lane, Piscataway, NJ
08854. Copyright © 2018 by the Institute of Electrical and Electronics Engineers Inc. All rights reserved. Periodicals postage
opment through tutorials and raising aware-
pending at New York and additional mailing offices. Postmaster: Send address changes to IEEE Power Electronics Maga- ness of engineering tools and technologies.
zine, IEEE, 445 Hoes Lane, Piscataway, NJ 08854 USA. Canadian GST #125634188 PRINTED IN THE U.S.A.

Promoting Sustainable Forestry

Digital Object Identifier 10.1109/MPEL.2017.2788966 SFI-01681

2 IEEE Power Electronics Magazine z December 2018

 Take the path
of least resistance

Increase the efficiency of your power electronics systems

with our ultra-low DCR power inductors

Our AGP Family of high current
power inductors feature flat wire
construction for exceptionally low
DC and AC resistance, letting your
power systems run cooler and more
efficiently.
They’re offered with inductance
values as high as 470 µH and
current ratings up to 108 Amps.
And all AGP Family inductors
meet stringent AEC-Q200 Grade 1
quality standards, making them
suitable for automotive and other
harsh environment applications.
AEC Visit www.coilcraft.com/AGP to
learn more and arrange a test drive.
Q200 Grade1 Free samples are always just around
the corner!

WWW.COILCRAFT.COM
 From the Editor
Investigating Developments in
Modular Multilevel Converters
T
 he modular multilevel con-
verter (MMC) has come a long
way since first proposed for
high-voltage dc (HVDC) applications
nearly 15 years ago. Ongoing improve-
ments and advances in MMC tech-
nology have extended the reach
of the MMC to many more ap­­plica­­
tions in the last 10–15 years, which
include flexible ac transmission sys-
tems and high-power motor drives.
To better understand the MMC, our
theme for this issue of IEEE Power
­Electronics Magazine is the modular
multilevel converter. Consequently,
we invited Prof. Rainer Marquardt
of the University of Bundeswehr,
Munich, Germany and a recipient of
the IEEE William E. Newell Power
Electronics Award for pioneering
the development of the MMC con-
cept and applications, to write an
article that covers the basics, the
state of the art of the MMC, and
future progress in the field.
Prof. Marquardt’s article “Modu-
lar Multilevel Converters” shows that
the MMC is enabling power electron-
ics in new ways to meet the needs
of demanding applications like mul-
titerminal HVDC, medium-voltage dc
(MVDC) networks, medium-voltage
drives, electric ships, and many other
similar applications. These are appli-
cations where neither conventional
Digital Object Identifier 10.1109/MPEL.2018.2875258
Date of publication: 19 December 2018
4 IEEE Power Electronics Magazine z December 2018
by Ashok Bindra
multilevel-voltage source converters conducted by the researchers. Based
nor thyristor converters have the abil- on this survey, the article describes
ity to meet the requirements. According failure analysis of the series resonant
to this article, due to dc–dc converter and
a dva nces i n M MC LLC converter. The
technology, fully elec- last feature article,
To better understand
tronic failure manage- “A Critique of Little
ment and protection of the MMC, our theme Box Ch a l lenge In-
associated equipment for this issue of IEEE verter Designs” by
in extended networks Power Electronics Clint W. Halsted and
has become feasible. Magazine is the Madhav D. Manjrek-
Likewise, it indicates ar of the Energy and
modular multilevel
that electronic current Power Conditioning,
converter.
and voltage limita- Energy Production
tion at the ac and dc and Infrastructure
sides controlled by Center, University of
the converters will be extremely valu- North Carolina at Charlotte, provides
able in future applications, especially a comparative analysis of the Little
for multiterminal HVDC. Box Challenge (LBC) inverter designs
The second article on multilevel presented by the finalists of the LBC
inverters is written by Xing Zhang competition. It presents a figure of
et al. In this article, “Multilevel Invert- merit that is used to illustrate why the
ers for Grid-Connected Photovoltaic gallium-nitride-based design from the
Applications,” several multilevel to- University of Illinois at Urbana–Cham-
pologies are discussed for photovol- pagne was able to achieve the highest-
taic (PV) applications. It also sum- density power conversion.
marizes the salient features of these
topologies. In conclusion, the article Columns, Society News, and More
suggests that the use of a common dc In the “Happenings” column, contrib-
bus, three-phase isolated cascaded H- uting editor Tom Keim investigates the
bridge, grid-connected PV inverter is search for improved materials for
a better solution. electrodes in lithium-ion batteries,
In addition, this issue presents primarily driven by new uses of the
two more interesting articles. The ar- battery. The “Passive Components”
ticle “Failure Analysis of the dc–dc column examines the right way of
Converter,” by Levy Ferreira Costa selecting power resistors. While the
a nd Ma rco L iser re of Ch r ist ia n- “Patent Reviews” column discusses
Albrechts-Universität zu Kiel, Ger- payments for utility patent infringe-
many, presents the results of a survey ment, the “Entrepreneur Viewpoint”
column provides a summary of the Oregon, 23–27 September, this issue tric powertrain for hybrid, electric,
proposals submitted by the 157 final- also celebrates the 30th anniversary and fuel-cell vehicles. Finally, this is-
ists of the Empower a Billion Lives of the founding of the IEEE Power sue’s “Event Calendar” provides a year’s
c o m p e t i t i o n . The Electronics Society listing of conferences and workshops.
focus of the “White (P E L S ). P r o f . Ge- Thank you for your continuous
Hot” column is on This issue also rard Hurley of the support. As we close our fifth year
the evolution of gate celebrates the 30th National University on a sound footing, I am looking
drivers, and the “Ex­­ of Ireland, Galway, forward to working with you next
anniversary of the
pert View” column presents minutes of year, bringing articles, columns, and
by Victor Veliadis, founding of the IEEE a town hall meeting news from around the world that is
chief technology Power Electronics at ECCE 2018, which of value to practicing power elec-
o f f i c e r of Power­ Society. was specially orga- tronics engineers. To serve you even
America, presents a nized in celebration better going forward, we are also
strategy for acceler- of the 30th anniver- looking for your feedback, which
ating the commercialization of wide- sary of PELS. The “Society News” helps make this magazine a valuable
bandgap power devices. articles look at activities of PELS resource for working power elec-
Besides providing an overview of Chapters around the world, includ- tronics engineers from around the
the IEEE Energy Conversion Con- ing information on an internation- world. I wish you all the best in the
gress and Exposition 2018 (ECCE al workshop on energy harvesting coming year.
2018), which was held in Portland, in Ireland and a book review on elec- 

OPTIMIZED HIGH-FREQUENCY MAGNETICS Agile Magnetics

Boost your new, higher
switching-frequency power
supply design to the top
echelon of performance,
value, and features, with
advanced magnetic
components from Agile
Magnetics.
With advanced design
knowledge and the
largest choice of high-
frequency materials and
manufacturing techniques
at our disposal, Agile will
optimize your magnetic
components’:
• Package size and shape
• Power efficiency
• Performance margin
• Reliability
• Manufacturability
• Thermal rating
• Cost
From rapid prototype
through production,
our ISO9001 / AS9100 –
certified operation will
propel your power products
to excellence.

Agile Magnetics, Concord, NH • www.agilemagco.com • 1-800-805-8991 • Since 1992

6 IEEE Power Electronics Magazine z December 2018

BridgeSwitch™

Integrated Half-Bridge BLDC Motor Drive IC

HV+

HD

• 98.5% efficient - No heatsinks

VDD HS
BPH Drive

•
XH

Hardware protection simplifies

HB

software
SM
FAULT
/INH LS
System Drive
INL
MCU &
IPH
Control
BPL
• Superior thermal performance
ID SG XL LS

• Works with any microcontroller

HV-
PI-8314-091718

www.power.com
President’s Message
Victory Lap? No, Pass the Baton!
T
his is my last column as pre­­
sident of the IEEE Power
Electronics Society (PELS),
and while it is tempting to take a
victory lap, that simply is not the
nature of our organization. Rather,
as my predecessors have done, I will
hand off the baton into the capable
hands of our next president, Frede
Blaabjerg. However, it is an appro­
priate time to give you a summary of
how far we have come in the last
two years, thanks in large part to
our volunteers and staff.
The Society has grown over the
last few years, currently heading to­
ward 9,500 members, and is on track
to surpass 10,000 next year! We have
consistently sustained a better than
5% growth rate and also enjoy a high
member retention rate. Because di­
versity and inclusion are cornerstone
goals of our strategic plan, two of our
fastest increasing segments include
Young Professionals and Women in
Engineering groups, but we still have
so much more growth to realize. This
takes all of us working together pro­
moting the creation of Chapters, de­
scribing the benefits of membership,
and getting involved! People often ask
if bigger is better for a Society like
ours. We do not seek growth for the
sake of growth, but rather we want
to reflect the number of professionals
that work in this field. To this end, we
will be conducting a market survey
Digital Object Identifier 10.1109/MPEL.2018.2875257
Date of publication: 19 December 2018
8 IEEE Power Electronics Magazine z December 2018
in 2019 to ascertain the total number
working in our field of interest around
the world, so that we can better target
our efforts and achieve a growth path
that is informed by these numbers.
Our Technical Committee (TC) struc­
ture has evolved to increase our volun­
teer engagement, improved conference
quality and attendance, and supplied
our journals with a steady stream of
quality and timely special issues. Al­
most all of our TCs have established
technical achievement awards to rec­
ognize the contributions of our PELS
members. Furthermore, they have es­
tablished student travel grants to sup­
port students to attend conferences,
network, and begin to establish their
own professional roots as our future
members. Finally, Technical Operations
is actively expanding to embrace new
and emerging technical areas such as
cybersecurity, wireless power, and grid-
connected power electronics—often in
partnership with other organizations
and sometimes on our own. Our TC
structure has served us well for almost
ten years, but to remain nimble and re­
flective of a changing field, we will also
be revisiting this structure to see if new
TCs are needed or if some existing ones
require updating.
Our flagship publication, IEEE
Transactions on Power ­Electronics
(TPELS), is one of the top in the elec­
tronics field across the entire IEEE
portfolio in terms of impact fac­
tor each year. Now more than 12,000
pages strong each year, it serves the
Society’s need to publish the world’s
by Alan Mantooth
most significant advances in the field.
The recently started IEEE Journal
of Emerging and Selected Topics in
Power Electronics (JESTPE) has al­
ready established a solid impact factor
that is growing. It is rising alongside
the tremendous growth of our field.
Big thanks go to Brad Lehman, Henry
Chung, and Don Tan in the chief editor
roles as well as to all of the associate
editors and reviewers for these publica­
tions! I recently appointed, and the Ad­
ministrative Committee (AdCom) ap­
proved, new editors-in-chief (EiCs) for
TPELS, IEEE Power Electronics Let-
ters, and JESTPE because the current
editors have come to the end of their
terms. The new incoming editors are
■■TPELS EiC: Yaow-Ming Chen
■■IEEE Power Electronics Letters
EiC: Yunwei (Ryan) Li
■■TPELS Co-EiCs: Hui (Helen) Li,
Paolo Mattavelli, and Chun Taek Rim
■■TPELS e d i t o r- a t - l a r g e : S u d i p
Mazumder.
You will note that we have changed
the structure of the chief editorial board
to include three co-EiCs to help the
EiC with the increasing workload of
TPELS. The new EiC of JESTPE will
be Ojo Olorunfemi from the IEEE
Industry Applications Society (IAS)
because we share this journal 50–50
with IAS. We will soon name a co-EiC
from PELS to step in and assist Ojo.
This person is likely to become the
next EiC of JESTPE.
Our conferences are truly global
i n their presence. With the IEEE
Energy Conversion Congress and
Exposition (ECCE) Global Meeting
Series, which includes ECCE North
America, ECCE Europe, ECCE Asia
a nd more recent ly t he Sout her n
Power Electronics Conference, we
have worked hard to create a global
footprint to serve our members. Fur­
ther, the TCs have key workshops
or conferences, such as the Interna­
tional Symposium on Power Elec­
tronics for Distributed Generation
Systems, the Workshop on Control
and Modeling for Power Electronics,
and the Workshop on Wide Bandgap
Power Devices and Applications,
that are outstanding forums for dis­
seminating the latest R&D. In fact, I
traveled to six continents in 2017 to
cover only a fraction of the confer­
ences that we sponsor or cospon­
sor in power electronics. This field
is booming! To the many volunteers
who help to make these events so
successful, I want to thank you for
your service.
Speaking of global presence, we
have worked to increase our presence in
many regions, but most notably in South
America and China, during my term.
This will continue region by region
around the world in the coming years.
Both our conferences and publica­
tions serve as financial cornerstones
for the Society. Through these activi­
ties, we have been able to turn invest­
ments back to the members such as
our PELS Mentorship program, which
will begin a worldwide rollout in 2019
after two years of piloting in the United
States, the International Technology
Roadmap for Wide-Bandgap Semicon­
ductors and other standards activities,
the Empower a Billion Lives competi­
tion, the Cyber-Physical Security Ini­
tiative, Design Automation for Power
Electronics, and others. These offer
you, the membership, opportunities to
engage, network, learn, and grow both
personally and professionally. All of
these either started or took off during
the last two years! Plus, we successfully
launched our first-ever Members Town
Hall meeting at ECCE North America in
Portland, Oregon, where we were cele­
brating the tenth ECCE and PELS’ 30th
anniversary as a Society. You can view
the recording of the town hall meeting
on our website. We intend to continue
the Members Town Hall meeting going
forward to interact with the member­
ship more regularly. If you have top­
ics or ideas for this interaction, please
send them to myself or Frede, and we
will try to include them. Speaking of
anniversaries, the IEEE International
Telecommunications Energy Confer­
ence celebrated its 40th this year in
Torino, Italy.
Another key development that you
will observe in 2019 is the first-ever
PELS Day, which the AdCom estab­
lished as 20 June each year to mark the
date that PELS was approved for be­
coming a Society from an IEEE coun­
cil. The Society became effec­t ive on
1 January 1988. So, on 20 June 2019
a variety of global events will be held,
so watch for them on the PELS website.
To this end, because of our core
values of inclusivity and diversity, we
established “Lunch with the President”
 forums to hear from various member
segments including young profession­
als, women in power electronics, in­
dustry members, and South American
professionals, so far. I can tell you that
we are always interested in hearing
from others, so feel free to reach out so
that further events can be organized.
Finally, I would like to bring aware­
ness to our membership about the
role PELS has taken in helping to es­
tablish and nurture the IEEE Trans­
portation Electrification Community
(TEC). TEC itself boasts more tha n
9,500 members and is a growing tech­
nical community.
TEC was created to be the central
hub of focus for transportation elec­
trification activities across 11 IEEE So­
cieties and the IEEE Standards Asso­
ciation, acting as the key connection
point for external stakeholders. While
this initiative began as an IEEE-level
activity, PELS was chosen to be the
home Society for TEC because of
both its global presence and impor­
tance technically, but also because
Vector Network Analyzer Bode 100
Use Bode 100 to measure:
• impedance of passive components
• input impedance and output impedance
• control loop stability (loop gain) of converters
• and much more from 1 Hz to 50 MHz...
Benefit from a
• very wide frequency range
• multifunctional hardware
• easy-to-use software
www.omicron-lab.com
2018-01-25_Halfpage.indd 1
PELS has been so well run through
the years. Many thanks goes out to the
hundreds of volunteers who have led
the Society to this point.
TEC embodies a global commu­
nity of engineers, academics, industry
partners, and passionate practitio­
ners for the electrification of rail,
aerospace, marine, and automotive
platforms; autonomous artificial intel­
ligence; manufacturing best practices;
standards development; and research
information sharing for coordinat­
ed applications.
I have been asked by some students
around the world if it really is a good
idea to go into power electronics as a
profession. The question stems from
the commoditization that can occur in
electronics. Given the breadth of our
field, it might be easy to conclude that
some areas would become that way,
but I feel that we are in a golden era of
power electronics and that it is a very
exciting time to be entering this field.
There is so much on the table for us
to realize, and there are more break­
throughs around the corner to keep
this trend going! To those students, I
say that this is definitely a rich field to
pursue for a career.
In conclusion as my last message
as president, let me reiterate how hum­
bled and honored I am to have been
given the opportunity to lead this great
Society. It has been hard work but a
true highlight of my career. I feel that I
have left it better than I found it, and
this has been true of each and every
president we have had in our 30 years. I
have had some great officers, staff, and
Advisory Committee members to work
with to see this mission through, and to
all of them I owe a great deal of thanks!
When I took office, I showed a picture
of Usain Bolt running with a baton in
a relay race. I’ve run hard and fast, and
now I hand the baton off—still running
at full speed. I am confident this pace
will continue under our future leader­
ship. Thank you for giving me the op­
portunity to serve as your president.
Happy holidays!

2018-01-26 09:55:56

by Tom Keim
New Uses Spur Lithium-Ion Battery
Research and Development
D
riven in part by increased
interest in electrification of
land transportation and by
the increased use of wind- and solar-
generated electricity, lithium (Li)-ion
battery research and development
(R&D) is exploring many innova-
tions. This article discusses poten-
tial new directions in the Li-ion bat-
tery industry. It starts with a few
paragraphs about the basics of the
technology and the evolution of the
product line and then offers a dis-
cussion of future directions.
Both electrodes in a Li-ion battery
are complex structures consisting of
active materials, binders, and current
collectors. In both electrodes, the
active materials have a crystal struc-
ture that comprises pronounced lay-
ers. Today, the active materials are
one of several complex metal oxides
for the cathode and graphite for the
anode. The Li is stored between lay-
ers in both electrodes. As the battery
charges and discharges, Li ions are
cycled back and forth between the
electrodes and physically move from
one electrode to the other. The cath-
ode is the positive terminal, and the
anode is the negative terminal.
There are a number of different
chemical formulations for the metal–
oxide cathodes, which enable the
design of batteries with usefully
­different properties. Further discus-
Digital Object Identifier 10.1109/MPEL.2018.2873994
Date of publication: 19 December 2018
sion of these differences is not within
the scope of this article.
The first commercial Li-ion battery
was introduced by Sony in 1991. Early
applications included cell phones, lap-
top computers, and some camcorders.
In these applications, the high specific
energy, combined with the low self-
discharge rate, and acceptable cycle
life were primary attractions. The tar-
get products had moderate expected
product life, and battery replacement,
while undesirable, was not unthink-
able. Wikipedia reports that, by 2012,
approximately 660 million cylindrical
Li-ion batteries were produced. Most
of these were 18650 cells, so named
because they are nominally 18 mm in
diameter and 65 mm long. For many
years, 18650 cells were the mainstay
of the laptop market, but by 2012, that
industry was migrating to thinner
form factors, which could not accom-
modate the 18-mm diameter. Today,
18650 cells are still produced in great
quantity. The Tesla Model S and Model
X battery-electric automobiles use
battery packs built up of thousands of
these cells, connected both in series
and in parallel to produce the desired
battery capacity and v­ oltage.
Dr. Brian Barnett is a longtime
industry contributor and observer.
Over his career, he has worked for
and led major battery activities at
Arthur D. Little, Tiax, and CAMX
Power LLC. Most recently, he has
established a small consultancy, Bat-
tery Perspectives LLC. Dr. Barnett
December 2018 z IEEE Power Electronics Magazine 13
Happenings
reports that, today, few (if any) new
laptop computers use cylindrical
cells; instead, they are built with thin
f lexible rectangular cells called
pouch cells. Dr. Barnett attributes the
continued production of 18650 cells
to other applications, including the
aforementioned Tesla models and
probably many rechargeable electric
power tools. Other, smaller consum-
er markets have adopted the 18650
cell, including some e-­cigarettes.
In addition to cylindrical cells and
pouch cells, other cells are produc­
ed with generally rectangular rigid
cases. Many automotive battery packs
are made up of such cells.
Dr. Barnett points out that a mod-
ern 18650 cell can deliver about three
times the energy of an early produc-
tion cell (approximately 3.5 Ah, or
about 13 Wh). Also, the power avail-
able from such a cell has increased
substantially more than a factor of
ten over the same period. Some of
these advances reflect a reduction in
conservatism, justified by accumulat-
ed experience. But much of the differ-
ence is attributable to advances both
in materials and in cell design.
According to Dr. Barnett, further
improvement, relying principally on
improved cell design with existing or
similar materials, is approaching
a ceiling. The technical community
believes that the next jump in energy
density will be achieved by more fun-
damental changes. For this reason,
R&D in Li-ion battery technology is, if
anything, more robust than in the
past. Certainly, both the electrifica-
tion of transportation and the conver-
sion to intermittent renewable gener-
ation of utility power are large poten-
tial markets, which may be enabled
or accelerated by continued advance
in battery capability. Furthermore,
even low estimates of the future
demand for batteries for vehicles will
14 IEEE Power Electronics Magazine z December 2018
require more than a doubling of the
production rate for batteries, and
storage for the power grid, when it
comes, will be a larger market still.
The U.S. Department of Energy
Vehicle Technologies Office issued a
progress report on Department of
Energy-funded battery R&D for 2016
that exceeds 1,000 pages. Looking at
the index, it is evident that there are
6/3/10 4:36:05 PM
many battery technologies under
investigation, not only Li-ion batter-
ies. But it appears that substantially
more than half the report is devot-
ed to Li-ion technology. The report
for 2017 is even longer. Much addi-
tional work is funded from other
sources, including investment in
start-up companies.
While there are already a large num­­­
ber of different formulations for cath-
ode materials on the market, more
are being developed. Dr. Barnett is
included among the authors of a
recent paper “High-Nickel Cathode/
Graphite Anode Cells for Diverse
DOD Applications.” This paper was
presented in June 2018 at a long-run-
ning conference series for developers
of power supplies for military uses
[1]. The high-nickel cathode material
that is the subject of the paper was
developed and patented by CAMX
Power and licensed for c­ ommercial
production by global battery materi-
als companies. The paper compares
cells with this cathode material to
cells with other common materials,
on the basis of specific energy, ener-
gy density, and specific discharge
capacity. The differences are mostly
small, but in every comparison, the
new cathode is superior.
There is also a substantial ongoing
body of work attempting to develop
higher-performance anodes. Unlike
cathode technology, where different
active materials are preferred for dif-
ferent uses, almost all commercially
produced cells today use very similar
technology based on graphite. The
layered structure of graphite accom-
modates the storage of Li when the
cell is in the charged state and the
release of ions to travel to the cath-
ode during the discharge cycle. Much
of the current work is directed to
substituting silicon for graphite. The
potential is huge. It takes six carbon
atoms to accommodate one Li atom,
whereas one silicon atom can accom-
modate four Li atoms.
But in absorbing that much Li, a
silicon structure grows in volume by a
factor of more than three. Few materi-
als can undergo such a large volume
change reversibly, and bulk silicon is
not among them. Today, there are a
number of start-up companies pursuing
a way forward to use silicon in anodes.
One of these is Amprius Corporation.
The company is working to commer-
cialize a process developed by Prof. Yi
Cui at Stanford University, California,
whereby silicon nanowires are used to
provide host sites for the Li while pro-
viding the flexibility to accommodate
the volume change without destruction.
The company has encouraging results
and substantial funding but does not
appear to have made any major
announcements in recent years.
An apparently balanced assess-
ment appeared in the magazine Joule
in 2017. A group of authors from the
Ulsan National Institute of Technolo-
gy in South Korea presented a per-
spective article, “Confronting Issues
of the Practical Implementation of Si
Anode in High-Energy Lithium-Ion
Batteries.” The article discusses the
reasons why silicon offers potential. It
also indicates that there are a number
of ongoing problems that need to be
resolved before the technology’s
potential is realized. This work has
been ongoing for 20 years. Yes, things
are happening in silicon anode
research, but the technical progress is
not currently disrupting the market.
The modern Li-ion battery was, to
a large degree, made possible by the
development of graph ite -ba sed
anodes. Much of the previous research
focused on metallic Li for the anode.
Li metal is an excellent source of Li
ions, but cycle life and safety proved
to be problems. When graphite-based
anodes were developed, interest in
using metallic Li diminished substan-
tially. Nevertheless, the search for
improved anode materials is leading
to renewed interest in developing
metallic Li anodes, even as most
developers acknowledge that more
work needs to be done before such
systems can be brought to commer-
cial fruition.
Efforts continue to make Li-ion
batteries safer. The principal safety
issue is that mechanical damage,
external overheating, or malfunction
in normal service can cause a ther-
mal runaway event. This possibility
requires attention during system
design to limit the consequences of
such events to acceptable levels and
to reduce the resulting ongoing costs
to produce safe systems. Malfunc-
tion in normal service is controlled
principally by adequate quality con-
trol during the production of the
cells. It is possible to produce cells
in great quantity without malfunc-
tions in normal service. But as long
as the possibility of thermal run-
away due to other causes remains,
safety will be an important issue.

We engineer optimally designed

custom magnetics, while offering thousands
of part numbers standard, off the shelf.
We can help you find one that meets your needs today.
and it can be on your desk, tomorrow.

Standard Products Custom Products

Power Current Sense Wall Plug In Power Supplies Switching Power Alternative Energy
Transformers Transformers Transformers & LED Drivers Supply Inductor

Audio Inductors Medical-Grade Medical-Grade Low- Power Supply Professional Audio

Transformers and Chokes Isolated Sources LeakageTransformers Precision Toroidal Impedance Matching
Transformer Transformer

Triad Magnetics
460 Harley Knox Blvd, Perris, CA 92571
Tel: 951.277.0757 Fax: 951.277.2757
Email: info@triadmagnetics.com www.triadmagnetics.com

December 2018 z IEEE Power Electronics Magazine 15

 Thermal runaway events can arise in several funda-
mentally different ways. An internally grown short circuit
is typically the result of an impurity introduced during
manufacture. But the other causes of thermal runaway
can be attributed to a range of events, almost all of which
can be characterized as cell abuse. Safety against thermal
runaway is thus highly correlated with abuse tolerance.
Long-term overcharging is one example of a condition
that can give rise to thermal runaway. But even if a cell is
removed from the circuit, exposure to a moderately hot
environment can cause a problem, as can mechanical abuse
of the cell, e.g., by crushing or puncture. The detailed
description of the steps in the process of failure varies sig-
nificantly among these alternatives, as does the range of
precautions against them. But the end result is often similar.
Much is made of the possibility of the internal conver-
sion of the stored electrochemical energy to heat. This is
significant, but it is far from the whole story. There are a
number of complex chemicals in a cell, and they begin
exothermic thermal decomposition at temperatures as
Custom Discrete and Multi Circuit Filters and EMI connectors for all your filtering requirements low as 80–100 °C. More significantly, the electrolyte is a
combustible substance with a specific heat of combustion

WEMS Electronics
Contact us today for all of your filter solutions. WEMS Electronics is a full service turn-key Small Business
not that far below petroleum-based fuels.
The total available heat of combustion can be several
with an extensive AS9100D/ ISO-9001:2015 registered complex specializing in state-of-the-art EMI custom
filters, engineering and manufacturing services. WEMS designs and manufactures high reliability multi-circuit
times the electrochemically stored energy. If the event is
& discreet feed-through (F/T) input, output, power, signal line filters and filter connectors for military, aero-
space, space, and commercial applications. initiated by a rupture of the cell, it is common for the elec-
4650 West Rosecrans Avenue Hawthorne, California 90250
Phone 310-644-0251 www.wems.com trolyte to burn, substantially contributing to the overall
event. Even if the cell is not ruptured at the onset, internal
heating can result in pressurization of the contents, which

Tape Wound Toroidal & Cut Cores can result in a flame-thrower event should the cell fail
under pressure. There is ongoing work on alternative elec-
For Sophisticated Electronic trolytes that are not flammable, but currently none have
Components had a combination of ­properties that has resulted in dis-
placement of the incumbent materials.
It is common for a thermal runaway result in one cell
to start the process in nearby cells in a battery pack,
often resulting in the failure of a second cell after a delay
of seconds or even minutes. It is possible for a cascading
failure to result in loss of the entire battery pack.
Another interesting development may have positive
implications for the safety of future batteries. Ionic Mate-
rials, a Massachusetts start-up company with ties to
Tufts University, has developed useful quantities of a
Magnetic Metals Achieve Near-Perfect solid polymeric material that replaces both the electro-
Uniquely Supplies: Magnetic Circuits in: lyte and the separator in an otherwise conventional bat-
• All types & thicknesses of soft • Power/navigation systems tery. The material is far less flammable than present elec-
magnetic materials with… • Transformers/GFCI’s trolytes and is claimed to allow the passage of Li ions as
• Advanced magnetic & • Power supplies well as a conventional electrolyte does.
performance properties. • Voltage monitors This development is the subject of an impressive seg-
ment in an episode of the long-running PBS television
• Standard & custom sizes, Contact Our Technical show Nova [2]. It is almost irresponsible to recommend it,
shapes and quantities. Team Today!
because it contains egregious violations of standard labo-
• Nanocrystalline & amorphous New Jersey & California
888-892-8392 ratory safety precautions, but the results are too impres-
alloys; silicon- nickel- and
cobalt-iron alloys. www.MagneticMetals.com sive to ignore. What is clear is that prototypes of pouch
cells the size of a digital tablet device are extremely resis-
75+ YEARS tant to thermal runaway due to puncture by metal objects.
OF CUSTOMER
SERVICE
Conspicuously absent from the dis-
cussion, and also from the company
website, are any claims that cells with
the new material have been built that
are competitive with conventional
cells on the basis of specific energy or
any of the many other properties on
which batteries are judged.
Even farther from being proven, at
least in public, is the assertion that
the new solid electrolyte will allow
the use of a metallic Li anode. It is
widely accepted that if Li metal can
be used, the specific energy of batter-
ies could be substantially advanced,
probably rendering both graphite and
silicon anodes obsolete. Ionic Materi-
als claims that specific energy can be
doubled by such a substitution. There
are historical reasons to be cautious
about the safety of a metallic Li anode
and other reasons to question wheth-
er such a large increase in energy per
unit mass would be achievable. But a
safe doubling of specific energy would
likely be a major market success if it
were to be achieved.
It is apparent from credible public
sources that Ionic Materials, like Ampri-
us, is a serious company, with serious
investors and capable employees,
working on an important problem. It
appears that the company currently has
enough funding and that it is visible and
well connected enough to have access
to more if progress warrants it.
Likewise, it has been established
that Li-ion research is being conduct-
ed on innovative anodes and cath-
odes, and literally on everything in
between. The industry is strong and
growing, and it is reasonable to
expect that changes in the capabili-
ties of Li-ion batteries will continue.
About the Author
Tom Keim (tkeim@alum.mit.edu) is a
late-career engineer and a longtime
December 2018 z IEEE Power Electronics Magazine 17
Member of the IEEE. His specialty is
high-performance electromechanical
systems and the power systems that
drive and control them. He has work­­
ed for a worldwide conglomerate, for
a small (50 employees) innovative
research and development company,
for a major research university, and
for an engineering consulting compa-
ny. He has 50 publications and 11 pat-
ents and is currently active as an
author, inventor, and consultant.
References
[1] B. Barnett, et al., “High-nickel cathode/graph-
ite anode cells for diverse DOD applications,”
presented at the 48th Power Sources Conf., Den-
ver, Colorado, 2018.
[2] Nova. (2017, Feb. 1). Search for the super
battery. Public Broadcasting Service. [Online].
Available: www.pbs.org/video/nova-search-super-
battery

Entrepreneur Viewpoint
I
t’s expensive to be energy poor.
A villager in rural Kenya spends
80 times more for lighting and
400 times more for phone charging
per energy unit than an average
American or European, often spend-
ing more than 35% of personal net
income on minimal lighting with a
kerosene lantern. Other activities
requiring energy are often unafford-
able. Addressing basic energy pover-
ty can bring about broad individual
and societal benefits, including
increased productivity, enhanced
educational outcomes, better health,
and greater food security.
The list of potiential benefits goes
on and on. Access to energy means
greater access to refrigeration, which
makes for more effective immuniza-
tion programs. It also means greater
access to communications tools, to
such services as online banking and
thus greater financial inclusion, and to
many digital tools and thus the many
benefits of digital technologies. Fur-
thermore, reduced energy poverty fos-
ters more efficient manufacturing and
distribution of goods, increases civic
engagement, and decreases emissions
of carbon dioxide.
Despite sustained heroic efforts
from numerous national govern-
ments, aid organizations, and philan-
thropic initiatives, 3 billion people
around the globe live in extreme
energy poverty, including 1.1 billion
people with no access to electricity,
Digital Object Identifier 10.1109/MPEL.2018.2874613
Date of publication: 19 December 2018
18 IEEE Power Electronics Magazine z December 2018
Update on the Empower a
Billion Lives Initiative
severely diminishing their ability to
maintain robust health, attain a good
education, and earn a living wage. To
get these people to a reasonable level
of energy access so they have a fair
chance of achieving their aspirations
requires initiatives on a giant scale.
“More of the same” is not the answer.
In keeping with the IEEE’s mission
of “advancing technology for humani-
ty,” the IEEE Power Electronics Soci-
ety (PELS) has launched a major glob-
al initiative: Empower a Billion Lives
(EBL). This is a recurring biennial
global competition to crowdsource
innovation and the development or
demonstration of new regionally rele-
vant, holistic, scalable, and economi-
cally viable ways to address the prob-
lem of energy poverty. EBL is expected
to leverage 21st-century technologies
that are exponentially declining in
price and have been adopted rapidly
around the world.
The first competition, EBL 1, was
launched in September 2017 at the
IEEE Energy Conversion Conference
and Exposition (ECCE) and is present-
ly ongoing. The online round, the first
stage of the competition, which closed
on 31 August 2018, attracted tremen-
dous attention with 459 teams re­­
gistered from 71 countries and with
157 final proposals submitted. The
teams represented a wide range of
organizations, including mature oper-
ating companies, start-ups, and univer-
sities. Among the teams were 71 from
Africa, reflecting the intense interest
in ad­­dressing the problem of energy
poverty on that continent.
by Deepak Divan
Qualifying teams will move to the
regional rounds, with five scheduled
over the next few months: in Shenzhen,
China (with the IEEE International
Power Electronics and Application
Conference and Exposition), Chennai,
India (with the IEEE International
Conference on Power Electronics,
Drives, and Energy Systems), Johan-
nesburg, South Africa (with a special
workshop), Seville, Spain (with a spe-
cial workshop), and Atlanta, Georgia
(with the IEEE Decentralized Energy
Access Solutions Workshop; visit www
.empowerabillionlives.org for more
details). Each location will have 10–15
teams competing for local awards that
will help them get ready for the field-
evaluation part of the competition and
the global finals to be held along with
ECCE in Baltimore, Maryland, in Sep-
tember 2019. The regional and final
rounds will feature team presentations
to a panel of judges, a “Shark Tank”-
style defense of their proposals, and a
poster presentation to the broader
group of attendees.
The teams are being judged on
technology, business, and social im­­
pact by a panel of 90 reviewers and
judges from around the globe. Unlike
typical competitions, which are focus­
ed on a specific narrow objective, EBL
encourages holistic thinking and seeks
novelty in technological, business, or
social factors. The fundamental chal-
lenge is to ensure that the teams align
with objectives that promise technical
and economic viability and address is­­
sues related to the effects of proposals
on society. The technical evaluation
and field trial will provide an unbiased data-based
validation of the performance of the proposal as
well as provide proof of end-customer acceptabili-
ty. This process of analyzing the projects for tech-
nical and market risk can dramatically reduce
risk and encourage investment in the new ideas.
Competing teams have proposed cutting-
edge concepts, such as self-organizing grids that
scale arbitrarily, solar “Avon” ladies for last-mile
distribution and support, income generation
through the use of lowest-cost energy, cloud-
based transaction platforms, multimedia devic-
es as energy and connectivity platforms, and
modular stackable solar home systems. Some of
these ideas are ahead of what we see presented
in IEEE journals and conferences and indicative
of the rapid pace of change and transformation
occurring in this sector. Proposals being devel-
oped definitely have a chance of changing the
energy world as we know it.
Meanwhile, IEEE PELS has launched a major
new initiative, the Energy Access Working Group,
which is responsible for all humanitarian initia-
tives within PELS. Traditionally, the PELS commu-
nity has not given energy access much attention.
Yet it is clear that initiatives in ensuring energy
access can provide a glimpse of what a future
energy infrastructure may look like: decentralized,
efficient, and inexpensive. A new annual event, the
IEEE Decentralized Energy Access Solutions
Workshop, provides the forum for disseminating
and presenting new ideas for energy access. EBL
provides the main global initiative by the IEEE
PELS to make a difference.
Please join the EBL community to stay abreast
of ongoing news and developments. Members
can support teams through assistance from
employers or through personal donations to
EBL. The IEEE is in the process of rolling out a
crowdfunding platform to make it easy to con-
tribute. Member support is very much needed
and will make a big difference.

About the Author

Deepak Divan (ddivan@gatech.edu) is a Nation-
al Academy of Engineering member, IEEE Fellow,
and past president of the IEEE Power Electronics
Society. He is a professor and director of the Cen-
ter for Distributed Energy at Georgia Tech. He
also serves as founder and chief scientist of Var-
entec, a company focused on grid edge control
that is funded by a clean-tech venture capital firm,
Khosla Ventures, and investor Bill Gates. Varentec
is his third start-up company.

Passive Components
S
electing the correct resistor
for a specific task can be a
mind-breaking puzzle, espe-
cially in applications with pulses of
high power when average power is
relatively modest. On the one hand,
the resistor must be big enough to
prevent risks or hazards. On the
other hand, the resistor should be
as small as possible to save space
and reduce costs. This article ex­­
plains the steps necessary to make
a wise selection.
Challenge
In power electronics, resistors are
used in many applications. They
are found, for example, in invert-
ers, devices for charging and dis-
charging capacitor banks, filters
Digital Object Identifier 10.1109/MPEL.2018.2874614
Date of publication: 19 December 2018
Thermal Model of an Alpha Resistor
RS
50
30
RM
R0
CS
1
15
IL
CM
Resistor Wire
FIG 1 Each resistor has its own unique thermal model.
20 IEEE Power Electronics Magazine z December 2018
Power Resistor Selection,
the Right Way
as snubbers, drive systems as brake
r e s i s t o r s , and wind turbines as
low-voltage ride-through energy-
dump resistors.
A resistor is, from an electrical
point of view, the simplest compo-
nent. In the field of electricity, only
a handful of variables need to be
calculated: current, voltage, ohmic
value, power, and energy. Although
each variable is easy to figure, the
appropriate values might not be so
easy to determine. Because, for some
applications, demands on electrical
systems are very difficult to predict.
In designing a dump application for
wind turbines, for example, an en-
gineer may need a crystal ball. How
often will there be a grid fault, and
how long will it last? National power
grid or power company grid codes
help us anticipate a possible “worst
case.” Still, determining the best re-
sistor for an application is tricky.
R1 R2 R3 CN
2 3 N N +1
C1 C2 C3 CN
CT
0
by Wim van Dijk and John Siefken
Underdimensioned resistors can lead
to excessive extra costs due to down-
time and repair or, in worse cases,
injury or death. However, overspeci-
fied resistors add to costs and take
up extra space.
An important step in defining the
specifications is analyzing risk. What
happens if the resistor breaks down?
Must it be overspecified or can we ac-
cept, in rare situations, a breakdown
of the resistor?
Concept
If all (and if not all, then most) of the
specifications are known, we can
start selecting a resistor. Such param-
eters as maximum voltage, way of
cooling, environmental protection
grade, electrical connection type, and
cable size are considered. This helps
determine which resistor family is
most suitable. Then, the size of the
resistor must be defined.
AI-Housing
RMI RMA
20 21
RA1
22
CMI CST
RA2
 K

Load
K : 1e–4

C1 + R1 W Power Tw
C: 0.12 R: 30 Tch
Wm Tcl
V_init: 3,200 Scope
Alpha-Model2A
Heat Sink Thermal Model

FIG 2 A simulation has three basic elements: the thermal model, the power load coming from the electrical circuit, and
the ­output.

Solution
With a power resistor, the task of
selection boils down to an assessment
of one factor: temperature. Making a
choice is very easy: If the internal or
external components stay within the
temperature parameters, the resistor
is okay. With a dynamic power load,
we need to view the resistor as a
dynamic component. Energy generat-
ed in the resistor must be expelled to
the outside. The ability to do this lies
within the properties of the resistor
itself. If we can establish a model that
shows us where the energy is generat-
ed and how it can be expelled, we can
simulate the different temperatures
and determine whether the resistor is
up to the job. From physics we can
build a thermal model. This is referred
to as modeling. Each resistor has its

December 2018 z IEEE Power Electronics Magazine 21


IEEE POWER ELECTRONICS
MAGAZINE APEC ISSUE
PELS SOCIETY a sponsor
of APEC 2019 will publish
their annual APEC Issue
This first Quarter Issue contains
a PREVIEW of APEC,
and extra copies of PELS magazine
will be given out to the attendees
For full details, email
Walter Chalupa
PELS Magazine NSM
e-mail:chalupapels@aol.com
own unique thermal model (Figure 1). The internal con-
struction and materials used define the thermal model.
Modeling is a key factor in the process of determining
the best characteristics of a resistor for a particular ap-
plication. It is not the scope of this article to explain how
the model is defined. The next step in the process is simu-
lating. The thermal model resembles an electric circuit. It
consists of many (thermal) resistors and capacitors. One
can use electrical simulation software and program the
thermal model in it as if it were an electrical circuit. The
output, although given with electrical quantities, should
be interpreted as thermal quantities. The units are differ-
ent but the numbers are the same. Measuring the tempera-
ture somewhere in the circuit will give a reading in volts.
Simply translate this to degrees. It is the same for power;
the thermal circuit is charged with a “power.” In the ther-
mal model, this is represented as a current source. The
advantage of handling thermal models and simulations as
if they were electrical circuits is that they can be used in
standard (electrical) simulation software.
Thermal Simulations
Typically, a simulation has three basic elements: the ther-
mal model, the power load coming from the electrical
circuit, and the output (see Figure 2). The power load
comes from the application. It could be a simple, well-
defined power shape that can easily be either programmed
in the software or built as an electric circuit. For example,
the charge or discharge power curve is very easy to simu-
late if one simply makes a circuit with one capacitor and
one resistor. The power dissipated in the resistor needs to
be calculated and fed into a “current source” that will be
the input for the thermal circuit.
The output is usually a scope plot. The thermal model
has two or more points where the temperature can be
measured: the internal active resistor wire and one or two
housing or casing temperatures. The user is most likely
more interested in the housing temperature than in the
resistor-wire temperature. The manufacturer of the resis-
tor needs to observe both values. If the temperatures are
within both physical limits and customer-specified limits,
the resistor is up to the job. If not, one should repeat the
simulation with either a bigger or smaller resistor to deter-
mine the best size.
Figure 3 shows an example of a high-energy dump simu-
lation. The blue line represents input power (in kilowatts).
The red line shows output active wire temperature (in °C).
The green line indicates case temperature (75 °C at 400 s).
The ambient temperature is set at 40 °C.
Conclusions
Thermal models of power resistors help us select the
correct resistor without over- or underdimensioning, sav-
ing both cost and time and leading to a high level of reli-
ability. For every company, time and cost are two key
factors that go into nearly every calculation in the search
 About the Authors
750 Wim van Dijk (dijk@danotherm
700 Load .dk) is a sales engineer at Danotherm
650 Twire Electric in Copenhagen, Denmark. He
600 Tcase High
Power (10 kw)/ Temp (°C)

550 has been actively selling power elec-

500 tronics for 20 years. He joined Dano-
450 therm Electric in 2012 and has been
400
350
involved in thermal modeling and
300 thermal simulations ever since.
250 John Siefken (John@5scomponents
200 .com) is the president and chief
150
100 executive officer of 5S Components,
50 East McKeesport, Pennsylvania. He
0 has been active in selling power elec-
0 1 2 3 4 5 6 7 8 9 10 11 12
Time (s) tronics components for more than
40 years. Prior to joining 5S Com-
FIG 3 High-energy dump simulation.
ponents, he was with General Electric
in Syracuse, New York, and Chicago,
Illinois as well as Powerex in Young-
for efficiencies. Thermal simulations, easy-to-use electrical simulation soft- wood, Pennsylvania.
useful in reducing both time and ware. They can be done fast and save
cost, can be performed in standard money in the engineering phase. 

UPGRADE YOUR PERFORMANCE

WITH SMALLER SIZE DIODES
EXTREME HIGH VOLTAGE
PERFORMANCE
Standard Performance vs. XOE
95 27
85 24 Contact DTI today for a quote
75 21
or to request FREE samples!!
IFAVM (mA)

65 18
VF (V)

55 15
45 12
35 9
25 6
15 3
G15FS XGF15 G15FS XGF15

*Based on same size 15kV diodes, VF @ 10mA www.deantechnology.com

December 2018 z IEEE Power Electronics Magazine 23


Multilevel
Converters
by Rainer Marquardt
M
odular multilevel converters (MMCs) have
paved the way for power electronics in
very demanding application areas. While
becoming the worldwide standard for
high-voltage dc (HVDC), many other future
applications are emerging. More and more, the function-
ality and cost of power electronics must be judged from a
superordinate system view. Passive components “around”
the converter, such as mechanical switchgear, fuses,
overvoltage surge suppressors, and filters offer very low
potential for future progress. With respect to protection,
the scope must be widened from the semiconductors to
Digital Object Identifier 10.1109/MPEL.2018.2873496
Date of publication: 19 December 2018
24 IEEE Power Electronics Magazine z December 2018
Modular
State of the art
and future progress
©iStockphoto/LaymanZoom
the protection of the associated equipment (e.g., motors,
drive trains, power net, and so on). Typical future require-
ments at the system level include:
■■the extremely high availability of the whole power elec-
tronic system from providing “inherent redundancy”
■■the improved scalability of the power converter as a
result of using identical submodules (SMs)
■■eliminating large passive filters (e.g., ac harmonic fil-
ters and dc capacitors)
■■eliminating bulky transformers
■■exchanging real power between several systems/drives
via a common network (i.e., a multiterminal dc).
Several multilevel converter topologies have been in­­
troduced to reduce harmonics and the associated ac fil-
ters [1], [2]. Reducing harmonics, however, is only one of
2329-9207/18©2018IEEE
several important future requirements. for the foreseeable future, all of these
Because of the increased complexity severe drawbacks, in total, restrict
of mechanical construction, mainly With respect to system matrix converters and CSCs to niche
three-level-converters have gained a requirements, the applications. An additional point is
considerable share of the applications. the nonexistence of a useful dc bus
These converters offer a good per-
availability of a useful required in the majority of future ap­­
formance-to-cost ratio, especially for dc bus is crucial for plications (multiterminal dc).
“stand-alone standard drives” up to The widespread availability of
the medium voltage level [3], [4]. For
many future fast and reliable gate turn-off semi-
converters with a higher-level count applications. conductors [e.g., i nt eg r a t ed gate-
(i.e., >3), critical failure propagation commutated thyristor (IGCT), IGBT,
after semiconductor defects is a sig- and FET] have made VSCs the most
nificant drawback that decreases suitable and universal topology. In
reliability. Especially for networks of several converters, the foreseeable future, this is unlikely to change (up to
short circuits at the dc bus remain very critical: the resulting the highest power range). Reverse conducting IGBT and
surge and resonant currents can damage even noninvolved FET structures made from wide-bandgap materials will
converters connected at the same dc bus [multiterminal- promote this trend even further. With respect to system
medium-voltage dc (MVDC) and -HVDC]. requirements, the availability of a useful dc bus is crucial
for many future applications.
Limitations of Basic Power-Converter Topologies In summary, with respect to demanding future applica-
Well-known basic topologies are: tions in the high power range, essential improvement of
■■matrix converters VSC and multilevel VSC is most useful. The main issues for
■■current-source converters (CSCs) improvement are:
■■voltage-source converters (VSCs). ■■the requirement of high-pulse frequencies—and, in con-
Historically, matrix converters and CSCs have originated sequence: high dv/dt of the semiconductors
from line-commutated (or load-commutated) converters ■■the danger of short circuits at the dc side causing extreme-
equipped with thyristors. To overcome the limitations ly high surge currents (destroyed semiconductors,
of line commutation, advanced versions of power convert- mechanical destruction of bus bars, and arcing)
ers using gate turn-off semiconductors were introduced ■■the unfavorable, slow dynamic behavior of dc-voltage
early [5]. Accordingly, this necessitates assuring com- controllability
mutation loops with very low stray inductance, which is ■■the complex construction and lack of “scalability/modu-
accomplished by adding ac-side capacitors. At first glance, larity” when increasing the voltage or power range.
this may seem advantageous because these capacitors The modular multilevel converter has been introduced
help to achieve a filter circuit for harmonics; however, the to eliminate these problems, which are relevant in many
new problems that occur from using ac-side capacitors are demanding high-power applications.
numerous and potentially severe, particularly at the line
side. These problems are: Basics and State of the Art of MMC
■■In combination with unknown and variable reactances of A main characteristic of MMC is its uniform implementation
the grid and filters of other converters, they lead to unwant- from basic building blocks (SMs). In the case of MMCs,
ed parallel or series resonances. these are two-terminal switching cells containing an inter-
■■Noncontrollable reactive currents are flowing, even when nal dc-storage capacitor (Figure 1) [6], [7].
the converter is blocked. The essentially reduced number of external con-
■■In contrast to the dc-side-capacitor (of VSC), ac-side ca­­ nections and the absence of a dc-side power
pacitors are not helpful to protect the semiconductors supply for the SMs greatly simplify the
against transient overvoltages (of high energy) from the scalability and mechanical construc-
grid side. tion of the converters. Some of the fea-
■■A high safety factor of the semiconductor voltage must tures of MMCs may at first seem to
be installed, therefore, leading to increased semiconduc- be strange because nothing similar is
tor losses. known from conventional or multi-
■■The requirement for reverse blocking semiconductors (or level VSCs. The essential differences,
additional diodes or bidirectional semiconductors) leads to be noticed, are:
to increased losses, in general. 1) The internal arm currents (i a) of
This last point becomes obvious when compared to opti- the converter are not chopped and
mized asymmetrical (field stop) or reverse conducting semi- flow continuously. They can be con-
conductors. The problem becomes even more severe when trolled to given set values by the con-
the level of blocking voltage is increased. At present, and verter control system.

December 2018 z IEEE Power Electronics Magazine 25

2) Stray inductances and additional a basic drawback of MMCs, originat-
chokes (L a) distr ibuted in the ing from the concept of “distributed
converter arms are not a problem
With respect to hard- capacitors.” Future progress concern-
(some level of inductance is neces- ware measures, both ing this issue will require increased
sary to limit high-frequency circu- investment in power semiconduc-
lating currents).
the “full-bridge func- tors and improved control schemes
3) The dc side of the converter does tionality” and “double [6], [8], [9]. With respect to hardware
not contain a dc-link capacitor; the measures, both the “full-bridge func-
connection” of SMs
elimination of capacitors and fil- tionality” and “double connection”
ters from the dc bus is a fundamen- are the main options of SMs are the main options enabling
tal advancement. enabling the essential the essential re­­duction of capacitor
4) The dc-bus voltage is controlled size [14], [15], [17], [18]. With regard
and impressed by the converter reduction of to motor dri­­ves requiring high-torque
control system quickly and directly ­capacitor size. near-zero output frequency, special
(similar to the multilevel voltage at control schemes have been success-
the ac side). fully developed [10], [11].
The last two mentioned points
enable a much better and faster control of the dc-side val- Control of the dc Bus
ues and the (directly related) instantaneous real power The MMC offers new degrees of freedom to control the dc-
flow. Additionally, severe fault conditions, especially short side voltage/current and the corresponding real power
circuits at the dc bus, can be managed in a safe manner. flows much better and faster than known from conven-
These new degrees of freedom are very valuable, particu- tional VSCs. The equivalent circuit of the dc side is quite
larly for future networks of converters and multiterminal different from that of the VSC; it consists of a controlled
dc applications. (multilevel) voltage source in series with a small induc-
During the early years of MMCs, their increased control tance. The inductance is given by the sum of stray induc-
complexity and semiconductor cost were regarded as criti- tances, part of the arm inductors (2L a /3), plus an optional
cal by many experts. Therefore, the simplest SM half bridge dc choke (if present).
(HB) was preferred. In the meantime, mainly the reduction With regard to the controllable voltage source, the high
of capacitor size has gained importance. This point presents number of voltage levels has an extremely positive effect on
the resulting dc-ripple current; therefore,
dc voltage and current can be controlled
much faster and better when compared to
P conventional VSCs or CSCs. This feature
ia1 ia3 ia5 is very important when the real power
SM SM SM changes rapidly (e.g., the sudden slip of a
drive) or when several converters share a
SM SM SM common dc bus (e.g., multiterminal dc).
Ud /2

SM SM SM Short Circuits at the dc Bus

Short circuits at the dc side impose signif-
SM SM SM icant problems with conventional VSCs.
La La La Extremely high surge currents from the
UCM L1 discharge of the dc-link capacitors are
Ud L2
L3 flowing superimposed to the rectified
La La La ac short-circuit currents. The resulting
SM SM SM
extreme mechanical forces are able to
destroy solid bus bars, cause subsequent
SM SM SM arcing, and other severe damages like an
Ud /2 explosion of IGBT modules. Additionally,
high-peak-resonant currents in the dc-bus
SM SM SM
network can destroy the semiconductors
SM SM SM of “noninvolved” converters.
ia2 ia4 ia6 MMCs have the advantage of not need-
ing capacitors or other passive filters at
N
the dc bus. Therefore, only the rectified ac-
FIG 1 The topology of a three-phase MMC. current flows following a dc short circuit.

26 IEEE Power Electronics Magazine z December 2018

This component is relatively harmless and is limited by the ac diodes (D 6, D 7) are now “converted” to IGBTs, too, by
reactances and the arm inductors when using HB SMs. The paralleling IGBT (T6, T7, marked red). This enables to
remaining disadvantages are that the dc short circuit leads to generate negative terminal voltages in both directions
an ac short circuit. A second disadvantage is the relatively high of the arm current. The freedom of using negative ter-
thermal load for the diodes in the SMs caused by the ac short- minal voltages, however, is limited by the fact that the
circuit current. The situation has, however, become manage- paralleling of differently charged capacitors causes
able and much improved as compared to conventional VSCs. current spikes. Limiting associated losses and current-
A much smarter solution is electronic dc-current limit- spike amplitudes require exact voltage measurements
ing or cut-off. This becomes possible for the MMC, when and careful switching. Typically, voltage differences
one of the following options is implemented: of the capacitors (prior to paralleling) must be kept
■■SMs, enabling bipolar terminal voltages (FB functionality) below 1% of the nominal dc-capacitor voltage. Alterna-
■■electronic or hybrid dc breakers at the dc side of the tive switching patterns may mitigate this drawback in
converter. the future.
The second of these options is apparent and may be
reasonable in some applications [12], [13]. The first option Double-Zero SMs
has some additional advantages that become evident upon By introducing a controllable dc capacitor, realized by a
closer examination. To limit or cut off each arm current reverse-conducting silicon carbide (SiC)–FET (T0 ; see Fig-
under all external conditions, the applied SMs must be able ure 3), a novel control scheme becomes possible, thus
to generate bipolar terminal voltages [6]. The well-known enabling reduced on-state losses as well as reduced switch-
topology of an FB is able to fulfill these requirements, but ing losses [16]–[18]. The applied control scheme shifts the
unfortunately, it leads to increased power loss. main switching operations to the SiC device (T0, D 0) . Addi-
tionally, the conduction loss of the SM is reduced by paral-
Advanced SMs for MMCs leling the upper (T1, T3) and lower semiconductors (T2, T4)
The absence of a dc-side supply for the SMs and the spe- during the zero-voltage (bypass) state [16], [17]. This to­­
cial operating conditions in an MMC are the two main pology is very suitable for implementing an essentially
points that enable essentially improved topologies. A con-
ventional FB is applicable for MMC SMs and offers the fol-
lowing advantages:
T6
■■electronic dc-current limitation and a wide dc-voltage
P1 P2
operating range
D6
■■higher modulation factor of ac voltage (k > 1), which is T1 D1 T3 D3
useful to minimize the size of capacitors.
C1 + T D5 + C2
Unfortunately, a conventional FB leads to increased con- X1 – 5 –
duction losses, which are not generally acceptable in most T2 D2 T4 D4
high-power applications. D7
N1 N2
Clamp-Double SMs T7
The clamp-double SM (Figure 2) has been introduced in [6] X2
and [7] to enable electronic over current protection under dc
FIG 2 A clamp-double SM [6] or semi-FB SM [14]
short circuit conditions and to reduce power loss compared (T6 and T7 added).
to FB SMs. It has the following advantages:
■■the added expenditure for the semiconductors is relative-
ly low
■■all of the semiconductors have the same blocking-voltage P
requirement
■■one on-state voltage is eliminated since only 50% of the T1 D1 T3 D3
D0 T0
maximum positive arm voltage is required in a negative
direction in typical MMC applications. X1
The resulting conduction loss can therefore be reduced C0 +
T2 – T4
by 25%, as compared to that of FBs. The basic idea of this D2 D4
double connection of suitable SMs is of general importance.
It is further applied in the following versions. N

Semi-FB SMs X2
The semi-FB SM (Figure 2) is an extension of the clamp-
double SM and has been proposed in [15]. The clamping FIG 3 A double-zero submodule (DZ-SM).

December 2018 z IEEE Power Electronics Magazine 27

improved double SM. The main ad­­ Protecting power semiconductors
vantages of this are that the power from explosion is essential for high-
losses are reduced further. Very va­­ To limit or cut off power converters. MMCs offer better
luable options for fully electronic each arm current conditions for solving these problems
protection against an explosion are because the SMs do not have a sepa-
enabled, too. under all external rate dc supply (which would deliver
The favorable double-connection ­conditions, the applied high-dc power into the SM after a
circuit can be achieved by externally fault). Nevertheless, the energy of
connecting two, single double-zero SMs must be able to the SM capacitor is high, and stan-
(DZ) SMs, as shown in Figure 4. The generate bipolar dard module packages are unable to
stray inductance (L S) of these connec- absorb the discharge energy. Com-
tions does not disturb the switching
­terminal voltages. paratively, press pack housings would
operations. This configuration also be more suitable, but the developing
enables the capacitor size to be cut in trend of improved semiconductors
half, almost exactly. will drive them beyond their limita-
tions as well. Multichip press pack arrangements, which
Future Prospects are necessary for modern semiconductors, are getting
Large networks dominated and controlled by converters unable to ensure the safe conduction (of the nominal cur-
will become a typical structure for many important future rent) after defects. These problems will be aggravated with
applications. The demanding requirements in these net- shrinking chip sizes (SiC), and increased number of paral-
works are lel chips. In summary, the future of press pack housings
■■fully electronic failure management and fast protection of remains questionable.
associated equipment (without the aid of mechanical Electronic protection is a promising future approach,
switches or fuses) however, it requires true internal redundancy of the SMs,
■■electronic current lim­itation at the ac and dc side (includ- both for the power circuit and the control electronics. For
ing overvoltage clamping). HB-type SMs, a possible solution has been presented in
The inherent redundancy of MMCs and the elimination [24]. This double SM of HBs restricts the redundant power
of large capacitors from the dc side facilitate fully elec- switches to one device per double SM. As expected, the
tronic failure management. These features are very impor- power loss increases (caused by the redundant power
tant at the system level in extended networks. switch), which carries a high portion of the arm current.
When assessing the converters, the three main consid- A suitable topology with FB functionality in which the
erations are: “redundant” switch leads to reduced (!) power losses of the
■■reduced size of the capacitors SM, is the DZ SM shown in Figure 3.
■■protection against explosion of submodules Concerning power loss and efficiency, there exists a
■■improved efficiency. permanent requirement for further improvement. Espe-
The first consideration is a common drawback of MMCs cially in the high power range, energy cost and expenditure
and originate from the concept of “distributed capacitors.” A for the cooling equipment are always of concern. Advanced
good option for improvement (enabling a reduction of almost SM topologies and improved semiconductors are the
exactly -50%) is the “double connection” of suitable SMs main drivers of progress. With respect to semiconduc-
(semi-FB-SM: Figure 2; DZ SM: Figure 3). Further reduction tors, reverse-conducting silicon Si-GBT chips and reverse-
requires improved control schemes, which must establish conducting SiC–FET chips will be the main options in the
a controlled energy exchange between the arms of the con- foreseeable future. A third option are SMs, which enable a
verter [9], [25]. useful combination of both technologies, such as the DZ
SM (Figure 3).

Comparison of Efficiency
Ls MMCs, with typical values near 99%, are known for their
X1 DZ – SM X2
(1) high efficiencies. The first large commercial converters
[23] have previously reached values of roughly 98.8%. The
P
main reason for the high efficiencies is the very low
N switching frequencies that are required by the semicon-
Ls X1 ductors of the SMs. Typical switching frequencies are
DZ – SM X2
(2) approximately fsw = 150 Hz in HVDC applications. On
the other hand, even at these low frequencies, the
switching-power losses of the applied 4.5-kV IGBT
FIG 4 The double connection of DZ-SMs. devices amount to roughly one-third of the total losses. SiC

28 IEEE Power Electronics Magazine z December 2018

devices for MMCs have been consid- components in the arm currents. A
ered previously [19] and would lead to conventional LCC converter with thy-
a nearly zero switching loss under
Using SiC-FETs, conduc- ristors performing the same task as
these conditions in MMC [22]. Using tion losses could also be the MMC would have three times this
SiC-FETs, conduction losses could minimum value (see Figure 5). MMCs
reduced; however, this
also be reduced; however, this will equipped with HB SMs have slightly
remain a commercial issue related to will remain a commer- higher values because the ac compo-
the size of the chip area as well as cial issue related to the nent of the arm current is relatively
cost in the foreseeable future. high, as shown in the shaded opera-
Line-commutated thyristor conver­ size of the chip area as tion-range HB of Figure 5. SMs with FB
ters (LCCs) represent a demanding well as cost in the fore- functionality provide for the dimen-
reference for comparison. In HVDC sioning of a higher ac voltage and, as
applications, the LCCs achieve effi- seeable future. a result, a lower nominal ac current.
ciencies of up to 99.4%; however, to Typical and reasonable values are in the
be fair, approximately 0.1% of power range of k . 1.5, with the possibility of
losses of the large, passive filters should be considered even higher values. In summary, the normalized arm cur-
as well. Moreover, for future development of MMCs, LCC rent values can therefore be as low as two times the theoreti-
efficiency will set the re­­ference value. To gain general cal minimum value, or even less (as illustrated by the round
insight into the development potential of MMCs and the dots in Figure 5).
options for improvement, the following analysis has A second parameter to consider is the squared RMS
been conducted according to [18]: it is based on the value of the current of SM dc capacitor (Figure 6), which
uniform use of SIC–FET devices with identical data is normalized in the same manner as described for the
per transistor. A per device, on-resistance of R SiC = 2 mX arm current. This figure is a measure of power loss of the
has been stated. Considering the state of the art of high- SiC–FET (T0, D 0) of the new SM (Figure 3). In fairly good
voltage SiC chips [21], this value leads to approximately approximation, however, it is an indicator of the necessary
50 cm 2 of the total chip area per transistor function. The size of the dc capacitor as well. These generalized results
absolute value, however, is not relevant if the following clearly indicate that SMs with FB functionality permit
results are regarded as a comparison of the different topol- essentially improved conditions for the dimensioning of
ogies. Under these conditions, computing the following the components.
parameters of the converters provides general insight. A comparison of the resulting power losses is given in
Figure 5 shows the total squared root mean square Figure 7. As expected, half-bridge (HB) submodules offer
(RMS) value of the arm current of an
MMC as a function of ac-modulation fac-
tor (k). For the MMC, k is defined as the
ratio between ac-to-dc-voltage per con- 2
Ia,RMS 1
verter arm. A value of k = 1 corresponds =f
Id 2 k
to the maximum available ac voltage of
3
an MMC solely equipped with HB SMs.
Because of some headroom for voltage 5.5
tolerances, approximately k = 0.8 must φ = 30°
5 Typical (FB)
be chosen as a nominal value.
4.5 φ = 15°
To obtain generalized knowledge— Typical (HB)
4 φ = 0°
independent of the power and voltage
ranges—the squared RMS value of the 3.5
2

arm current must be normalized to a 3

Ia,RMS

LCC (for Comparison)

Id
3

“universal” reference value. The theoreti- 2.5

2

cal minimum value of arm current that 2

is necessary for delivering the required
1.5
real power has been selected for this pur-
HB
FB

pose. This value corresponds to one- 1

third of the dc current (for a three-phase 0.5
1
ac/dc converter). The minimum value of 0
0 0.25 0.5 0.75 1 1.25 k
squared RMS arm current is therefore
equal to ^ I d /3 h2. Any type of real con-
verter will have a higher value of squared FIG 5 The normalized squared root mean square (RMS) value of an arm current as
arm current because of additional ac a function of modulation factor k.

December 2018 z IEEE Power Electronics Magazine 29

 by parallel operation in the bypass-state
[16]. In addition, this topology enables
2
IC,RMS 1 the option to keep the main advantages
=f
Id 2 k by using a combination of one SiC- and
3 four Si-devices only. With respect to the
“benchmark of efficiency” LCC, it can be
1.75 concluded that it will be possible in the
Typical (FB) φ = 30° future to compete with LCC concerning
1.5
efficiency. Solely, higher dc current may
Typical (HB)
remain the essential advantage of LCC in
1.25 φ = 15°
the future.
φ = 0°
2
IC,RMS

1
Id

Conclusions
3
2

0.75 The progress of high-power VSCs re­­

mains important for many future applica-
0.5 tion fields, particularly grid infrastruc-
ture, the efficient use of solar and wind
0.25 power resources, multiterminal HVDC
FB HB 1 and MVDC networks, medium-voltage
0 drives, electric ships [20], [26]. MMCs
0 0.25 0.5 0.75 1 1.25 k
have opened new roads for power elec-
tronics in these very demanding applica-
FIG 6 The normalized squared RMS value of capacitor current as a function of mod- tion areas, where neither conventional
ulation factor k. multilevel-VSC nor thyristor converters
meet the requirements. Fully electronic,
the lowest conduction losses. On the other hand, the impor- fast failure management and the protection of associated
tant requirements for fully electronic fault management and equipment in extended networks is now feasible. Electronic
current limitation are not fulfilled. Hybrid denotes a mix- current and voltage limitation at the ac- and dc-side con-
ture of HB- and FB- submodules per converter arm. Similar trolled by converters will be extremely valuable in future
to the HB-version, it suffers from a restricted, narrow dc- applications, especially for multiterminal dc.
voltage operating range and large capacitors.
All the advanced submodules do not suffer from these About the Author
drawbacks. The new double-zero submodule offers the low- Rainer Marquardt (rainer.marquardt@unibw.de)
est power losses. The main reason is that conduction losses received the Dipl.-Ing. (M.Sc.) and Dr.-Ing. degrees (Ph.D.) in
of the “main semiconductors” (T1 fT4) have been reduced 1978 and 1982, respectively, both in electronic communication

Conventional Submodules Advanced Submodules

Conventional Submodules
Double SM
nHB 2 k = 1.5
Pon,HB = r . RSiC . Ia,RMS 1 Id = 3 kA
nFB Typical LCC (for Comparison)
2
Pon,FB = r . 2RSiC . Ia,RMS
0.8 k = 1.25 k = 1.5
n n 2
. RSiC . Ia,RMS Id = 2.5 kA k = 1.5 Id = 3 kA
Pon,Hybrid = 2 FB + HB
(1 – η) (%)

r r Id = 3 kA
0.6
Advanced Submodules k = 0.8 k = 1.5
Id = 2 kA
nFB Id = 3 kA
Pon,DZ = r . RSiC . (Ia,RMS + 2IC,RMS)
2 2
0.4
3 .
Pon,Semi – FB = Pon,FB
4
3 . 0.2
Pon,DC–DZ = Pon,DZ
4

r = 0.9; φ = 0°; Vc = 2.25 kV; RSiC = 2 mΩ 0

HB FB Hybrid DZ Semi-FB DC–DZ

FIG 7 General calculation and comparison of power loss of different SM topologies.

30 IEEE Power Electronics Magazine z December 2018

and power electronics, from the University of Hannover,
where he worked until October 1983 as a research scientist
at the Institute for Power Electronics (Prof. Klemens Heu-
mann). He then joined Siemens AG/Erlangen working as an
R&D engineer, leader of power electronics development of
traction and innovation for power electronics. He per-
formed numerous industrial research and development
projects for high-power applications in power transmission
and advanced ac-drive systems for traction applications. He
has filed more than 50 patents in these areas. Currently, he
leads the institute of “Power Electronics and Control” at
the University of Bundeswehr/München as a full professor
and is the director of the associated Research Institute
ITIS. His inventions include low-loss snubber circuits for
gate turnoff power semiconductors—first applied in high-
speed trains (ICE, Germany) and modular multilevel
converters for high-performance/high-power VSC—first
applied in HVDC-transmission (San Francisco Bay). He
received the highest award “Ring of Honor” from the Insti-
tution of German Electrical Engineers (VDE 2012), the Out-
standing Achievement Award from the European Power
Electronics Association (EPE 2015), the Uno Lamm HVDC
Award from the IEEE (PES 2015), and the William E. New-
ell Award from the IEEE (PES 2018).
References
[1] A. Nabae, I. Takahashi, and H. Akagi, “A new neutral-point-clamped PWM
inverter,” IEEE Trans. Ind. Appl., vol. IA-17, no. 5, pp. 518–523, 1981.
[2] T. Meynard and G. Gateau, “PWM strategies for multilevel converters,” in
Power Electronic Converters: PWM Strategies and Current Control Tech-
niques, E. Monmasson, Ed. London: Wiley-ISTE, 2013, pp. 243–286.
[3] T. Brückner, S. Bernet, and H. Guldner, “The active NPC converter and
its loss-balancing control,” IEEE Trans. Ind. Electron., vol. 52, no. 3, pp.
855–868, 2005.
[4] J. Rodriguez, S. Bernet, P. K. Steimer, and I. E. Lizama, “A survey on
neutral-point-clamped inverters,” IEEE Trans. Ind. Electron., vol. 57, no. 7,
pp. 2219–2230, 2010.
[5] S. Bernet, T. Matsuo, and T. A. Lipo, “A matrix converter using reverse
blocking NPT-IGBTs and optimized pulse patterns,” in Proc. 27th Annu.
IEEE Power Electronics Specialists Conf., 1996, pp. 107–113.
[6] R. Marquardt, “Modular multilevel converter: An universal concept for
HVDC-networks and extended DC-bus-applications,” in Proc. IEEE Int.
Power Electronics Conf. (IPEC), Sapporo, Japan, 2010, pp. 502–507.
[7] R. Marquardt, “Modular multilevel converter topologies with dc-short
circuit current limitation,” in Proc. Int. Conf. Power Electronics and ECCE
Asia (ICPE & ECCE), 2011, pp. 1425–1431.
[8] D. Dinkel, C. Hillermeier, and R. Marquardt, “Direct multivariable control
of modular multilevel,” in Proc. European Conf. Power Electronics and
Applications (ECCE Europe), Riga, Latvia, 2018, pp. 1–10.
[9] D. Karwatzki, L. Baruschka, M. Dokus, J. Kucka, and A. Mertens, “Branch
energy balancing with a generalised control concept for modular multilevel
topologies—Using the example of the modular multilevel converter,” in Proc.
European Conf. Power Electronics and Applications (EPE), 2016, pp. 1–10.
[10] A. Antonopoulos, K. Ilves, L. Angquist, and H.-P. Nee, “On interaction
between internal converter dynamics and current control of high-performance
high-power ac motor drives with modular multilevel converters,” in
Proc. IEEE Energy Conversion Congr. Exposition (ECCE), 2010, pp.
4293–4298.
[11] J. Kolb, “Optimale betriebsführung des modular multilevel-umrichters
als antriebsumrichter für drehstrommaschinen,” Ph.D. thesis, Karlsruhe
Institute of Technology, Germany, 2013.
[12] D. Schmitt, Y. Wang, T. Weyh, and R. Marquardt, “DC-side fault cur-
rent management in extended multiterminal-HVDC-grids,” in Proc. 9th Int.
Multi-Conf. Systems, Signals and Devices, 2012, pp. 1–5.
[13] D. Schmitt, “Modular multilevel converter M2C für multiterminal
HVDC,” Ph.D. thesis, Universität der Bundeswehr München, Germany, 2012.
[14] K. Ilves, L. Bessegato, L. Harnefors, S. Norrga, and H.-P. Nee, “Semi-full-
bridge submodule for modular multilevel converters,” in Proc. 9th Int. Conf.
Power Electronics and ECCE Asia, 2015, pp. 1067–1074.
[15] K. Ilves, “Modeling and design of modular multilevel converters for grid
applications,” Ph.D. thesis, KTH Royal Institute of Technology, Stockholm,
Sweden, 2014.
[16] R. Marquardt, “Modular multilevel converter: Impact on future applica-
tions and semiconductors,” in Proc. Power Electronic Components and
Their Applications, ETG-Symp. VDE, Bad Nauheim, Germany, 2017,
pp. 1–10.
[17] C. Dahmen and R. Marquardt, “Progress of high power multilevel con-
verters: Combining silicon and silicon carbide,” in Proc. Int. Exhibition
and Conf. Power Electronics, Intelligent Motion, Renewable Energy and
Energy Management (PCIM Europe), 2017, pp. 1679–1685.
[18] C. Dahmen, F. Kapaun, and R.Marquardt, “Analytical investigation of
efficiency and operating range of different modular multilevel converters,”
in Proc. Int. Conf. Power Electronics and Drive Systems (PEDS), 2017,
pp. 336–342.
[19] D. Peftitsis, G. Tolstoy, A. Antonopoulos, J. Rabkowski, J.-K. Lim, M.
Bakowski, L. Ängquist, and H.-P. Nee, “High-power modular multilevel con-
verters with SiC JFETs,” in Proc. IEEE Energy Conversion Congr. Exposi-
tion, 2010, pp. 2148–2155.
[20] M. Winkelnkemper, L. Schwager, P. Blaszczyk, M. Steurer, and D. Soto,
“Short circuit output protection of MMC in voltage source control mode,” in
Proc. IEEE Energy Conversion Congr. Exposition (ECCE), 2016.
[21] K. Kawahara, S. Hino, K. Sadamatsu, Y. Nakao, T. Iwamatsu, S. Nakata,
S. Tomaohisa, and S. Yamakawa, “Impact of embedding schottky barrier
diodes into 3.3 kV and 6.5 kV SiC MOSFETs,” Silicon Carbide Related Mate-
rials, vol. 924, pp. 663–666, 2017.
[22] Y. Ishii and T. Jimichi, “Verification of SiC based modular multilevel
cascade converter (MMCC) for HVDC transmission systems,” in Proc. Int.
Power Electronics Conf. (IPEC, ECCE Asia), 2018, pp. 1834–1839.
[23] J. Dorn, H. Huang, and D. Retzmann, “A new multilevel voltage-sourced
converter topology for HVDC applications,” in Proc. Cigré Session 2008, Paris.
[24] H.-G. Eckel and H. Gambach, “Double module for a multistage convert-
er,” U.S. Patent, US9.263.969B2, Feb. 2016.
[25] Y. Wang and R. Marquardt, “Novel control scheme for the internal ener-
gies and circulating currents of modular multilevel converter,” in Proc. Int.
Exhibition and Conf. Power Electronics, Intelligent Motion, Renewable
Energy and Energy Management (PCIM Europe), 2017, pp. 606–611.
[26] M. Spichartz, V. Staudt, and A. Steimel, “Modular multilevel converter
for propulsion system of electric ships,” in Proc. IEEE Electric Ship Tech-
nologies Symp. (ESTS), 2013, pp. 237–242.

December 2018 z IEEE Power Electronics Magazine 31
Multilevel Inverters
for Grid-Connected
Photovoltaic
Applications
Examining emerging trends
by Xing Zhang, Tao Zhao, Wang Mao, Don Tan, and Liuchen Chang
A
s the cost of photovoltaic (PV) modules and invert-
ers continues to decline, PV power generation is
gaining more and more share in the electricity
market. The market and its customers are
demanding higher-performance inverters in
terms of efficiency, power density, module-level control,
and increasingly higher voltage and power levels. Because
of their outstanding performance, multilevel inverters have
attracted the attention of researchers and technology devel-
opers alike. This article presents commonly used multilevel
inverter technologies for grid-connected PV applications,
Digital Object Identifier 10.1109/MPEL.2018.2874509
Date of publication: 19 December 2018
32 IEEE Power Electronics Magazine z December 2018
©istockphoto.com/MeggiSt
including five-level inverters, single-phase nonisolated
inverters, and three-phase, isolated cascaded H-bridge
inverters. Detailed discussions are presented, along with
characteristics of PV applications.
Global Outlook of Solar Energy
In the last ten years, solar energy has become the fastest-
growing renewable energy source in terms of installed
capacity. At least 98 GW of PV capacity was installed in
on- and off-grid markets in 2017, increasing the total
installed capacity by nearly one-third, for a cumulative
total of approximately 402 GW, as shown in Figure 1 [1].
The International Renewable Energy Agency reports that
prices for solar and wind auctions have decreased in
2329-9207/18©2018IEEE
recent years. As shown in Figure 2, solar energy was con-
tracted at an average global price of US$50/MWh in 2016,
compared with the average price of nearly US$250/MWh
in 2010 [2]. In 2016, a bid of US$24.20/MWh was submit-
ted by China’s Jinko Solar and Japan’s Marubeni for
1.17 GW of capacity in Abu Dhabi [2], which was the lowest
solar price reported during that time. Meanwhile, the mar-
ket for different types of inverters has revealed new trends.
A shipment of 80-kW string inverters increased dramati-
cally, suggesting that the power of string inverters will
become higher in the next five years. In 2016, global ship-
ments of microinverters and power optimizers reached
3 GW (up by 28% from 2015) and are expected to reach
9 GW by 2021. Centralized inverters above 2 MW will be­­
come the norm for the market, and centralized inverters
below 1 MW will be shipped less and less.
With the continuous reduction of global PV system pric­
­es, inverter manufacturers are competing more intensely.
The general trend indicates that inverters are becoming
smaller in size, lighter in weight, and lower
in price. One way the size of inverters has
been reduced is through the use of silicon
Solar PV Global Capacity and Annual Additions, 2007–2017
carbide (SiC) power-switching devices that
increase the switching frequency, thereby 450
reducing the size of the inductors. How- 400
ever, currently, the technology of SiC pow­­ 350
er devices is at an emerging stage, their 300
prices remain more expensive than silicon
(GW)
(MPPT), module-level monitoring, and module shutdown.
In general, CHB inverters can be divided into two cate-
gories: single-phase and three-phase. If a high-frequency
transformer is present in the front stage of an inverter,
three-phase CHB inverters can be divided into three-
phase nonisolated CHB inverters and three-phase isolated
CHB inverters. Single-phase CHB inverters, which are
mainly used in low-voltage, low-power applications, are
commonly used in residential PV generation units. Three-
phase nonisolated CHB inverters can be used in low- and
medium-power applications, while three-phase isolated
CHB inverters are commonly employed in high-voltage
and high-power generation systems.
Five-Level Grid-Connected PV Inverters
To further increase power capacity, the five-level topology is
applied to more and more PV systems in the field. In 2010,
the simplest five-level active neutral point clamped (ANPC)
topology was unveiled by ABB. As shown in Figure 3(a)
World Total
Annual Additions 402 GW
Previous Year’s Capacity
+98 Total
303 Global
+76
Capacity
228

250 +51
177
devices, and their applications are limited. 200 +40
137
Another way of reducing the size is to use 150 100 +38
multilevel inverter technologies, which can 70 +29
100 40 +31
decrease the voltage rating and losses of 15 23 +17
50 8
the devices, consequently reducing the size +2.5 +6.6 +8
0
of the cooling systems and increasing the 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
switching frequency, thus reducing the size
of inductors. FIG 1 Global PV capacity and annual additions from 2007 to 2017 [1].
Compared with the traditional three-
level inverters, five-level technology has
attracted much attention from inverter
manufacturers and scholars because 300
of its exceptional performance. In 2017,
Solar Prices Wind Prices
the National Electrical Code (NEC) set 250
stricter rules for rapid shutdown. Accord-
ing to the NEC’s newest regulations, a 200
(US$/MWh)

PV system requires the rapid shutdown

of its operation at the module level to 150
improve system safety, which will aggres-
sively encourage the development of 100
module-level power electronics technolo-
gies. Additionally, the trend of centralized 50
inverters is toward higher power, higher
voltage, and higher integration. Conse- 0
2010 2011 2012 2013 2014 2015 2016
quently, research interest in the area of
cascaded H-bridge (CHB) [3] inverters Year
is growing due to their modularity, mod-
ule-level maximum power point tracking FIG 2 The average auction prices from 2010 to 2016 [2].

December 2018 z IEEE Power Electronics Magazine 33

 dc+ dc+
+ +
V1 V3 V5
Vdc A
V1 V5 F
Vdc
D1
D1 D3 – –
D Vdc C
NP E PH
– + +
V2 V6
NP D2
+ D2 D4 G
Vdc B
V2 V4 V6
V3 V7
Vdc
–
dc–
First Mode Fifth Mode
Second Mode Sixth Mode
V4 V8
– Third Mode Seventh Mode
dc– Fourth Mode Eighth Mode
(a) (b)

FIG 3 (a) A conventional, neutral point (NP) clamped topology and (b) a five-level ANPC topology from ABB. PH: phase.

S1A The Main S1C

Circuit
S5A Topology
S2A S2C
S6A
+ L3 S3A S2 S3C
Vdc/2
– S6B
3Vdc + + S4A S4C L
+ C3 C4
Vdc o 2 C C2 1
– – – a S5B
Vdc Cdc S4B
Vdc S4D
+ 4
Vdc/2 2
– S3B S1 S3D

S2B Interphase S2D

Active Capacitor Balancing
Block
S1B S1D
L2
n
(a) (b)

FIG 4 (a) A conventional, five-level flying-capacitor topology and (b) a modified, five-level flying-capacitor topology of SolarEdge.

and (b), when compared to the conventional topology, the
number of power semiconductors is decreased, and the
voltage rating of each of the power semiconductors can be
kept lower by using the ANPC topology. Therefore, this
topology not only reduces cost, it also improves the sys-
tem’s reliability. This topology has also been used by ABB
for the medium-voltage drives series ACS2000 in the
4–6.9-kV voltage range.
In grid-connected PV systems, SolarEdge Technolo-
gies introduced its new high-definition wave inverters,
which have 16 times fewer magnetics and 2.5 times fewer
cooling components than conventional inverters. As shown
in Figure 4(a) and (b), the main disadvantage of the con-
ventional, five-level flying-capacitor topology is that it
requires a large number of capacitors, which increases
costs and reduces system reliability. To address this prob-
lem, SolarEdge presented the active capacitor circuit and
interphase balancing block, which considerably reduce
both the size and cost of input bus capacitors and flying
capacitors. Furthermore, SolarEdge uses thin film instead
of electrolytic capacitors, resulting in further-improved
system reliability. By using low-voltage MOSFETs at high

34 IEEE Power Electronics Magazine z December 2018

 + Q1
First Unidirectional
Element Q7
Second Unidirectional
Vdc/2 Element
+ Q2
C1 Q3
Vdc/4
Third Unidirectional
N Element
PV + R
Fourth Unidirectional
C3
Element
+
C2 Q5 Q4
Vdc/2
Fifth Unidirectional
Element
Sixth Unidirectional Q8
Element
–
Q6

First Mode VRN = Vdc/2 Fifth Mode VRN = 0

Second Mode VRN = Vdc/4 Sixth Mode VRN = –Vdc/4
Third Mode VRN = Vdc/4 Seventh Mode VRN = –Vdc/4
Fourth Mode VRN = 0 Eighth Mode VRN = –Vdc/2

(a)

dc–dc C1 Five-Level Five-Level Five-Level

PV Converter Inverter Inverter Inverter
C2
–

Power Grid

(b)

FIG 5 (a) The five-level topology of Sungrow and (b) the schematic diagram of a three-phase, four-wire, five-level inverter.

frequency, the power losses are significantly lower, the
size and cost of passive components (particularly mag-
netic components) are reduced, and the inverter’s volume
is greatly decreased.
To solve the problem of power capacitor self-balance
in the conventional neutral point clamped topology, Sun-
grow introduced another five-level topology in 2015. As
depicted in Figure 5(a) and (b), two working modes, i.e.,
the second and third modes, can be selected for output-
ting a positive voltage, with currents flowing through the
clamping capacitor C3 in the opposite directions in the
two working modes. Therefore, the voltage of the clamp-
ing capacitor C3 can be balanced by flexibly selecting the
two working modes. Similarly, when outputting a negative
voltage, the voltage of the clamping capacitor C3 can be
balanced by f lexibly selecting the other two working
modes, i.e., the sixth and seventh modes. A balance of
capacitor voltage can be achieved at full power and full
modulation without adding extra hardware circuitry.
Sungrow released the world’s first 1,500-V string inverter
SG80HV featuring the five-level topology at the Asia Solar
2016 solar tradeshow.

December 2018 z IEEE Power Electronics Magazine 35

Single-Phase, CHB Grid-Connected PV Inverters
One of the goals of PV inverters is to achieve module-level
control that has many advantages, such as module-level
MPPT, module-level monitoring, and panel shut-off. A tradi-
tional micro-PV inverter structure is shown in Figure 6(a).
Each PV module is connected to a microinverter, whose
output is directly connected to an ac grid. To match the grid
voltage, this topology contains a high-frequency isolation
transformer to boost the output voltage of a PV module,
which increases the system’s cost and reduces its efficiency.
Figure 6(b) depicts the topology of series-connected, power-
optimized PV inverters in which each PV module is con-
nected to a dc–dc converter and multiple dc–dc converters
are connected in series. The dc–dc converter in this topol-
ogy is also called a power optimizer; it controls a PV mod-
ule independently, usually using the buck-boost topology
shown in Figure 6(c). However, since there are multiple dc–
dc converters in this topology, its efficiency is relatively low
and cost is expected to rise. Figure 6(d) shows a low-volt-
age, ac-side series PV inverter topology that was presented
by Empower/Kaco at the 2016 Munich Exhibition, consist-
ing of n inverters in series. The ac side of each inverter is a
sinusoidal voltage of the same frequency and is in phase,
while all of the ac sides of the inverters are connected in
series to the voltage level of the ac grid. This topology can
also gain independent control of the PV modules; however,
many filters are used, and the switching frequency needs to
be increased to reduce the filter volume, which, to some
extent, reduces its efficiency.
To achieve module-level control while taking efficiency
and cost into consideration, the single-phase CHB topology
(Figure 7), which has a modular structure and enables mod-
ule-level monitoring and MPPT, was developed [4]. Although
each H-bridge unit operates at a lower switching frequency,
the ac-side voltage VAB can output a multilevel staircase wave
with a relatively high equivalent switching frequency, which
is necessary for reducing the switching loss of inverter and
the volume of inductors L 1 and L 2 . Switches S 11 + S n4 can
use low-voltage MOSFETs with small on-resistance (R ds(on)),
which also helps reduce cost and improve efficiency. With

PV Module ac Bus PV Module Power Optimizer

dc
PV1 Microinverter PV1
dc dc

dc
PV2 Microinverter PV2
dc
Grid
… … … …
ac
dc
PVn Microinverter PVn
dc

Grid

(a) (b)

dc/ac Filter ac Bus

PV Module

L PV1
Q1 Q3

PV2
C1 Q4 C2
Q2 … …

PVn

Grid

(c) (d)

FIG 6 (a) A microinverter structure, (b) a power-optimized PV grid-tied structure, (c) a buck boost, and (d) an Empower/Kaco structure.

36 IEEE Power Electronics Magazine z December 2018

the CHB topology, it is possible to eli­ has one more degree of freedom (DoF)
minate the transformer between the than does the H-bridge, this topology
inverter and gird, further improving
Because the current can, to some extent, suppress the leak-
the efficiency and power density of flowing through each age current. However, the cost and
the system. Therefore, CHB multilevel switching losses are increased com-
topology has significant advantages
H-bridge is the same pared to the conventional CHB to­­
in the field of PV power generation. but the power differ- pology. In [6], a modulation strategy to
However, there are three issues in this reduce the current leakage by re­­
ence is too large, the
topology worth examining. ducing the change rate of the sum of
H-bridge units with the parasitic capacitance voltage is
Leakage Current Problem higher power will be proposed. However, this modulation
Since there is no isolation trans- strategy is based on two H-bridge units
former in the entire system, a leak- overmodulated, result- and is not applicable to the system in
age current (i leakage) will be gener- ing in distorted grid which multiple H-bridge units are cas-
ated on the parasitic capacitance caded. In [7], the common-mode filters
(C PVij, i = 1, 2, f, n, j = 1, 2) between current as well as sys- are added on the dc side and ac side of
the PV module and the earth, with the tem instability. the CHB PV inverter, resulting in bet-
leakage current loop, as shown in Fig- ter performance in suppressing the
ure 8. The application of cascaded leakage current than the methods in
structure leads the leakage current [5] and [6], however, the cost and power
in parasitic capacitance with respect to not only the com- density of the system are compromised. Therefore, the
mon-mode voltage, but also the difference-mode voltage, leakage current suppression of the single-phase CHB PV
so the problem of leakage current is even more promi- inverter warrants additional study.
nent. Several technical papers have proposed a few meth-
ods to address the issue related to leakage current. A cas- Overmodulation Caused by Power Imbalance
caded H5 topology is proposed in [5], which replaces the As seen in Figure 9, the input power of some H-bridges
H-bridge units shown in Figure 7 with H5 units. Since H5 may drop drastically due to dark clouds or damaged PV

H-Bridge Unit
S11 S13 CPV11
S11 S13
A L1 L1
A
C1 C1
PV1 PV1
S12 S14 S12 S14
CPV12

CPV21 S21 S23

S21 S23

C2 PV2 vg
PV2 vg
S22 S24
S22 S24
CPV22
……
……
CPVn1 Sn1 Sn3
Sn1 Sn3

PVn Cn
Cn L2 O
B
PVn Sn2 Sn4
B L2
Sn2 Sn4 CPVn2
ileakage

FIG 7 The schematic diagram of a single-phase, CHB grid-con- FIG 8 The leakage current loop of a single-phase, CHB grid-
nected PV inverter. connected PV inverter.

December 2018 z IEEE Power Electronics Magazine 37

modules. Because the current flowing through each H-bridges in CHB inverters from overmodulating under
H-bridge is the same but the power difference is too large, heavy power imbalance conditions. Even if the MI of
the H-bridge units with higher power will be overmodu- some H-bridges reach up to 1.155, the amplitudes of their
lated, resulting in distorted grid current as well as system modulation waveforms will be no greater than unity by
instability. A hybrid modulation strategy is described in [8] the third-harmonic compensation. Although the capacity
and can extend the linear operating region of CHB invert- of this method to deal with power unbalance is weaker
ers more than those of sinusoidal pulsewidth modulations than that of the schemes in [8] and [9], it can ensure that a
because the maximum modulation index (MI) of the CHB inverter operates in unity power factor with the
square wave is 4/r. However, this method could increase small fluctuation of dc-link voltages under some imbal-
dc-link voltage fluctuations. A reactive power-compensa- anced power conditions.
tion strategy, which exploits the power factor as a DoF to
stabilize the converter operation and may limit its applica- Interunit Communication Problem
tion, is discussed in [9]. In [10], an optimized, third-har- CHB inverters require all H-bridge units to work together
monic-compensation strategy is proposed to prevent the to achieve high-performance grid-connection operation,
therefore, interunit communication is needed. Conven-
tional Controller Area Network and 485 communication
technologies require additional communication lines,
H1
L1 which increases the complexity and unreliability of the
Irradiation Level
H2 system. Accordingly, power line communication (PLC)
1,000 W/m2
has been used to solve this problem. In [11], a PLC-based,
500 W/m2 Hn PV module-monitoring application, which uses the dc
… vg power lines as the communication channel, is proposed;
300 W/m2 however, the bit rate is low, and it can only be employed
L2 during low-demand PV-monitoring situations. A new PLC
scheme related to dc–dc converters in which the commu-
nication signal is generated by switching the frequency
H1
L1 modulation of power converters has been presented [12].
Irradiation Level
H2 Here, however, the amplitude of the data signals is small
1,000 W/m2
and cannot be regulated. Another PLC system is used on
1,000 W/m2 Hn cascaded H-bridge inverters [13] while the communication
… vg signal is transmitted between two adjacent output voltage
1,000 W/m2 levels and superimposed on the combined multilevel out-
L2 put. However, the transmission line effect will severely
interfere with the signal, leading to data corruption in cas-
caded H-bridge inverters. Therefore, PLC utilized in inter-
FIG 9 The power imbalance of a single-phase, CHB grid-con- unit communication of a cascaded H-bridge inverter still
nected PV inverter. faces challenges.

ac Breaker

EMI LCL EMI 35-kV

PV Transformer
Filter Filter Filter ac Grid
Arrays

ac Main Contactor

dc Breaker
Inverter Circuits

FIG 10 A centralized, grid-connected PV inverter. EMI: electromagnetic interference; LCL: Inductor (L) Capacitor (C) Inductor (L).

38 IEEE Power Electronics Magazine z December 2018

Three-Phase, CHB Grid-Connected Performance of dc–dc Converters
PV Inverters To adapt to the 1,500-V PV system, the
Along with module-level control, other
Since most commercial dc–dc converter is best used as a three-
current trends are high-voltage and PV modules can with- level topology so that common 1,200-V
high-power inverters. A traditional cen- switches can be used. Since most com-
stand no more than
tralized inverter topology is shown in mercial PV modules can withstand no
Figure 10. At present, the dc voltage 1,000 V between the more than 1,000 V between the active
level of mainstream inverters is 1,500 V; active part and the part and the grounded frame, a dc–dc
the inverter bridge uses the three-level converter must achieve electrical isola-
inverter circuit, and the ac output is grounded frame, a tion between the grid and the PV array.
connected to 35-kV grid through a line- dc–dc converter must In general, the isolation level of trans-
frequency transformer. This topology formers needs to be more than twice
can have a power rating of several achieve electrical isola- the grid voltage; therefore, how to
MW and an efficiency of more than tion between the grid realize high-voltage isolation while
99%. One of its drawbacks, however, is achieving small volume is a challeng-
that there is only one overall MPPT, and the PV array. ing issue worthy of study. Since the
and a 35-kV line-frequency trans- dc-dc converters are connected in
former with a large volume and weight series with the H-bridges, they need to
is required. Additionally, since it is limited by power devices, be carefully optimized and designed so that a high over-all
the power level of this topology is difficult to increase. Fig- efficiency is obtained.
ure 11 shows the hybrid PV, grid-connected power-genera-
tion structure. Compared with the centralized topology Interphase Power Imbalance
shown in Figure 10, it adds multiple, distributed MPPT in Since each PV array in each phase operates at its own
the front stage, which improves the power generation of the maximum power point, the three-phase input power may
system. However, similar to the centralized inverter, its not be the same. Consequently, the control method must
power level is limited and this structure requires a bulky be modified to deliver balanced three-phase current to the
line-frequency transformer. grid even under unequal power generation from each
To avoid this cumbersome 35-kV line-frequency trans- phase. One way to solve the interphase power imbalance
former, a three-phase isolated CHB inverter is shown in Fig- of a star-connected, three-phase CHB converter is to inject
ure 12. By configuring the number of H-bridge modules for a zero-sequence component into the converter output volt-
each phase, this topology can be directly connected to 35-kV ages. Various zero-sequence injection methods have been
power grid without a line-frequency power transformer, developed, as discussed in [14]–[17]. A fundamental-fre-
since galvanic isolation has been provided by embedded, quency, zero-sequence injection method is derived from
compact high-frequency transformers. Additionally, each
isolated dc–dc converter can connect to a PV array, so this
topology can be applied to high-voltage and high-power N
PV generation systems. This topology not only has all the
advantages of single-phase CHB PV inverters, but also has
Phase A
no leakage current problem owing to high-frequency iso-
Cell A1
lated transformers. Isolated H-Bridge
PV

dc

PV dc
DMPPT
String

PV Cell A2
DMPPT Centralized 35-kV
String Inverter ac Grid Ă Ă

… … Cell An

PV Phase B
DMPPT
String Phase C
n

FIG 11 A hybrid PV inverter system. FIG 12 A three-phase, isolated CHB grid-connected PV inverter.

December 2018 z IEEE Power Electronics Magazine 39

the instantaneous power theory [14]. A weighted, max–min
zero-sequence injection method was proposed in [15].
However, the injection of the zero-sequence components
increases the required converter output voltages, which
are usually limited by the dc-side voltages. Once the limit
is reached, the converter saturates and three-phase bal-
anced grid currents cannot be provided. To deal with
severe power imbalances, three methods, i.e., the double-
one-sixth third-harmonic injection, reduced third-harmonic
injection, and double max–min zero-sequence injection,
are proposed in [16]. Additionally, the optimal zero-se­­
quence injection and simplified, optimal zero-sequence
injection are presented in [17]. In these methods, the opti-
mal zero-sequence injection can maximize the converter’s
power-balancing capability, however, its calculation also
becomes complicated.
A better solution is to use the common dc bus, three-
phase, isolated CHB grid-connected PV inverter shown in
Figure 13, in which the parallel input ports produce a com-
mon, low-voltage dc bus connected to nonisolated dc–dc
array adaptors that can achieve MPPT of the PV array [18].
This is a typical solid-state transformer structure. It can not
only keep the converter free from interphase power imbal-
ance, but it also has multiple-channel MPPT to improve
the system’s power generation. Although this topology is a
three-stage converter structure, the front-stage dc–dc con-
verter can usually operate in a bypass state. Therefore, its
overall efficiency is not significantly lower than the topol-
ogy presented in Figure 12.
Conclusions
In this article, several multilevel topologies for PV applica-
tions were discussed in detail. The salient features of these
topologies are summarized as follows.
1) Five-level inverter technology can reduce the required
voltage ratings and the losses associated with the de­­
vices, thereby reducing the size of the cooling systems
and indirectly increase the switching frequency to
decrease the size of the inductors. However, along with
the increase in the number of devices, the power stage of
the five-level topology becomes bigger, which leads to
the increase of stray inductance and decreased system
reliability. Furthermore, the control algorithm of the
five-level topology is very complicated.
2) With module-level control, single-phase CHB inverters
have considerable market value. This topology also has
a modular structure, higher efficiency, and lower cost.
However, leakage current, power imbalance, and inter-
module communication problems are still challenging
issues that warrant further study.
3) Three-phase isolated CHB PV inverters can achieve
higher voltage and power levels, and multichannel MPPT.
Therefore, they have promising application potential.
However, high-voltage isolation and high-efficiency
dc–dc converters still need to be developed. Moreover,
the interphase power-imbalance control methods also
require further improvement. A better solution is to use
the common dc bus, three-phase, isolated CHB grid-con-
nected PV inverters.

Phase C
Phase B
Phase A
N
Isolated H-Bridge
Nonisolated

dc
Boost
dc Inductors

Number 1
Cell A1

Cell A2

Boost

Cell An n

Number m

FIG 13 A typical dc bus, three-phase, isolated CHB grid-connected PV inverter.

40 IEEE Power Electronics Magazine z December 2018

About the Authors
Xing Zhang (honglf@ustc.edu.cn) received his B.S., M.S.,
and Ph.D. degrees in electrical engineering and automation
from Hefei University of Technology, China, in 1984, 1990,
and 2003, respectively. Since 1984, he has been a faculty
member of the School of Electric Engineering and Automa-
tion, Hefei University of Technology, where he is a professor
in the Photovoltaic Engineering Research Center of Ministry
of Education. His research interests include photovoltaic
generation technologies, wind power generation technolo-
gies, and distributed generation systems.
Tao Zhao (zt_kyyx@163.com) received his B.S. degree
in electrical engineering and automation from Xi’an Techno-
logical University, China, and his M.S. degree in power elec-
tronics and power drives from Hefei University of Technolo-
gy, China, in 2014 and 2017, respectively. He is currently
working toward his Ph.D. degree in electrical engineering at
Hefei University of Technology. His research interests
include the modeling and control of power converters and
multilevel converters.
Wang Mao (hfut_maow@163.com) received his B.S.
degree in electrical engineering from the School of Electrical
Engineering and Automation, Anhui University, Hefei, China,
in 2013. He is currently working toward his Ph.D. degree in
electrical engineering and automation at Hefei University of
Technology. His research interests include power line com-
munication technologies, the modeling and control of power
converters, and photovoltaic generation technologies.
Don Tan (dong.tan@ieee.org) earned his Ph.D. degree
from the California Institute of Technology, Pasadena. He is
currently a Northrop Grumman Aerospace Systems distin-
guished engineer and a power products manager. He served
as the president of the IEEE Power Electronics Society from
2013 to 2014 and is the founding editor-in-chief of IEEE
Journal of Emerging and Selected Topics in Power Elec-
tronics. He is recognized as an authority in adiabatic power
conversion and energy systems and has pioneered break-
through innovations. He has given more than 50 keynotes
addresses and invited talks. He is a Fellow of the IEEE.
Liuchen Chang (lchang@unb.ca) received his B.S. degree
in electrical engineering from Northern Jiaotong University,
Beijing, China, in 1982, his M.Sc. degree in electrical engineer-
ing from the China Academy of Railway Sciences, Beijing, in
1984, and his Ph.D. degree in electrical engineering from
Queen’s University, Ontario, Canada, in 1991. In 1992, he joined
the University of New Brunswick, Canada, as a professor of
electrical and computer engineering. He has published more
than 350 refereed papers in journals and conference proceed-
ings. His research interests include the research, development,
demonstration, and deployment of renewable energy-based
distributed generation systems and direct load control systems.
He is a fellow of the Canadian Academy of Engineering.
References
[1] REN21. (2018). Renewables 2018 global status report. REN21. [Online].
Available: http://www.ren21.net/status-of-renewables/global-status-report
[2] IRENA. (2016). Renewable energy auctions: Analysing 2016. IRENA,
Abu Dhabi, UAE. [Online]. Available: http://www.irena.org/-/media/Files/
IRENA/Agency/Publication/2017/Jun/IRENA_Renewable_Energy_­
Auctions_2017.pdf
[3] J. Lai and F. Peng, “Multilevel converters—a new breed of power convert-
ers,” IEEE Trans. Ind. Appl., vol. 32, no. 3, pp. 509–517, 1996.
[4] E. Villanueva, P. Correa, J. Rodriguez, and M. Pacas, “Control of a single-
phase cascaded H-bridge multilevel inverter for grid-connected photovoltaic
systems,” IEEE Trans. Ind. Electron., vol. 56, no. 11, pp. 4399–4406, 2009.
[5] X. Guo, “Hardware-based cascaded topology and modulation strategy
with leakage current reduction for transformerless PV system,” IEEE Trans.
Ind. Electron., vol. 63, no. 12, pp. 7823–7832, 2016.
[6] R. Selvamuthukumaran, A. Garg, and R. Gupta, “Hybrid multicarrier mod-
ulation to reduce leakage current in a transformerless cascaded multilevel
inverter for photovoltaic systems,” IEEE Trans. Power Electron., vol. 30,
no. 4, pp. 1779–1783, 2015.
[7] Y. Zhou and H. Li, “Analysis and suppression of leakage current in cascad-
ed-multilevel-inverter-based PV systems,” IEEE Trans. Power Electron., vol.
29, no. 10, pp. 5265–5277, 2014.
[8] M. Miranbeigi and H. Iman-Eini, “Hybrid modulation technique for grid-
connected cascaded photovoltaic systems,” IEEE Trans. Ind. Electron., vol.
63, no. 12, pp. 7843–7853, 2016.
[9] L. Liu, H. Li, Y. Xue, and W. Liu, “Reactive power compensation and opti-
mization strategy for grid-interactive cascaded photovoltaic systems,” IEEE
Trans. Power Electron., vol. 30, no. 1, pp. 188–202, 2015.
[10] T. Zhao, X. Zhang, W. Mao, F. Wang, J. Xu, Y. Gu, and X. Wang, “An
optimized third harmonic compensation strategy for single-phase cascaded
H-bridge photovoltaic inverter,” IEEE Trans. Ind. Electron., vol. 65, no. 11,
pp. 8635–8645, 2018.
[11] J. Han, J. D. Jeong, I. Lee, and S. H. Kim, “Low-cost monitoring of photo-
voltaic systems at panel level in residential homes based on power line com-
munication,” IEEE Trans. Consum. Electron., vol. 63, no. 4, pp. 435–441, 2017.
[12] J. Du, J. Wu, R. Wang, Z. Lin, and X. He, “DC power-line communication
based on power/signal dual modulation in phase-shift full-bridge convert-
ers,” IEEE Trans. Power Electron., vol. 32, no. 1, pp. 693–702, 2017.
[13] Optistring Technologies, “Common line communication in cascaded
inverters,” Eur. Pat. WO2017086862A1, May 26, 2017.
[14] C. Townsend, T. Summers, and R. Betz, “Control and modulation
scheme for a cascaded H-bridge multi-level converter in large scale photo-
voltaic systems,” in Proc. IEEE ECCE, 2012, pp. 3707–3714.
[15] S. Rivera, B. Wu, S. Kouro, H. Wang and D. Zhang, “Cascaded H-bridge
multilevel converter topology and three-phase balance control for large
scale photovoltaic systems,” in Proc. 3rd IEEE Int. Symp. Power Electron-
ics for Distributed Generation Systems (PEDG), 2012, pp. 690–697.
[16] Y. Yu, G. Konstantinou, B. Hredzak, and V. G. Agelidis, “Power balance of
cascaded H-bridge multilevel converters for large-scale photovoltaic integra-
tion,” IEEE Trans. Power Electron., vol. 31, no. 1, pp. 292–303, 2016.
[17] Y. Yu, G. Konstantinou, B. Hredzak, and V. G. Agelidis, “Power balance
optimization of cascaded H-bridge multilevel converters for large-scale
photovoltaic integration,” IEEE Trans. Power Electron., vol. 31, no. 2,
pp. 1108–1120, 2016.
[18] F. Zhang, X. Ma, L. Huang, P. Xu, Y. Xuan X. Yang, X. Hao, Z. Li, “Design
and demonstration of a SiC-based 800V/10 kV 1MW solid-state transformer
for grid-connected photovoltaic systems,” in Proc. IEEE Int. Future Energy
Electronics Conf. and ECCE Asia, 2017, pp. 1987–1990.

December 2018 z IEEE Power Electronics Magazine 41
 Failure
Analysis of
the dc–dc
Converter
A comprehensive survey
of faults and solutions
for improving reliability

by Levy Ferreira Costa

and Marco Liserre

I
n many industrial applications, power interruption
is not tolerated, and a highly reliable power elec-
tronics system is required. In fact, a failure on the
power converter penalizes not only the mainte-
nance cost (to repair or change the converter),
but also the operational cost, because the service is
interrupted. Thus, this article investigates the failure in
dc–dc converters, with the aim to identify the most vul-
nerable devices and discuss solutions for improving the
converter’s availability.
More precisely, this study focuses on the series-res-
onant converter (SRC) and LLC converter, which are
very popular in industry. Through a survey carried out
involving industries and universities, the primary-side
Digital Object Identifier 10.1109/MPEL.2018.2874345
Date of publication: 19 December 2018
semiconductors and resonant capacitors are recognized
as the most susceptible device to fail. Thus, avoiding the
failure of those devices might considerably enlarge the
converter’s availability. Among the most adopted solution
in industry, the oversizing of components, redundancy
of converters, and fault-tolerant (FT) methods are high-
lighted. Each of these methods are discussed in detail in
this article.
The State of dc–dc Converters
The rapid evolution of power electronics and informa-
tion theory in the last decades enables the rise of new
­technologies in diverse areas of application, such as elec-
tric vehicles (EVs), fast-charger stations, uninterruptible
power systems (UPSs), medicine, data processing (and
data centers), and others. All of these applications con-
tain at least one dc–dc converter for adapting the voltage

42 IEEE Power Electronics Magazine z December 2018 2329-9207/18©2018IEEE


and/or controlling the power and current transferred
between distinguished buses. Figure 1 shows examples of
those aforementioned applications, in which the dc–dc
converters are highlighted. The dual-active-bridge and
SRC are the most adopted converters in this application
because of their characteristics of soft switching, low
electromagnetic interference, and high power density. In
industry, the SRCs became even more popular, and its
variant, the LLC converter, has gained more attention.
Today, the SRC and LLC are widely adopted in industry in
many applications.
The dc–dc converters are mostly used as battery
chargers in EVs, fast-charger stations, and UPSs, while
in data centers, they are also used to adapt the volt-
age and to supply the server. Thus, each application
imposes different requirements in terms of power and
voltage levels, and control for their dc–dc converters,
©istockphoto.com/triloks
as shown in Figure 1. On the other hand, all of these con-
verters have the following common requirements, inde-
pendent of the application:
1) high efficiency
2) high reliability
3) low production cost.
Indeed, the overall cost is the most relevant aspect of the con-
verter from the industry’s viewpoint, and the reliability and
efficiency directly affect the operation cost. The ­efficiency
is directly related to the energy savings in each application,
resulting then in low energy cost over the years, while reli-
able converters reduce the cost of maintenance and war-
ranties. Thus, both features deserve much attention during
the converter design.
Over the past few years, much research has been focused
on efficiency improvement of dc–dc converters, and several
design optimization methods and power converter topologies
December 2018 z IEEE Power Electronics Magazine 43
[1]–[9] have been discussed. On the other hand, a lack of infor-
mation about the reliability assessment for dc–dc converter
exists. Yet, robustness is considered even more important
than the efficiency in many industrial applications because
an interruption of the power flow can be very costly. In con-
trast, very few papers have attempted to improve the convert-
er’s reliability by the means of the FT approach [10], [11].
To improve the robustness and availability of the con-
verter, it is necessary to understand precisely the possible
failures and subsequently propose some strategy to avoid
or postpone them as much as possible. In this context, this
article aims to identify the most vulnerable devices to fail-
ure of the dc–dc converter, considering different appli-
cations, and how these failures can be avoided. To take
advantage of the industry experience and knowledge from
academia in this field, we carried out a survey to recognize
the most faulty devices, which is explained in detail in the
section “Failure Analysis of the dc–dc Converter: A Survey.”

Fast-Charger Station for EVs Data Centers

Rack/Server

dc dc dc 400 Vdc 48 Vdc

dc dc dc
dc dc dc UPS ac dc
ac ac ac dc–dc System dc dc
Converter
Application
Power Range: Output Voltage Range: Power Range: Output Voltage Range:
50 → 250 kW 300 → 1,000 V 50 → 10 kW 12 → 400 V

Power Range: Output Voltage Range: Power Range: Output Voltage Range:
1.5 → 60 kW 300 → 1,000 V 1 → 50 kW 1 → 25 kV

EV–Battery Charger Special High-Voltage Power Supply

UPS
MRI
dc dc ac 400 Vdc dc
dc dc dc ac dc dc
Batteries ac dc
dc
dc

FIG 1 An example of different applications of the dc–dc converter, pointing out the usual range of power and voltage in each
application. MRI: magnetic resonance imaging.

100
5–100 kW High Power High
23.8% 19.2%
80
Industry > Ten Years Final
Development 69.1% Product PSU/UPS
Power Level

60 71.4% 71.4% 500 W–5 kW Telecom

(%)

42.9% 42.8%

40
Five–Ten Battery
Industry R&D Years Charger (EV)
20 14.3% 23.8% 50–500 W 23.8%
Prototype
Academia < Five Years 26.8% 33.3%
LED 14.2% Low
14.3% 14.3%
0
Development Designer Status Power Level Application
Reason Experience

FIG 2 Information related to the participants of the survey (companies and universities) and status, specification, and application
of the converter. LED: light-emitting diode; R&D: research and development.

44 IEEE Power Electronics Magazine z December 2018

Thereafter, a review of possible solu- questions related to the power con-
tions proposed in the literature to verter, such as the application, volt-
avoid such failures is presented in the
To improve the robust- age/power level, devices used, and so
“Improving the Converter’s Availabil- ness and availability on. Moreover, questions related to the
ity: Solutions” section, in which the authenticity of the source were also
of the converter, it is
advantages, drawbacks, and applica- required, such as affiliation, business
bility of each solution are discussed. necessary to under- case (in case of industry), years of
stand precisely the ex­perience, and so on, to ensure the
Failure Analysis of the dc–dc validity of the collected data. The
Converter: A Survey possible failures and headquarters of companies involved in
Two different approaches can be used subsequently propose the survey are located in different
to analytically quantify the reliabil- countries, such as the United States,
ity of the converter: the statistical some strategy to avoid Canada, Italy, Sweden, Germany, Swit-
approach based on the constant fail- or postpone them as zerland, Brazil, China, Tunisia, the
ure rate [12], [13] or the physics-of- United Kingdom, and India. On the
failure (PoF) [12], [14]–[16]. The first
much as possible. other hand, some export their product
method uses the constant failure rate worldwide, e.g., Huawei Technologies
for a converter’s components accord- Co., Fulham Inc., Valeo, and Power-
ing to the military handbook MIL-HDBK-217F [17], regard- box Benelux BV. It means that most of the data obtained
less of the application or condition that the devices are sub- from these companies are not related to a unique sample of
jected to. It means that each component of the converter is the power converter but to their experiences from several
considered a black box with a constant failure rate deter- samples, which is an additional advantage for the accuracy
mined by the military handbook MIL-HDBK-217F, and then of this survey.
standard metrics (e.g., mean time between failure and Even more importantly, the survey involves compa-
mean time to failure) are adopted to quantify the robust- nies from different market segments, with lots of experi-
ness of the converter. This approach might lead to inaccu- ence and mature products on the market. It strengthens
rate results once all metrics are considered constant, the accuracy of the obtained data. Indeed, 72% of the
re­gardless of the condition, and the wear-out is not consid- answers comes from companies with more than ten years
ered. In contrast, the second methodology deeply inves- in the market. Information related to the experience of
tigates the possible causes of failures for each device, the survey participant is summarized in Figure 2. With
considering the conditions (voltage, current, temperature, respect to the application of the products involved on the
and others) and the mission profile of the converter. The survey, these data are also summarized in Table 1 and
PoF has a high level of accuracy, but it is paid off by the Figure 2. The survey covers different application prod-
high effort of modeling and computational implementation. ucts that are characterized in three different power lev-
Besides, the results are truthful for this specific converter els: low power (1500 W), medium power (500 W–5 kW),
and application (mission profile). Although the first ap­­ and high power (25 kW). Although the category of high-
proach is considered outdated, the second one requires too power application starts at 5 kW, it is not limited only to
many data and very high computational effort, and it is a few kilowatts, but it covers also converters with hun-
valid in a single application case. dreds of kilowatts.
This work does not intend to analytically quantify the
reliability of the dc–dc converters but to identify the device
most prone to fail in the field when considering different Table 1. The dc–dc converter application
applications and conditions. Moreover, only substantial and according to the power level.
trustworthy data obtained either from academia and indus-
Power Level Application
try were considered.
Low power (1500 W) Light-emitting diodes
Methodology of the Study Battery charger
PSU (auxiliary)
To collect meaningful data, a survey of the converter’s fail-
ures was carried out with participants from industry and Medium power (500 W–20 kW) UPS
Battery charger (automotive)
academia working in different segments of power electron- Power supply for space application
ics application and different places (countries). The partici- Power supply unit for motor driver
pants were asked not only technical questions related to the Data center/telecom
converter but also questions related to the application High power (25 kW) Data center/telecom
(either for research institutes and companies) and to the Distribution system (microgrid)
experience of the participants. To carefully characterize the Renewable energy system
Welding power supply
way each participant used the converter, they were asked

December 2018 z IEEE Power Electronics Magazine 45

Results
The results of the survey are summarized in Figures 3 and 4.
Figure 3 shows the most failure devices of the converter.
Regardless of the application and power level, the primary-
side semiconductor is listed as the major failure source of
the converter, followed by the resonant capacitor. These
two devices together are responsible for 72% of the total
failure faced in industry for this kind of converter. The
fault of the semiconductor on the primary side alone is
responsible for almost half of the damage to the converter,
and it results mostly in a short circuiting of the device,
according to the participants. The resonant capacitor, on
the other hand, opens the circuit during the fault, avoiding
the power being delivered to the secondary side. The last
failure is, at least, not catastrophic because the circuit is
Most Susceptible Device to Failure
9.5%
19%
47.6%
23.8%
Semiconductors: Primary Side
Semiconductors: Secondary Side
Resonant Capacitor
Control/Command
opened, avoiding further damage. Failures on the control
and command (e.g., gate driver) platform comes in third
place, followed by failures of the semiconductor on the
secondary side.
The occurrence of the failure on the SRC converter is
presented in Figure 4(a). Note that the converter fault
occurs occasionally in 60% of the cases. Only 15% of the
products covered in this survey frequently present failures.
In this group of converters, all failures reported are related
to the semiconductor on the primary side. Additionally,
these converters are applied for high power (i.e., higher
than 5 kW), and they employ IGBT devices on the primary
side. Figure 4(b) shows the type of semiconductor used
on the primary side of the converter covered in this sur-
vey. Note that most of the products use the superjunction
MOSFET (SJ-MOSFET) because of its high performance
and low price. Moreover, most applications use input volt-
ages up to 900 V, enabling the use of such a semiconductor.
Note that SiC-MOSFETs have been used in 25% of the cases,
illustrating a significant increase of the application of such
technology. The SiC-MOSFET generally has better electric
characteristics than the SJ-MOSFETs. On the other hand,
its higher price is the limiting factor for using this technol-
ogy. Because the cost is a crucial aspect in industry, most
of the companies choose to use SJ-MOSFETs instead. It is
important to note that Figure 4(b) does not intend to com-
pare the SJ-MOSFETs with the SiC-MOSFETs in terms of
reliability. This graphic shows the semiconductor type used
in the converters, and the SJ-MOSFETs have more failures
because they have been adopted more frequently. No con-
clusion can be drawn from this graphic regarding the reli-
ability of the semiconductors, which is outside of the scope
of this article.

FIG 3 The results from the survey: most vulnerable device to Improving the Converter’s Availability: Solutions
failure of the SRC/LLC converter. From the previous analysis, the semiconductors of the pri-
mary side and the resonant capacitor have been recognized
as the most vulnerable devices to fail. Most of the events
Failure Occurrence Semiconductor Used that cause the semiconductors failure are related to the
malfunction of the auxiliary circuit (e.g., gate driver, auxil-
5%
iary power supply, forced cooling system, and control sys-
15% 10%
tem) or unexpected behavior of the load. Those events lead
to three main causes of failure for the semiconductors:
25% 60% overcurrent [short circuit (SC) current], overvoltage, or
60% 25%
overtemperature. In fact, a bug on the auxiliary circuit or
an unexpected load behavior are classified as the primary
root of a failure, and they lead to the real failure mecha-
nisms. Figure 5 shows the fundamental failure mechanisms
Occasionally SJ-MOSFETs and the predominant primary events that lead to such
Rarely SiC-MOSFETs causes. As a matter of fact, the semiconductors might
Frequently IGBT Others assume two possible states, depending on the failure mech-
(a) (b)
anisms: open circuit (OC) or SC [18]. For a voltage source
­converter, which is the case for the SRC/LLC, the OC fault
FIG 4 The results from the survey: (a) occurrence of the failure
is not catastrophic because the power transfer will be natu-
and (b) technology of the semiconductors used on the primary rally interrupted. Yet, in some applications, power flow
side of the converter. interruption is not tolerable. Instead, the SC fault is the

46 IEEE Power Electronics Magazine z December 2018

 Primary Cause Failure Mechanism Failure Mode Device

Reduced Load Impedance

Converter Control Error Overcurrent

Faulty Driver Pulses for the Switches SC Current

Human Errors (Wrong Connection, etc.)

Supply Voltage Error in the Driver Stage

Emitter/Source di/dt Feedback
Overvoltage SC Semiconductor
Overvoltage on the Gate

Parasitic Oscillations in the Gate Circuit

Supply Voltage Error in the Driver Stage

Emitter/Source di/dt Feedback
Overtemperature
Overvoltage on the Gate
Parasitic Oscillations in the Gate Circuit

Reduced Load Impedance

Converter Control Error Overcurrent

Faulty Driver Pulses for the Switches SC Current
OC Resonant Cap
Human Errors (Wrong Connection, etc.)

Start-Up of the Converter

Overvoltage
SC on the Load

FIG 5 The failure mode effect analysis of the SRC/LLC converter.

main issue, because it can cause destructive damage to the

power ­converter.
Even more importantly, the SC failure is not only the
most destructive type that should be avoided, but it is
also the most likely to happen in the real application; so, it
deserves more attention. This can be observed in Figure 5,
in which the failure modes are related to the failure mecha-
nisms and primary causes for each device.
With respect to the capacitor failure, the major primary
causes are related to the start-up of the converter and load
behavior. Indeed, the start-up leads to the overvoltage of
the capacitor if a soft start-up is not used. Furthermore,
sudden variation on the load impedance entails not only
an overcurrent on the capacitor, but also an overvoltage.
Normally, most of the failures results in an OC condition
[19], and the relation between the primary cause, failure
mechanism, and failure mode are also shown in Figure 5.
Note that the FT approach for the capacitor failures has
not been discussed in the ­literature.
Avoiding failure on those devices might considerably
enlarge the converter’s availability. Thus, the most often
Methods for Avoiding Failure
Overdesign of Redundancy
FT Approach
Components of Converters
FIG 6 The most adopted solutions in industry for avoiding fail-
ure on the converter.
adopted solutions in industry and academia for such pur-
poses are presented in Figure 6 and discussed next.
Oversized Components
The simplest solution is to oversize the most critical com-
ponents. This is usually adopted in industry, mainly when
the failure cause in unknown, though this increases the
cost. In the case of LLC or SRC, this approach is very
suitable for the resonant capacitor. The component regu-
larly fails because of overstressing of the voltage or cur-
rent because of unexpected operations, like overload, SC,

December 2018 z IEEE Power Electronics Magazine 47

or start-up. As a result, the damaged faulty converter (composed by two
capacitor behaves as an open the cir- cells and one HFT) is replaced, in
cuit, interrupting the power flow.
Redundancy has been the method described in Figure 7(d),
Therefore, oversizing the capacitor employed in many only a single cell is replaced. This
is a simple and relatively cheap solu- reduces the maintenance cost.
applications in which
tion. Contrarily, oversizing the semi-
conductors is not recommended and power interruption is FT Approach
is very rarely adopted. Indeed, over- not acceptable, such The FT feature contributes to an in­­
sizing the semiconductors in voltage crease in the availability of the sys-
greatly penalizes the cost and effi- as for data centers and tem, and it is a cost-effective solution
ciency of the converter because the distribution systems. for such purpose. Several FT meth-
price and the on resistance or drop ods have been proposed in the litera-
voltage during the on-state of the ture, and most of them include a sig-
semiconductors is directly related to nificant amount of extra hardware,
the voltage rating of the device. Thus, a higher-voltage-rated such as semiconductors/leg redundancy [22], [23] or a
device means higher cost and higher losses. Oversizing series connection of fuses/switches to isolate the fault [22],
the current rating of the semiconductors might not solve [24], [25]. Yet, this method is still more cost-effective than
the problem in most cases, so then it is not adopted. the redundancy approach.
From the existing FT methods presented in the litera-
Redundancy ture, the use of a redundant leg is the most popular. It is
The most frequently adopted solution for increasing the shown in Figure 8(a) [22]. In this solution, an extra leg is
availability of the converter is to use redundant convert- used to replace the damaged leg in the case of a fault. To
ers. Redundancy has been employed in many applications isolate the faulty leg, fuses must be connected in series
in which power interruption is not acceptable, such as for with the semiconductors, opening the circuit for the SC
data centers and distribution systems (e.g., a smart trans- case. Consequently, this solution requires many addi-
former [20]). In this approach, the dc–dc converter is con- tional components, considerably increasing the cost. Yet,
ceived by connecting multiple modules as exemplified in the fuses used in series with the main devices dissipate
Figure 7(a) for a data center application. Once a module power, deteriorating the converter’s efficiency. Another
fails, the remaining ones are in charge to provide the FT approach was proposed in [25] for the full-bridge
required power, maintaining the continuity of power (FB) phase shift dc–dc converter, in which it is reconfig-
flow. There are two kinds of redundancy: the stand-by ured in an asymmetrical half-bridge (HB) converter. The
and power sharing, as shown in Figure 7(b). In the stand- proposed topology is illustrated in Figure 8(b), and it is
by redundancy, the extra converter (or converters) re­­ valid only for OC failure. Moreover, this technique also
mains inactive in normal operation, and it is activated in requires fuses in series with the main devices, deterio-
case of damage on the main converter. Contrarily, in the rating the converter’s efficiency. Thus, this method has a
power-sharing redundancy, all modules operate simul- limited application.
taneously by sharing the power amongst them, and A promising FT solution with minimum of additional
when one of them fails, the remaining ones assume the hardware and no impact on the converter’s efficiency has
total power. Regardless of the type, the redundancy been proposed in [10]. The modified topology of the SRC
ap­p roach requires an extra cost because more con- with FT capability, i.e., FT-FB-SRC, is shown in Figure 9(a).
verters are needed. Thus, it is considered a very expen- This topology is resilient to OC or SC, making it a suit-
sive solution. able solution for every kind of semiconductor fault of the
Traditionally, the redundancy approach uses an extra primary side. The basic idea of this topology is to keep
dc– dc conver ter a s a replaceable pa r t of the entire operating the converter after the fault, (i.e., SC or OC of a
system, as depicted in Figure 7(c). Alternatively, a re­­ semiconductor), but as an HB-SRC instead of as a FB-SRC,
dundancy approach using a cell of the converter as the isolating the faulty leg. In fact, damaged device is used as
replaceable part of the entire system has been reported a circuit path, and the FB-SRC became a HB-SRC, as illus-
in [21]. In this case, the dc–dc converter is based on a trated in Figure 9(b). After the reconfiguration, the output
multiwinding transformer, instead of the traditional LLC voltage (Vo) drops to half of its value, and for this reason,
or SRC, and several cells are coupled to the same high- a reconfigurable rectifier is required to keep the output
frequency transformer (HFT) as depicted in Figure 7(d). voltage constant, according to [10]. Compared to the previ-
In this approach, all cells of the converter operate si­­ ous FT solution, this one uses only one extra device and
multaneously in normal conditions, sharing the power no additional fuse.
amongst them. In case of a fault, only the faulty cell is Table 2 shows a comparison of the existing methods
isolated from the entire system, as shown in Figure 7(d). that have been described along with the proposed one,
Unlike the traditional redundancy scheme where the in terms of components and efficiency impact. As can

48 IEEE Power Electronics Magazine z December 2018

 Data Center with Redundant dc–dc Converters Stand-By Power-Sharing
dc–dc Converter dc–dc Converter
dc–dc Converter Po
P1 = 0 P1 =
dc dc 3
400 Vdc dc dc dc
dc P
ac P2 = o
dc P 2 = Po dc 3
dc Po
dc dc dc
dc P
P3 = 0 P3 = o
dc dc dc 3
dc dc dc

(a) (b)

Standard Redundancy
Redundant Module (Replaceable)

Main Module

(c)

Multiwinding Redundancy Approach

Redundant Module (Replaceable) Faulty Module

Main Module Main Module

(d)

FIG 7 The redundancy approach of the converter: (a) an example of the application of redundancy in a data center application;
(b) the types of redundancy operation (stand-by or power sharing); (c) the standard redundancy scheme of dc–dc converters; and
(d) the alternative redundancy scheme using multiwinding transformer presented in [21].

December 2018 z IEEE Power Electronics Magazine 49

 FT-Redundant Leg FT-FB Phase Shift
s1 s3 D1 D3 s1 s3
D1 D3

iL Lr Cr iL Lr

Ci Co Ci Co
Vi vp vs vo Vi vo

D2 D4 1:n
s2 s4 s2 s4
D2 D4

(a) (b)

FIG 8 Existing FT approaches: the (a) redundant leg [22] and (b) FT-FB-phase shift proposed in [25].

FT-FB-SRC
s1 s3 D1 D3 s1 s3 D1
C1 Co
iL Lr Cr iL Lr Cr
a
Vi Ci Vi Ci
vp vs vo vp vs Vo
Sf
b
s2 C2 s2 1:n
s4 D2 D4
s4 Co
D2
(a) (b)

FIG 9 (a) The FT-FB-SRC proposed in [10] and (b) its operation under fault.

the SRC, and they are responsible for almost half of the
Table 2. The evaluation of different FT
failure. The resonant capacitor is listed as the second
approaches: Redundant leg.
most susceptible device to failure, causing almost 25% of
Redundant FT-FB-Phase FT-FB- the converter’s failure. These data are valid for different
Extra Devices Leg Shift SRC power level applications and for semiconductor technol-
Semiconductors 6 4 2 ogy used on the primary side. Most companies reported
Fuses 6 4 0 that the failures of the resonant capacitors normally occur
during the start-up or overload. Both conditions result in
Capacitors 0 0 1
overvoltage of the capacitor and, subsequently, its failure.
Others — Central tap trafo —
Oversizing the resonant capacitor is normally adopted in
Applied converters FB/SRC/LLC FB SRC/LLC industry, but the unpredictable behavior of the load still
Failure type OC/SC OC OC/SC leads to this problem.
Impact on ­efficiency Yes Yes No The major failure cause, i.e., semiconductors failure,
can be overcome using a FT approach. By doing so, the con-
verter might be resilient to semiconductor failure, remain-
be noticed, the FT-FB-SRC provides an efficient and ing in operation. Among the possible methodologies, the
low-cost solution. FT-FB-SRC topology presented in this article is the most
suitable one for such purpose.
Conclusions
This article has presented a failure analysis of the SRC and About the Authors
LLC. In this study, a survey involving companies and academia Levy Ferreira Costa (lfc@tf.uni-kiel.de) received his
was carried out with the aim of recognizing the devices of the B.S. degree in electrical engineering from the Federal Uni-
converter that are most vulnerable to failure. versity of Ceara, Fortaleza, Brazil, in 2010 and his M.S.
The results from the survey shows that the primary- degree in power electronics from the Federal University
side semiconductors are the most susceptible device of of Santa Catarina, Florianópolis, Brazil, in 2013. He was a

50 IEEE Power Electronics Magazine z December 2018

power electronics design engineer at Schneider Electric,
Brazil, from 2013 to 2014, working with uninterruptible
power supply systems. From 2014 to 2018, he was a re­­
search assistant with the chair of power electronics at
the Christian-Albrechts Universität zu Kiel University,
Germany. He is currently a power electronics scientist
with ABB Corporate Research, Switzerland. His research
interests include modular converter designs, dc–dc con-
verters, and high-efficiency conversion topologies for
industrial and renewable energy applications. He is a
Member of the IEEE.
Marco Liserre (ml@tf.uni-kiel.de) received his M.Sc.
and Ph.D. degrees in electrical engineering from the Bari
Polytechnic, Italy, in 1998 and 2002, respectively. He has
been an associate professor at Bari Polytechnic and, from
2012, a professor in reliable power electronics at Aalborg
University, Denmark. Since 2013, he has been a full professor
and holds the Power Electronics Chair at Kiel University,
Germany, where he leads a team of more than 20 researchers
with an annual budget of €2 million and cooperates with 20
companies. He has received the IEEE Industrial Electronics
Society (IES) 2009 Early Career Award, IES 2011 Anthony J.
Hornfeck Service Award, 2014 Dr. Bimal Bose Energy Sys-
tems Award, 2011 Industrial Electronics Magazine Best
Paper Award, Third Prize Paper Award by the Industrial
Power Converter Committee at the 2012 IEEE Energy Con-
version Congress and Exposition, 2017 IEEE Power Elec-
tronics Society Sustainable Energy Systems Technical
Achievement Award, and the 2018 IEEE IES Mittelmann
Achievement Award, which is the highest award of the IES.
He is a Fellow of the IEEE.
References
[1] M. Liserre, G. Buticchi, M. Andresen, G. D. Carne, L. F. Costa, and
Z. X. Zou, “The smart transformer: Impact on the electric grid and tech-
nology challenges,” IEEE Ind. Electron. Mag., vol. 10, no. 2, pp. 46–58,
June 2016.
[2] M. Liserre, M. Andresen, L. Costa, and G. Buticchi, “Power routing in
modular smart transformers: Active thermal control through uneven loading
of cells,” IEEE Ind. Electron. Mag., vol. 10, no. 3, pp. 43–53, Sept. 2016.
[3] X. She, A. Q. Huang, and R. Burgos, “Review of solid-state transformer
technologies and their application in power distribution systems,” IEEE J.
Emerg. Sel. Topics Power Electron., vol. 1, no. 3, pp. 186–198, Sept. 2013.
[4] J. Wang, A. Q. Huang, W. Sung, Y. Liu, and B. J. Baliga, “Smart grid tech-
nologies,” IEEE Ind. Electron. Mag., vol. 3, no. 2, pp. 16–23, June 2009.
[5] X. She, X. Yu, F. Wang, and A. Q. Huang, “Design and demonstration of a
3.6-kv 120-v/10-kva solid-state transformer for smart grid application,” IEEE
Trans. Power Electron., vol. 29, no. 8, pp. 3982–3996, Aug. 2014.
[6] L. Costa, G. Buticchi, and M. Liserre, “Highly efficient and reliable sic-
based dc–dc converter for smart transformer,” IEEE Trans. Ind. Electron.,
vol. 64, no. 10, pp. 8383–8392, 2017.
[7] R. Yu, G. K. Y. Ho, B. M. H. Pong, B. W. K. Ling, and J. Lam, “Computer-
aided design and optimization of high-efficiency LLC series resonant con-
verter,” IEEE Trans. Power Electron., vol. 27, no. 7, pp. 3243–3256, July 2012.
[8] C. Gammeter, F. Krismer, and J. W. Kolar, “Comprehensive conceptual-
ization, design, and experimental verification of a weight-optimized all-sic
2 kV/700 V DAB for an airborne wind turbine,” IEEE J. Emerg. Sel. Topics
Power Electron., vol. 4, no. 2, pp. 638–656, June 2016.
[9] R. M. Burkart and J. W. Kolar, “Comparative h– _– v– pareto optimiza-
tion of Si and SiC multilevel dual-active-bridge topologies with wide input
voltage range,” IEEE Trans. Power Electron., vol. 32, no. 7, pp. 5258–5270,
July 2017.
[10] L. Costa, G. Buticchi, and M. Liserre, “A fault-tolerant series-resonant
dc–dc converter,” IEEE Trans. Power Electron., vol. 32, no. 2, pp. 900–905,
Feb. 2017.
[11] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, “Fault-tolerant strategy
for a photovoltaic dc–dc converter,” IEEE Trans. Power Electron., vol. 28,
no. 6, pp. 3008–3018, June 2013.
[12] H. Chung, H. Wang, F. Blaabjerg, and M. Pecht, Reliability of Power
Electronic Converter Systems, (Energy Engineering Series). Stevenage,
U.K.: IET, 2015.
[13] Y. Song and B. Wang, “Survey on reliability of power electronic sys-
tems,” IEEE Trans. Power Electron., vol. 28, no. 1, pp. 591–604, Jan. 2013.
[14] H. Wang, M. Liserre, and F. Blaabjerg, “Toward reliable power electron-
ics: Challenges, design tools, and opportunities,” IEEE Ind. Electron. Mag.,
vol. 7, no. 2, pp. 17–26, June 2013.
[15] H. Wang, M. Liserre, F. Blaabjerg, P. de Place Rimmen, J. B. Jacobsen,
T. Kvisgaard, and J. Landkildehus, “Transitioning to physics-of-failure as a
reliability driver in power electronics,” IEEE J. Emerg. Sel. Topics Power
Electron., vol. 2, no. 1, pp. 97–114, Mar. 2014.
[16] K. Ma, H. Wang, and F. Blaabjerg, “New approaches to reliability assess-
ment: Using physics-of-failure for prediction and design in power electronics
systems,” IEEE Power Electron. Mag., vol. 3, no. 4, pp. 28–41, Dec. 2016.
[17] U.S. Department of Defense. (1991, Dec.). Military handbook: Reliability
prediction of electronic equipment. [Online]. Available: https://snebulos.mit
.edu/projects/reference/MIL-STD/MIL-HDBK-217F-Notice2.pdf
[18] R. Wu, F. Blaabjerg, H. Wang, M. Liserre, and F. Iannuzzo, “Catastrophic
failure and fault-tolerant design of IGBT power electronic converters—An
overview,” in Proc. IEEE Conf. Industrial Electronics Society (IECON),
Nov. 2013, pp. 507–513.
[19] H. Wang and F. Blaabjerg, “Reliability of capacitors for dc-link applica-
tions in power electronic converters: An overview,” IEEE Trans. Ind. Appl.,
vol. 50, no. 5, pp. 3569–3578, Sept. 2014.
[20] L. F. Costa, G. D. Carne, G. Buticchi, and M. Liserre, “The smart trans-
former: A solid-state transformer tailored to provide ancillary services
to the distribution grid,” IEEE Power Electron. Mag., vol. 4, no. 2, pp. 56–67,
June 2017.
[21] M. Liserre, G. Buticchi, L. Costa, M. Andresen, “Spannungswandler,
Verfahren zu dessen Betrieb und Computerprogramm,” German Patent
DE102017104138, 2017.
[22] W. Zhang, D. Xu, P. N. Enjeti, H. Li, J. T. Hawke, and H. S. Krishnamoor-
thy, “Survey on fault-tolerant techniques for power electronic converters,”
IEEE Trans. Power Electron., vol. 29, no. 12, pp. 6319–6331, Dec. 2014.
[23] E. Ribeiro, A. Cardoso, and C. Boccaletti, “Fault-tolerant strategy for a
photovoltaic dc–dc converter,” IEEE Trans. Power Electron., vol. 28, no. 6,
pp. 3008–3018, June 2013.
[24] Y. Song and B. Wang, “Survey on reliability of power electronic sys-
tems,” IEEE Trans. Power Electron., vol. 28, no. 1, pp. 591–604, Jan. 2013.
[25] X. Pei, S. Nie, Y. Chen, and Y. Kang, “Open-circuit fault diagnosis and
fault-tolerant strategies for full-bridge dc–dc converters,” IEEE Trans.
Power Electron., vol. 27, no. 5, pp. 2550–2565, May 2012.

December 2018 z IEEE Power Electronics Magazine 51
 A
Critique
of Little Box
Challenge
Inverter Designs
Breaking from traditional design tradeoffs
S
ince the introduction of wide-bandgap (WBG)
power semiconductor devices, high power
density (PD) inverter design has become an in­­
creasingly popular topic in modern research
literature. The 2014–2015 Little Box Challenge
(LBC) cosponsored by Google and the IEEE sparked
even more interest in this area because WBG-based
designs aspired to achieve upward of 215- W/in 3 power
Digital Object Identifier 10.1109/MPEL.2018.2873992
Date of publication: 19 December 2018
52 IEEE Power Electronics Magazine z December 2018
densities. This article reviews several key topics of im­­
portance that are related to realizing such high PD gal-
lium nitride (GaN) WBG-based inverter designs. More
specifically, this article includes a comprehensive survey
of technologies employed by participating LBC finalists
in their efforts to achieve new levels of PD in 400-Vdc to
single-phase 240-Vac conversion applications while
maintaining tight specifications on dc ripple, ac ripple,
and aggressive requirements on efficiency. Both single-
phase dc-to-ac conversion and the related active energy
buffer circuit topologies used by all LBC finalists are
2329-9207/18©2018IEEE

by Clint W. Halsted and
Madhav D. Manjrekar
©istockphoto.com/nzphotonz
documented and compared. Additionally, a graphical trend
line and figure of merit (FOM) are developed that com-
pare PD vis-à-vis efficiency of all proposed solutions.
These are used to gain insight into how particular designs
were able to achieve relatively higher PD levels and to
illustrate how the GaN-based design from the Univer-
sity of Illinois at Urbana-Champaign (UIUC) was able to
achieve the highest density power conversion in this
competition. Furthermore, enhanced auxiliary devices
and techniques used in the CE+T and Virginia Tech (VT)
designs are discussed as well.
Overview
The need for high PD power conversion is expected to
increase as renewable energy interfacing requirements con-
tinue to accelerate. As converter PD increases to higher
levels, converters can be integrated within the host device
they are intended to support. These unprecedented levels
of PD are supported in part by the commercial availability
of WBG power devices and drivers. Although it has been
shown in other literature that all of the LBC finalist designs
contained WBG devices, it will be seen in this article that
the power device chosen does not necessarily guarantee a
relatively higher-density, higher-efficiency design. Other
review papers and technical documents [3]–[10] have been
published that detail inverter topologies and specifica-
tions of LBC finalist designs, including those for the active
energy buffer circuitry. However, a detailed comparative
analysis has not been performed in the previous literature.
In this article, we develop a graphical trend line and FOM
that are used to rank the LBC finalist designs. We also use
these to draw insight into inherent constraints built into the
design topologies.
WBG devices are not alone in their support of these
enhanced levels of power integration. This article also dis-
cusses auxiliary components that were used by the top
three finalists to achieve relatively enhanced performance.
Review of LBC Requirements and Implementations
The LBC participants were challenged to design and build a
2-kVA inverter with PD and efficiency exceeding 50 W/in 3
and 95%, respectively [1], [2]. The inverter was required to
have an input voltage of 450 Vdc with a 10-W series resistor
included. The included resistor would produce a ripple volt-
age on the dc bus due to the ripple current at the input
because of the single-phase inverter output. The ripple volt-
age allowed at the input dc bus was specified to be less than
3%. Additionally, the input current ripple was limited to
20%. The output voltage of the inverter was specified to
provide a 240-Vac +/− 12-Vac output with voltage and cur-
rent total harmonic distortion plus noise less than 5%.
Finally, the inverter was required to maintain a case temper-
ature of 60 °C (at 30 °C ambient) while fully complying with
electromagnetic interference (EMI) standards as specified
in FCC Part 15B. The sponsors committed to awarding a
prize of US$1 million to the team with the highest PD while
meeting all other specifications [1].
LBC Challenges
The main challenges in designing a converter at the required
PD are reducing the size of the magnetics and capacitance,
extraction of internally generated heat, and reduction of
ripple voltage and current due to 120-Hz power fluctuations.
Based on the required specifications, the overall volume of
the design is limited to 40 in 3. For a design of 1.6-in height,
all components, including the case and heatsink, must fit
within a 5-in × 5-in footprint. Because magnetics, capaci-
tance, and heatsinks typically occupy the most volume in a
December 2018 z IEEE Power Electronics Magazine 53
design, optimization of these compo- be shown, GaN has improved switch-
nents must receive attention. Active ing characteristics compared to all
energy buffers have historically relied
Because magnetics, other available devices.
on either high voltage capacitance or capacitance, and heat- High-density capacitive ­s torage
high current inductance to store energy is required to reduce the 120-Hz rip-
during the low-energy output cycle of
sinks typically occupy ple to within the limits of the LBC
the 60-Hz grid voltage. This approach the most volume in a specifications. Advances in film and
needs to be re-examined to realize a multilayer ceramic (MLC) capacitors
design, optimization
high-efficiency, high PD design. enable reduced inverters and active
of these components energy buffer size compared to elec-
LBC Solutions must receive trolytics and enjoy longer lifetimes.
Solutions to the aforementioned chal- Even higher-density capacitance will
lenges include high switching speed, attention. be needed in the future, as will be dis-
high efficiency, and high-density cussed in the “LBC Participant Imple-
capacitive storage. High switching mentations” section.
speed will allow the size of the mag-
netics to be reduced significantly. Further magnetics reduc- LBC Inverter Implementations
tion can be achieved by using planar cores with inductive The inverter design specifications for finalists in the LBC can
traces integrated inside the printed circuit board (PCB). be seen in Table 1, as reported in [3]–[10]. As listed, power
Although this technique is costlier in terms of board fabrica- densities ranged from 56 to 216 W/in 3, whereas efficiencies
tion, performance is enhanced by copper traces that are rel- fell within the 95.1 to 98.6% range. As stated previously, the
atively thin, and trace width can be adjusted by computer- highest PD of 216 W/in 3 was achieved by the UIUC team
aided tools and fabricated precisely and efficiently. The high [4], [5]. Figure 1 shows a picture of the inverter design with
switching frequency required for small magnetics means active energy buffer compared with a standard deck of play-
switching losses will exceed conduction losses in silicon ing cards. However, it has been reported that, due to power-
devices. However, WBG devices, especially GaN, have greatly up issues during the National Renewable Energy Laboratory
reduced switching losses. The lack of a body diode in GaN (NREL) testing, the UIUC team did not pass the NREL test-
devices also eliminates the traditional reverse recovery ing [14]. According to [14], only three teams successfully
losses associated with other device technologies, such as completed the 100-h testing. Winners, in order of first place
silicon IGBTs. GaN-based switching devices enable the use to third place, were as follows: CE+T, Schneider Electric,
of small magnetics in this application. and VT. Of note is that the first-place team had lower PD
High efficiency is required to reduce the size of the heat- than the UIUC team by a factor of 1.5. Unfortunately, the effi-
sink and case to within reasonable limits. Although the ciency of the CE+T and Schneider Electric teams has not
LBC specifies an efficiency as low as 95%, the size of the been reported. However, other teams did report and publish
heatsink can be reduced significantly if higher-efficiency information detailing their designs. Of particular interest is
design can be achieved. This applies equally to both the the dimensions of the UIUC design: 4.02 in × 2.42 in × 0.95 in.
design of the inverter and the design of the active energy A detailed comparison of the designs is presented later, but
buffer. WBG devices enable higher efficiencies due to the first a comparative evaluation of inverter topologies and
decreased gate charge properties of these devices. As will energy buffer designs is presented.

Table 1. The inverter design specifications [3]–[10].

Institution Efficiency (%) PD (W/in3) Dimensions (in)
TT 96.5* 91 6 × 3.69 × 1
Texas A&M 98 56 5.3 × 5.2 × 1.3
ETH !verter 95.1* 134 —
UIUC 97* 216 4.02 × 2.42 × 0.95
VT 98.6** 61 —
CE+T Unknown 145 2.5 × 1.615 × 3.41
Schneider Unknown 100 —
Electric
UT 96.9* 102 4.38 × 3.47 × 1.29

*CEC method used.

**Peak used. FIG 1 A photo of the 1-kW 240-Vac inverter with active energy
buffer from the UIUC finalists [5].

54 IEEE Power Electronics Magazine z December 2018

 It may be observed that predominantly most inverter
designs were based off the traditional full-bridge inverter
topology as shown in Figure 2(a). Teams from Taiwan Tech
(TT), Texas A&M, Schneider Electric, and the University
of Tennessee (UT) utilized the full-bridge inverter topol-
ogy [3], whereas the parallel full-bridge inverter topology
shown in Figure 2(b) was utilized by ETH Zurich and CE+T
Belgium [6], [9]. The parallel full bridge can take advantage
of increased effective switching by phase shifting a second-
ary half-bridge by 180°, also called multiphase operation.
In this case, inductor ripple is reduced by a factor of two,
which allows one to size the inductor smaller to achieve the
same ripple as a noninterleaved design. The seven-level fly-
ing capacitor multilevel inverter (FCMLI) topology shown
in Figure 2(c) was utilized by the UIUC team [4], [5]. The
FCMLI is preceded by a filter and followed by an unfolder
circuit as shown. The FCMLI produces a rectified sine
wave output between ground and the voltage bus; the
unfolder flips the polarity at twice line frequency to pro-
duce ac voltage with zero dc offset. The transistors switch
at 120 kHz but can achieve an effective switching frequency
of 720 kHz [4]. This design required the highest number (16)
of power switches but also achieved the highest PD and
smallest size. Also, it may be noted that only one inductor
is required for this design, instead of two or four. As seen
in Figure 2(d), the VT team utilized the highly efficient and
reliable inverter concept (HERIC) topology, which is simi-
lar to the traditional full bridge except for the inclusion of
the bidirectional switch bypass branch between the out-
put nodes [8]. Not included in the diagram is the required
S1a
inductance-capacitance-inductance (LCL) for filtering pur-
poses. The purpose of the bypass branch is to help decouple
the dc bus voltage from the grid voltage [8].
All of the finalist participants used GaN power devices
from GaN Systems except ETH Zurich, who used Infineon
devices, and UIUC, who used Efficient Power Conversion
(EPC) components [3]–[10]. The device information for the
Schneider Electric design was not readily available.
LBC Active Energy Buffer Implementations
Most of the active energy buffer designs have been observed
to be of the synchronous buck power decoupling type, as
shown in Figure 3(a) [3]–[10]. This circuit decouples active
power by transferring power to and from a buffer capacitor,
which allows output power to vary at double line frequency
without causing significant voltage/current ripple at the
inverter input. The TT, ETH Zurich, and CE+T participants
used this topology in their designs [3].
Texas A&M utilized the power decoupling circuit shown
in Figure 3(b) [3]. This circuit connects from one phase of
the ac output to the dc output through a third half-bridge.
The advantage of this topology is that the decoupling
capacitor carries an ac voltage in such a manner that the
capacitor is utilized to its full capacity. This circuit can be
thought of as a third leg of an inverter that helps to bal-
ance power.
UIUC used the shunt-connected full-bridge power de­­
coupling circuit shown in Figure 3(c) [4], [5]. The signal I ac
adds to the inverter current I out such that the twice line
frequency signal is canceled and I in remains constant dc.
S1b

S3a

S5b

C1
S3b

S5a

S7a

S7b

S1 S3 L1 L1

C L2 L3
S2 L2 L4
C2 S4
S2a

S8a
S2b

S4a

S4b

S6a

S6b

S8b

(a) (b)

S6a S5a S4a S3a S2a S1a S1 S3

S11 S21 S5
L C
C5 C3 C2 S6
C4 C1
S6b S5b S4b S3b S2b S1b C0 S2 S4
S12 S22

(c) (d)

FIG 2 The inverter topologies used by LBC finalists. (a) Basic full-bridge inverter topology, as in TT, Texas A&M, Schneider, and UT;
(b) parallel full-bridge inverter topology, as in ETH Zurich and CE+T; (c) seven-level flying capacitor inverter topology, as in UIUC;
and (d) HERIC inverter topology, as in VT.

December 2018 z IEEE Power Electronics Magazine 55

 This differential power processing architecture has the
advantage that its energy storage capacitor is fully utilized
S1 S3 S5 and power losses can be minimized by reducing the amount
La
S1 of power processed. However, one disadvantage is that
L a Lb C2
C1 there are four switches instead of two.
b Lc C3 Schneider Electric utilized a series-connected full-bridge
S2 C c
S2 S4 S6 active power buffer circuit shown in Figure 3(d) [7]. This cir-
cuit is like the one reported by UIUC except that it applies
(a) (b)
the decoupling capacitance in series with the input to can-
Iin
cel out the twice line frequency voltage ripple.
Iout
The remaining participants did not use active power
Iac decoupling techniques. Instead, VT opted to use a two-phase,
S1 S3
L C3 interleaved dc–dc buck converter before the dc–ac stage
to achieve power decoupling, whereas UT used a passive
C1 C2
power decoupling method [10]. It consisted of a resonant
S2 S4 notch filter tuned to 120 Hz. This filter allowed for high
attenuation at twice line frequency with relatively small
capacitance required [3].
(c)

Comparison of LBC Inverter and Buffer Technologies

C2
S1 S3 Trend-Line Comparison
L
To assist in comparison of various technologies reported by
C1 C3 LBC participants, efficiency versus PD is plotted, as shown
S2 S4 in Figure 4. As may be seen, the UIUC design is an outlier. In
all other designs, there appears to be a tradeoff between
efficiency and PD, with this design being an exception. The
(d)
trend line for the remaining designs has been observed to fol-
low the empirical equation
FIG 3 The active energy buffer topologies used by LBC final-
ists. (a) Synchronous buck power decoupling topology, as in TT, h = 100 - ^0.04 # PD h.(1)
ETH Zurich, and CE+T; (b) half-bridge power decoupling topol-
ogy, as in Texas A&M; (c) series-stacked power decoupling
topology, as in UIUC; and (d) active ripple filter in series with It may be deduced that the efficiency is reduced by 0.04%
the inverter input capacitor, as in Schneider. per 1-W increase in PD, or approximately 4% per 100 W/in 3.
It may also be seen that both GaN Systems
and Infineon GaN power devices were
selected for the designs that follow this
LBC Inverter Performance
trend, whereas the outlier inverter uses
99 an EPC part. On further investigation, it
VT
98.5 is seen that all of the inverter designs that
A&M GaN Seven-Level
98 GaN
Flying Capacitor
follow the trend line are of the traditional
Systems full-bridge type and use GaN transistor
97.5 GaN Full-Bridge
Efficiency (%)

UT Inverters UIUC power devices. The outlier inverter design

97
TT EPC by UIUC also uses GaN power devices but
96.5 uses a seven-level flying capacitor inver­­
96 ter topology.
95.5 Eff. = 100 – (0.04 × PD) Investigation of the power devices used
ETH
95 Infineon in each design may also lead to drawing a
few inferences about advantages in perfor-
94.5
0 50 100 150 200 250 mance. Power device manufacture and part
Schneider CE+T GaN number are taken from [3]. Table 2 shows
Systems
the device type used in each design. It
PD (W/in3) may be noted that the designs by Texas
A&M and UT both use the GA66508 650-V
FIG 4 The plot of the inverter FOM shows the UIUC design as an outlier, and tradi- 30-A GaN Systems device, and they both
tional designs contain design tradeoffs that are restricted to a particular trend line. operate at 100-kHz switching frequency.

56 IEEE Power Electronics Magazine z December 2018

It may also be seen that both designs shown in Figure 4 have on the x axis, as can be seen in Figure 5. As the corner of
different efficiencies and PDs, but they remain on the same the square is shifted up to a higher point on the trend line,
trend line as all of the other traditional half-bridge designs. the width of the rectangle increases from unity to a higher
This implies that PD may not necessarily correlate to device value that increases with efficiency. As can be seen, the ideal
selection. The VT and TT designs use a GS66516 650-V GaN inverter is at the upper right-hand corner of the graph. Practi-
Systems device rated at twice the current level (60 A) as cal inverters will be to the lower left of this ideal converter.
the ones previously mentioned. For the GS66516 device, it Also, practical inverter designs will be able to trade off effi-
is seen that, even at increased frequencies up to 800 kHz, ciency for PD. Switching frequency can be increased, e.g., so
neither team was able to exceed a PD metric of 90 W/in 3. that magnetics and capacitance sizes are reduced, but losses
It does appear that the higher current rated devices, with are increased. Applying this FOM to compare LBC participant
lower resistance, allow for high efficiency and PD, but as designs, we obtain the bar chart shown in Figure 6. It is seen
the switching frequency is increased, the efficiency drops that the UIUC team obtains an FOM score of 324, whereas VT
off at a much faster rate. This would be expected because achieves an FOM score of 130. The VT design has the highest
the gate capacitances increase, leading to higher switching efficiency, whereas the UIUC team has the highest PD.
losses. The Infineon CoolGaN device used by ETH Zurich
had a much lower efficiency with slightly higher PD. Also, it Size Comparison
may be observed that UIUC used 150-V, 48-A rated devices It is also of interest to investigate the size of the inverter
and operated at 120 kHz but obtained an effective switching relative to the active energy buffer circuits. The available
frequency of 720 kHz. data from the UIUC and VT teams allows the comparison,
as shown in Figure 7. It is seen that, although both teams
Development of FOM used different active energy buffer design topologies, the
Further insight into relative performance of competing volume of the designs is roughly the same, 4.9 in 3 for UIUC
inverter technologies reported for the LBC may be obtained and 6.3 in 3 for VT. However, the inverter volume difference
by development of an FOM. As stated previously, the inverter
design that achieves the highest PD rating, while also meet-
ing all of the other design requirements, was to be chosen as Table 2. The LBC power device selection.
the winning team. Based on this design goal, we develop an Voltage/ Switching Manufacturer
FOM based on PD that is useful to evaluate the relative per­­ Current Frequency and
Institution Rating (kHz) Part Number
formance of each design. Also, because efficiency affects the
heatsink design, the largest component, we select ­ef­­­­­­ficiency VT 650 V/60 A 60 GaN Systems/GS66516
as the second most important parameter after PD. TT 650 V/60 A 800 GaN Systems/GS66516
From the previous trend-line comparison, it was shown Texas A&M 650 V/30 A 100 GaN Systems/GS66508
empirically that there exists a recognizable trend line for all UT 650 V/30 A 100 GaN Systems/GS66508
five traditional half-bridge inverter designs. The FOM devel- ETH 600 V/- — Infineon/CoolGaN
oped in this section seeks to normalize the converters to this
UIUC 150 V/48 A 120 (720) EPC/EPC2003
trend line for the purposes of comparison. The VT design,
e.g., is seen to outperform the Texas A&M design because it
achieves both higher efficiency and higher PD. Additionally,
Ideal
it is desired to compare the VT design to the ETH Zurich
Converter d
design that has lower efficiency but higher PD. The FOM 100
then needs to assign a weighting factor used to normalize 1
1 d= (η – 0.95) + 1
m d1
two converters at different points along the trend line such 99 2
that converters above the trend line are assigned a higher Nonideal
FOM1 Converters
value, whereas converters below this trend line have a lower (Area of
η (%)

value. The FOM that accomplishes this is proposed as Square 1)

2 d2
95 1

FOM = &: m # ^h - 0.95 hD + 1 0 # PD, (2)

1
FOM2 Trend Line
(Area of
Square 2)
where m is the negative of the slope of the trend line, h is 0
efficiency, and PD is PD in W/in 3. Therefore,
PD1 PD2 ∞

FOM = "625 # ^h - 0.95 h@ + 1 , # PD.(3) PD

FIG 5 The development of the FOM graph showing an ideal

Graphically, the FOM can be viewed as the area of a square converter versus nonideal converters and visualization of the
on a Cartesian plan with efficiency on the y axis and the PD weighting factor.

December 2018 z IEEE Power Electronics Magazine 57

is substantial, 3.6 in 3 versus 26.7 in 3. For the UIUC design,
the buffer volume is larger than the inverter volume. For this
design, 47% of the buffer volume is composed of 365 MLC
capacitors. As can be seen in Figure 8, the efficiency of the
UIUC and VT inverters is 98 and 98.97%, respectively,
whereas the efficiency of the buffer circuits is 99 and 99.6%,
respectively. Because these are close, the size difference
does not appear to be due to differences in efficiencies.
For traditional topologies using WBG devices, there
appears to be a tradeoff between efficiency and PD, i.e.,
they follow a downward sloping trend line, as can be seen
in Figure 4. To break away from this trend line, the data
indicate advanced topologies may be required, including
multilevel topologies. The supporting technology for all of
the designs previously mentioned is enhanced auxiliary
components, which are discussed in the next section.
Review of Auxiliary Components
Used by LBC Finalists
LBC finalists used enhanced auxiliary devices that enabled
breakthroughs in their design of high PD inverters. Specifi-
LBC Inverter FOM
350
FOM (PD × Efficiency)
cally, designs from CE+T, VT, and UIUC will be discussed.
The auxiliary components important in the design of the
inverters can be categorized as follows: passives, digital
hardware, control/processing, sensors, PCB, thermals, and
the use of PC software tools. A discussion of how these
components were used and why they are important is
included in the remainder of this section.
CE+T used enhanced passive components in their design,
which contributed directly to higher-density power levels.
The use of MLC capacitors was one essential key factor
that contributed to reaching the reported level of 145 W/in 3
[6]. The design operated up to 240 kHz and used magnetics
with low loss at high frequency, which helped in achieving
the required efficiency of greater than 95%. VT used a planar
inductor core with PCB winding to help increase PD [8]. Full
utilization of magnetic components was important in opti-
mizing PD. CT+E used Litz wire in the construction of their
magnetic windings to further improve efficiency.
For the CE+T team, digital hardware was an essential
auxiliary component in their inverter design. A fast micro-
controller capable of sensing and protecting high current
events, controlling zero voltage switching (ZVS), and mini-
mizing dc bus ripple with a dynamic learning algorithm was
critical. The exact processor used is not known; however,
the UIUC team used a TMX320F28377, which is a dual-core
32-b 200-MHz microcontroller from Texas Instruments [4],
[5]. The CE+T team also used a dedicated configurable pro-

300
grammable logic device for ZVS control.
250 Control and processing algorithms were an important
200 enabling technology for the inverter designs as well. First,
150 ZVS was used to 1) control fast switching of the GaN device,
100 2) lower EMI, 3) lower reverse conduction losses during the
deadtime, and 4) lower switching losses. Second, process-
50
ing and control algorithms were used for reducing current in
0 Texas VT Taiwan UT ETH UIUC filter inductors. Third, as mentioned previously, a learning
A&M Tech algorithm was used for reducing the input bus voltage of the
Institution
active power filter. Fourth, processing was used to control
the variable switching frequency between 35 and 240 kHz
FIG 6 A bar chart of inverter FOM showing that UIUC outper- depending on output current. Fifth, control of deadtime
forms other designs. from 50 ns to 3 µs to was used to optimize switching losses.

Inverter/Buffer Efficiency Comparison

Inverter Volume Inverter Efficiency

35 100
Buffer Volume Buffer Efficiency
30
99.5
Volume (in3)

25
20 99
15
(%)

98.5
10
5 98
0 UIUC VT 97.5
Institution
97 UIUC VT
FIG 7 A bar graph showing volume size comparison of the
UIUC and VT inverter designs. The buffer volume for each FIG 8 A bar graph of the UIUC and VT inverter volume show-
design is approximately the same. ing an efficiency difference lower than 1%.

58 IEEE Power Electronics Magazine z December 2018

 Enhanced sensors for control were vias with thermal conductivity of
used by the CE+T team for success- 1.1 °C/W were used to extract heat to
ful implementation of their design. Enhanced sensors for the copper heatsink below. For the
Fast measurement of output current control were used by top layer, the inductor board was laid
using a hall-effect sensor having fast over the GaN board. The ceramic in­­
response time allowed the proper con- the CE+T team for suc- ductor air gap was positioned so
trol bandwidth and provided for the cessful implementation that air could flow through the gap
crucial overcurrent protection need­ material. Finally, the entire board
­ed to safeguard the design. For this of their design. assembly was encased in a copper
design, state observers were used for EMI shield, followed by thermal in­­
all other current measurements. This ter­­f ace m a ­­t er ia l (TIM), a nd then
helped to reduce the overall size of the another final outer copper shield that
converter by eliminating physical sensors and provided fast contained the EMC filter. An external fan forced cool air
current sensing of all of the required currents. Fast voltage through the honeycomb walls of the copper heatsink and
sensors also contributed to a successful design. over all of the internal components.
Design of the PCB has been an essential enabling com- Thermal management played a large role in the design
ponent in high-density designs. The use of proper power of the 2-kW inverter for all teams involved in the LBC. All
planes, small power loop inductance, minimal switch finalist participants used one or more copper heatsinks
node capacitance, and placement of decoupling capaci- and microfans to remove heat from within the inverter. The
tance were needed to reduce the possibility of overshoot CE+T team used a custom copper honeycomb heatsink
on device nodes. Because they can impede airflow, proper design. The team also used an aluminum oxide foil inside
placement of MLC capacitors was an important factor in the ferrite inductor air gap and enhanced TIM to remove
achieving the required thermal management benchmarks. heat from the material.
Use of an ultrathin PCB (0.012-in thick) enabled high PD The use of PC software tools was also critical for first-
design because the inverter was built using three PCB pass design success. The use of three-dimensional mechani-
boards sandwiched together with a heatsink in between. cal modeling software for the design of the inverter cases
As can be seen in Figure 9, the bottom PCB layer contained was important. The use of SPICE modeling software was
MLC capacitors for the active energy buffer and was posi- critical for ensuring proper operation of the power stage and
tioned component-side down. Next, the copper honeycomb modulation. In addition, thermal modeling software was
heatsink was laid on top of the capacitor board. Then a critical in the design of the heatsink and case. Finally, finite
layer of electrically insulating but thermally conductive element analysis was used as well, primarily COMSOL.
ceramic insulation was placed next. On top of this layer, the
GaN power device PCB was laid. Silicon foam was placed Conclusions
on the upper side of the GaN devices to apply even pres- A comprehensive review of the high PD inverter and active
sure so that thermal conductivity is maximized through energy buffer designs of the LBC finalist teams is presented
the thermal vias down to the copper heatsink. Thermal in this article. Topologies and converter specifications have

TIM

Inductors

EMC Filter
GaN MOSFET GaN MOSFET GaN MOSFET GaN MOSFET GaN MOSFET
PCB

Copper Heatsink

Ceramic Capacitors

FIG 9 A cross-section view of CE+T inverter design showing the path for air flow across components and relative area of devices.

December 2018 z IEEE Power Electronics Magazine 59

been discussed and compared. An 1997, and 1999, respectively. He is an
FOM has been developed to help associate professor at the University of
designers objectively compare high- It has been shown that North Carolina in Charlotte holding a
density converter designs. For tradi- permanent tenure. He also serves as an
tional half-bridge designs, it is shown teams could have assistant director of the Energy Produc-
that there is an inherent tradeoff improved their chanc- tion & Infrastructure Center, a research
between efficiency and PD. Inverters center at the university founded by
that use the traditional half-bridge es of winning the com- industry champions, such as Duke
topology are constrained within a petition if they had Energy, EPRI, Areva, Siemens, and
narrowly defined downward sloping Westinghouse. He holds ten U.S. and
trend line. However, it is observed
targeted the minimum international patents, has published
that the GaN seven-level flying ca­­ required efficiency. more than 55 journal and conference
pacitor topology employed by the papers, and has received multiple IEEE
UIUC team breaks away from this prize paper awards. He is a Senior
trend line and outperforms other tra- Member of the IEEE.
ditional topologies.
To fully utilize the advantages of GaN in future designs, References
advanced topologies, including multilevel structures, should be [1] Little Box Challenge. (2015). [Online]. Available: http://www.littlebox
investigated further. The active energy buffer volume con- challenge.com
sisted of 47% MLC capacitance. Higher-density capacitors [2] Google AI Blog. (2016). And the winner of the $1 million Little Box Chal-
are needed to significantly reduce active power decoupling lenge is... CE+T Power’s Red Electrical Devils. [Online]. Available: https://
volume, or new decoupling methods must be used. Because research.googleblog.com/2016/02/and-winner-of-1-million-little-box.html
the active energy buffer occupied larger volume than the [3] K. A. Kim, “Opening the box: Survey of high PD inverter techniques from
UIUC inverter, as inverter density increases beyond 216 the Little Box Challenge,” CPSS Trans. Power Electron., vol. 2, no. 2, pp.
W/in 3, the active energy buffer is expected to limit further 131–139, 2017.
system size reduction in single-phase inverter designs. It has [4] Y. Lei, “A 2 kW, single-phase, 7-level, GaN inverter with an active energy
been shown that teams could have improved their chances buffer achieving 216 W/in3 PD and 97.6% peak efficiency,” in Proc. IEEE
of winning the competition if they had targeted the mini- APEC and Expo 2016, Long Beach, CA, pp. 1512–1590.
mum required efficiency. The use of WBG technology in [5] UIUC Pilawa Group—545-rk37Vu-57827. (2015). Finalist technical
converter design is accompanied by high-speed switching approach documents. [Online]. Available: https://littleboxchallenge.com/pdf/
and thus requires an advanced PCB layout, fast controllers, finalists/57827-Tech.pdf
high-frequency sensors or state observers, and magnetics [6] Red Electrical Devils by CE+T (Grand Prize Winner), Belgium. (2015).
with low loss at high frequencies. With this, and the other Finalist technical approach documents. [Online]. Available: https://littlebox-
previously mentioned design considerations, future power challenge.com/pdf/finalists/56568-Tech.pdf
converter designs are poised to continue to advance toward [7] Schneider Electric Team, France. (2015). Finalist technical approach
ever-higher PD levels. documents. [Online]. Available: https://littleboxchallenge.com/pdf/
finalists/57137-Tech.pdf
About the Authors [8] Future Energy Electronics Center, VirginiaTech, USA. (2015). Finalist
Clint W. Halsted (chalsted@uncc.edu) received his B.S. and technical approach documents. [Online]. Available: https://littleboxchallenge
M.S. degrees in electrical engineering from the University of .com/pdf/finalists/59316-Tech.pdf
Arkansas, Fayetteville, in 1997 and 1999, respectively. He is [9] !verter Team. (2015). Finalist technical approach documents. [Online].
currently pursuing his Ph.D. degree in electrical and computer Available: https://littleboxchallenge.com/pdf/finalists/56618-Tech.pdf
engineering from the University of North Carolina at Char- [10] University of Tennessee, USA. (2015). Finalist technical approach docu-
lotte. Since 2018, he has been an adjunct instructor at the Uni- ments. [Online]. Available: https://littleboxchallenge.com/pdf/finalists/58474-
versity of North Carolina at Charlotte. His research interest is Tech.pdf
in power electronics design for renewable energy applica- [11] S. Dusmez, “Designing a 1kW GaN PFC stage with over 99% efficiency
tions. He holds two U.S. patents and has an Amateur Extra and 155 W/in3 PD,” in Proc. IEEE WiPDA, 2017, Albuquerque, NM, pp.
Federal Communications Commission license. 225–231.
Madhav D. Manjrekar (madhav.manjrekar@uncc.edu) [12] A. Gupta, “GaN-FET based grid-connected solar microinverter: Some
received his B.E. degree in instrumentation and control from design insights,” in Proc. IEEE WiPDA, 2017, Albuquerque, NM, pp.
the Government College of Engineering, Pune, India, his M. 233–237.
Tech. degree in electronic design and technology from the [13] Z. Wang Transphorm Inc., “Design and implementation of a high-effi-
Indian Institute of Science, Bangalore, his M.S. degree in elec- ciency three-level inverter using GaN HEMTs,” in Proc. IEEE PCIM Europe
trical engineering from Montana State University, Bozeman, 2015, Nuremberg, Germany, pp. 486–492.
and his Ph.D. degree in electrical and computer engineering [14] R. Pilawa-Podgurski, private communication, May 2018.

from the University of Wisconsin, Madison, in 1993, 1995,

60 IEEE Power Electronics Magazine z December 2018


by Art MacCord
Payments for Utility
Patent Infringement
W
e have not discussed the
issue of what a patent
owner gets if a court rules
that someone else i n f r i n g e d the
owner’s patent. In particular, how
do you determine the amount of
money the infringer must pay to the
patent owner?
This question recently featured in
a case involving Power Integrations
and Fairchild Semiconductor Inter­
national, Inc. Power Integrations
owned two patents on a power sup­
ply controller chip used in power
sources such as chargers for elec­
tronic devices. One patent covers
switching regulators for the pow­­
er supply controller chip. Prior-art
switching regulators were inefficient
during periods when the dc output
required little power. During these
low-power periods, prior-art switch­
ing regulators would skip on/off
cycles to decrease the dc power pro­
vided; the power remained off during
the skipped cycle. However, skipping
cycles created a loud noise and deliv­
ered power in an intermittent fash­
ion. The patent addressed this prob­
lem by reducing the frequency of on/
off cycles rather than by skipping
cycles altogether. The frequency of
on/off cycles is determined by feed­
back signals. Thus, the switching fre­
quency varies based on the feedback
signal. However, for a certain range
Digital Object Identifier 10.1109/MPEL.2018.2873995
Date of publication: 19 December 2018
Patent Reviews
of feedback signals, the frequency of sate for the infringement is not easy,
the on/off cycles does not change. so as a shortcut, in most cases the
The other patent covers a power measure is the reasonable royalty.
supply controller, which is an inte­ There has been guida nce on what
grated circuit that can perform a vari­ royalty percentage is reasonable.
ety of power-regulation functions. (The guidance is not crystal clear, so
The power supply controller includes there are still many arguments about
a multifunctional circuit coupled to how to determine a reasonable per­
receive a signal at a multifunctional centage.) The Power Integrations
terminal for adjusting a current limit case addressed the following ques­
of a power switch. The current limit is a tion: What base number of dollars is
value of current that to be multiplied by
can be used by the the royalty percent­
circuit to turn off the The switching regula- age to arrive at a rea­
power switch when sonable royalty dol­
tor increases the effi-
the amount of cur­ lar amount?
ciency of the chip
rent passing through The easy answer
the power switch when little output is that the number of
reaches the threshold power is required. dollars is the selling
value. The switching price of the infring­
regulator increases ing goods times the
the efficiency of the chip when little number of items sold, and that can be
output power is required. the case if the patent covers the entire
The chips that Fairchild sold were product that was sold and found to
found to infringe, and the court was infringe. But previous binding cases
tasked with determining the award of indicate that a patent owner seeking
money (in legal terms, these are reasonable royalty damages for an
called damages, as in a measure of infringing product that has both pat­
how much the defendant has to pay ented and unpatented features must
for damaging the plaintiff). apportion the damages only to the
The statute passed by Congress patented feature. The patent owner
says that “…the court shall award the must identify the smallest saleable
claimant damages adequate to com­ unit that infringes the patent, and for
pensate for the infringement but in multicomponent products with signif­
no event less than a reasonable royal­ icant unpatented features, the paten­
ty for the use made of the invention tee also has the burden of proof to
by the infringer, …” (U.S. Code, Title apportion the selling price among the
35, Section 284). Determining the features when calculating the base for
amount that is adequate to compen­ a reasonable royalty.
December 2018 z IEEE Power Electronics Magazine 61
 In the Power Integrations case, the of the infringing product that the pat­
patent owner said it was entitled to ent covered. One takeaway from this
the entire market value of the chip or case is to include patent claims to Get in the
the controller chip,
contending that the
cover various fore­
seeable commercial conversation!
efficiency increase products that in­­clude
One takeaway from Want to collaborate and get involved
was the feature that the technological
this case is to include with the IEEE? Use social media!
drove sales of the ad­v ance, so that if
chips covered by the patent claims to cover infringement occurs, Follow and engage with the
pat ent . T he U.S. various foreseeable the recoverable roy­ IEEE on YouTube, LinkedIn,
Court of Appeals for commercial products alty base is as large Facebook, and Twitter!
the Federal Circuit as possible. For a list of registered IEEE sites,
that include the tech-
agreed that people If this technology visit www.ieee.org/about/social_media.
nological advance.
bought the infringing seems old-fashioned,
chip instead of earli­ it may be because
er-generation chips the patent applica­
because of the efficiency increase, but tion that became the litigated patent
it noted that the royalty base could was filed in 2000, so we are not refer­
still not be the full sales price of the ring to 2018 state-of-the-art technology.
chip because Fairchild’s chip had
other significant features. Fairchild About the Author
indicated that its chips had jittering Art MacCord (amaccord@maccord
and other features that some custom­ mason.com) has practiced patent,
ers found appealing. The federal court trademark, copyright, and trade-secret
of ap­­peals said that the patent owner law for over 35 years and is a gradu­

IMAGE LICENSED BY INGRAM PUBLISHING

did not bear its burden of proving that
the other features of the Fairchild
chips did not influence purchasing
decisions and sent the case back to the
trial court to det er m i ne a ro­­yalty
base properly apportioned to the part
ate of the University of Virginia and
George Washington University Law
School. He currently practices with
MacCord Mason PLLC in Greensboro,
North Carolina.


62 IEEE Power Electronics Magazine z December 2018


by Victor Veliadis
Accelerating Commercialization of
Wide-Bandgap Power Electronics
S
ilicon (Si) power devices have
dominated power electronics
because of their low-cost vol-
ume production, excellent starting
material quality, ease of processing,
and proven reliability. Although Si
power devices continue to make sig-
nificant progress, they are approach-
ing their operational limits, primarily
because of their relatively low band-
gap and critical electric field, which
result in high conduction and switch-
ing losses and in poor high-tempera-
ture performance. Si carbide (SiC) and
gallium nitride (GaN) power devices
are revolutionizing power electronics
because of their favorable material
properties, which allow for highly
efficient power devices with a re­­
duced form factor and reduced cool-
ing requirements.
Several government-funding ini-
tiatives around the world promote
the adoption of wide-bandgap (WBG)
power electronics to exploit their
energy saving and technological
innovation promise. In the United
States, the Advanced Manufacturing
Office of the Department of Energy
and North Carolina State University
formed PowerAmerica, a National
Network for Manufacturing Innova-
tion Institute. PowerAmerica focuses
on enabling large-scale production of
WBG power electronics with the goal
of revolutionizing energy efficiency
Digital Object Identifier 10.1109/MPEL.2018.2875169
Date of publication: 19 December 2018
across multiple application plat-
forms. It started operations in 2015
with a five-year budget of US$140 mil-
lion and a mission to address manu-
facturing gaps in WBG power tech-
nology to enable U.S. leadership in
job creation, energy savings, and
technological innovation.
In the United States, the devel-
opment of advanced WBG crystal
growth, wafer fabrication, and device
processing technologies owe their
beginnings to the support from a num-
ber of U.S. government programs.
Beginning in the late 1980s, organiza-
tions such as the Air Force Research
Laboratory, Army Research Laborato-
ry, Office of Naval Research, Missile
Defense Agency, and the Defense
Advanced Research Projects Agency
sponsored hundreds of millions of dol-
lars and decades of work at universi-
ties, industry, and government labora-
tories. Initially, this work focused on
proof-of-concept critical enabling
technologies such as high-quality sub-
strates, epitaxy, and unit process
steps like ion-implantation, implant
activation, gate oxidation, and so on,
to assure a domestic source for cur-
rent and future U.S. Department of
Defense system needs. WBG devices
have now advanced well past this
point with the introduction of the first
commercial SiC Schottky diode in
2001. However, widespread commer-
cialization of WBG power electronics
has been slow, which is primarily
attributed to two factors:
December 2018 z IEEE Power Electronics Magazine 63
Expert View
1) high cost: chiefly due to low man-
ufactur ing volumes a nd dedi-
cated foundries that are not fully
loaded, WBG devices are more
expensive than their mass-pro-
duced Si counterparts
2) workforce lacking expertise in
WBG technologies: the power elec-
tronics industry is traditionally
slow to change and adapt to new
technologies.
These two factors have limited
interest in WBG devices for real ap­­
plications, and the benefits in terms
of higher efficiency, reduced weight
and volume, and lower overall sys-
tems cost have not been ubiquitously
quantified. With the value proposi-
tion of WBG system integration not
being shared in a widespread man-
ner, industry executives have been
reluctant to commit internal research
and development resources, further
hindering large-scale commercializa-
tion. In addition, many of the WBG
fabrication, packaging, reliability,
ruggedness, and system insertion
issues that must be resolved for opti-
mal WBG device operation in actual
applications are only now being rigor-
ously addressed.
I was named the deputy executive
director and chief technology offi-
cer of PowerAmerica in 2016 after
21 years in the semiconductor indus-
try, where my technical work includ-
ed the design, fabrication, and testing
of 1–12-kV SiC static induction transis-
tors, JFETs, MOSFETs, thyristors,
junction barrier Schottky, and PiN States, WBG power devices can be
diodes, as well as GaN radio frequen- manufactured with the relatively
cy switches for advanced radar sys- small investments necessary to sup-
tems on 150-mm GaN-on-Si wafers. port unique WBG fabrication steps
In my PowerAmerica role, I manage a such as high-temperature implanta-
budget in excess of US$30 million per tion and annealing, contact forma-
year that I strategically allocate to tion, backside processing, and so on.
over 35 industrial, university, and This requires WBG-specific equip-
national laboratory projects to enable ment at a cost of US$12–15 million
U.S. leadership in WBG power elec- shared between PowerAmerica and
tronics manufacturing, work force the foundry partner. Minimizing capi-
development, job creation, and ener- tal expenditures by exploiting the
gy savings. My approach for catalyz- mature Si-processing capability low-
ing the manufacturing of low-cost ers fabrication costs, provided the
SiC and GaN power electronics is foundry is loaded close to capacity
schematically depicted in Figure 1. with standard Si processes running
PowerAmerica funds building-block on the same line. In addition, aggre-
projects in multiple areas of the gating the demand for WBG sub-
WBG supply chain that synergisti- strates and epilayers in a few volume
cally culminate in large-scale WBG foundries contributes to lower mate-
power electronics adoption (green rial costs. Lower process costs in a
boxes). The red boxes represent tech- fully depreciated and Si+SiC capa­
nology areas the advancement of city-loaded foundry coupled with
which can boost WBG growth but decreased material costs lead to sig-
that are presently outside the Powe- nificant price reductions for WBG
rAmerica ­mission. devices. This approach offers a new
WBG device fabrication in large- opportunity for outdated Si foundries
volume Si foundries exploits econo- that have not kept up with the chan-
mies of scale and is key in lowering nel length reductions of the last two
cost. Through repurposing 150- and decades to continue manufacturing
200-mm Si foundries in the United legacy Si parts while ramping up SiC
Large-Scale WBG Power Electronics Adoption
Lowers Cost of Devices and Power Electronics
Large-Volume Manufacturing
Creates Device Demand
Funding Areas: Develop
• Foundry, Devices Demonstrate Compelling
Peripheral
• Modules, Reliability System Advantages
Technologies
• Power Electronics Applications
• Education
Establish Reliability
Improve Develop WBG Modules Train
Streamline
Material and Circuits, Reliability Workforce
Fabrication
Quality Work in WBG
FIG 1 PowerAmerica funds projects in areas that synergistically culminate in large-
scale WBG power electronics adoption (green boxes). The red boxes represent rele-
vant technology areas that are presently outside the PowerAmerica mission.
64 IEEE Power Electronics Magazine z December 2018
fabrication that requires relatively
modest 0.5-μm design rules.
Another area of PowerAmerica’s
work is WBG device packaging and
modu le s. This is an area in which
the United States is highly deficient
because semiconductor assembly
and packaging was off-shored in the
1970s. Standard Si modules do not
allow WBG devices to reach their full
performance potential. Thus, WBG
devices offer a unique opportunity
for industry growth because they
require modules with reduced para-
sitic inductance, reduced thermal
impedance, higher-voltage isolation,
and higher-temperature capability.
PowerAmerica funds WBG modules
with special layouts that minimize
parasitics, utilize new base-plate
materials for lower thermal imped-
ance, allow for double-sided cooling,
feature low inductance, and employ
new potting compounds for higher-
temperature operation.
In addition to better performance,
reliability and ruggedness are prerequi-
sites for WBG wide adoption. Improve-
ments in material quality and fabrica-
tion contribute to device reliability.
Ruggedness is one of the numerous
device design tradeoffs. PowerAmerica
has funded honest broker ruggedness/
reliability testing efforts to build confi-
dence in the use of WBG devices in
power electronics applications.
A workforce well trained in WBG
power electronics is key in creating
the large device demand that will spur
volume manufacturing with its cost-
lowering benefits. PowerA mer ica
hands-on projects, carried out by uni-
versities and their industry collabora-
tors, provide upper undergraduate
and graduate students with real-world
WBG power electronics experience. In
the next few years, some of these stu-
dents will enter the workforce and
accelerate the insertion of WBG devic-
es in industrial products. Other stu-
dents will become faculty members at
universities and train a new genera-
tion of students creating a WBG edu-
cation snowball effect. Additional
PowerAmerica educational activities
include industry-driven WBG short
courses that are taught by world-renown experts
SOLUTIONS P R OT E C T I O N
A N D SW I TC H I N G

from around the United States, WBG tutorials pre- FOR E-MOBILITY F O R B AT T E R Y
PAC K / B D U / M S D
sented at major conferences, and internship oppor- χp SERIES
tunities with WBG-trained students.
Finally, PowerAmerica projects highlight the
compelling advantages of WBG devices in power EVpack-fuse
electronics systems that include smaller weight/
volume, higher efficiency, and a reduced bill of
materials. New circuit topologies, novel gate driv-
χs SERIES
ers for faster switching and protection, and print-
P R OT E C T I O N ,
ed circuit board layouts that minimize inductance COOLING, &
and eliminate ringing are among the many demon- CONNECTIVITY
FOR INVERTER/
strated technological innovations. PowerAmerica P D U & B AT T E R Y
projects that prove the WBG power electronics MODULE
value proposition include laptop adapters, photo- M-fuse
DC FUSES LIQUID
voltaic (PV) inverters, uninterruptible power sys- COLD PLATE
HEAT SINKS

tems, data centers, electric vehicle (EV) fast char-

gers, EV on-board chargers, solid-state circuit SOLUTIONS
breakers, microgrid power conditioning systems, F O R E V/ H E V
DC FUSE
medium voltage variable speed drives, traction AND HYB BRAT
I D DT
CEPRY RY-O FUSE
inverters, and auxiliary power converters. R HEYLB AT
RID E DP O W ECRO ROELLI NAYG
FLEXIBLE MONITORING BUS BAR
DC
Over the past eight years, prices of commercially A P P L I C AT I OBN U SSB A R S
available SiC MOSFETs have dropped by approxi-
mately 80% as a result of an increase in manufac-
turing volume, technological innovation, and an
increase in wafer area to the current 150 mm. Pres- E P. M E R S E N . C O M
ently, prices of WBG devices are about three to four
times higher than those of similarly rated Si compo-
nents. As PowerAmerica continues to implement its
lower-cost WBG manufacturing strategy and with
the introduction of 200-mm wafers in the near
future, WBG devices are expected to reach a 1.5-
times cost parity with Si within five years. It should
be noted that device cost is only one element of the
system’s bill of materials cost. For example, with
the system-level simplifications that WBG devices
introduce (reduced size and weight of passives,
reduced system cooling requirements, and so on),
the overall cost of a SiC-based PV system is now
comparable to that of its Si-based system counter-
part. Also, this is before the efficiency savings over
the life of the WBG system are considered.
PowerAmerica is a 46-member-driven institute
accelerating the commercialization of WBG power
electronics. I am greatly indebted to my team and
Department of Energy Technical Manager Al Hef-
ner for valuable technical contributions and gener-
ous support.

About the Author

Victor Veliadis (jvveliad@ncsu.edu) is the deputy
executive director and chief technology officer at
PowerAmerica and professor of electrical and com-
puter engineering at North Carolina State Universi-
ty, Raleigh. He is a Fellow of the IEEE.

 Society News
by Ashok Bindra

ECCE Plenary Talks Highlight

Challenges in Energy Conversion

T
his year marked the tenth anni­
versary of the annual IEEE
Energy Conversion Congress
and Exposition (ECCE), which was
held 23–27 September at the Oregon
Convention Center in Portland. Spon­
sored by the IEEE Power Electronics
Society and the IEEE Industry Appli­
cations Society, ECCE boasted a
record number of digests from across
Digital Object Identifier 10.1109/MPEL.2018.2875500
Date of publication: 19 December 2018
the globe. After welcoming the attend­
ees, General Chair Avoki M. Omekan­
da proudly stated, “This popular con­
ference is already bigger and better
than previous years. This year the
pivotal international conference was
attended by a record 1,789 profes­
sionals, practicing engineers, and
researchers working on various as­­
pects of energy conversion systems
and technologies.” While ECCE is
known for presenting the latest in
energy conversion research, along
with innovations in traditional compo­
nents, this year the focus was on the
challenges faced in the energy conver­
sion industry.
Five plenary speakers, ­f eaturing
Victor Veliadis of PowerAmerica,
Stephanie Watts Butler of Texas Instru­
ments, Sean James of Microsoft, Jason
Busch of Pacific Ocean Energy Trust
(POET), and Jiaqi Liang of Hyperloop
One, emphatically conveyed this mes­
sage. PowerAmerica Chief Technology
Officer Veliadis kicked off the plenary

SiC Device Market Size Split by Application

Rail (Including Auxiliary Power)

1.2 PFC/Power Supply
EV/HEV (Including On-Board Charger) Other
Wind CAGR2020–2022 = 40% UPS
xEV Charging Infrastructure Motor Drive
PV
Motor Drive (Including Air Conditioning)
UPS PV
(US$B)

0.6 Others (Oil and Gas, Military, Medical, R&D,...)

xEV Charging
CAGR2016–2020 = 28%
Wind
EV/HEV

PFC/PS

Rail
0
2016 2017 2018
2019 2020 2021 2022
Year
(Source: Power SiC: Materials, Devices, Modules, and Applications Report, Yole Développment, August 2017.)

FIG 1 SiC devices are projected to capture US$1 billion in device market shares by 2022. EV: electric vehicle; HEV: hybrid EV; PV:
photovoltaic; R&D: research and development; CAGR: compound annual growth rate; PFC: power factor correction; xEV: hybrid
and electric vehicles; PS: power supplies; UPS: uninterruptible power supply. (Data courtesy of PowerAmerica.)

66 IEEE Power Electronics Magazine z December 2018

 PEM Ad _Layout 1 16/04/2013 09:19 Page 1

CWT wide-band
AC current probes
PowerAmerica is
addressing these • Thin, flexible, clip-around
ac current probes
challenges by funding • Wide-band from below 0.01Hz
projects in multiple to greater than 20MHz (-3dB)
• Ideal for power electronics CWT Standard
areas of the WBG • Robust coils
development work
supply chain that will in lengths from
• Current ratings 30A 300mm to > 1m
synergistically to 600kA
• Insulation voltage
culminate in 10kV peak (coil
thickness 8.5mm)
large-scale WBG CWT
Ultra-
power electronics mini CWT LF
adoption. CWT
Mini

• Coil size
• Two miniature coil (max 1.7mm • Extended low-
session. His talk, “SiC Power Devices: High-Impact sizes (3.5mm or cross-section), frequency cut-off
Applications and Path to Wide Adoption,” indicated 4.5mm) up to 80mm length • Available with
200mm length • Insulation voltage standard or
that silicon carbide (SiC) power devices are mak­ mini coil
• Insulation voltage 1.2kV peak
ing progress and projected to capture a US$1 billion 2kV or 5kV peak
device market share by 2022 (Figure 1). As per the
Please contact us to discuss
talk, SiC devices are expected to displace their in­ your application
cumbent silicon counterparts in high-impact ap­
plications, such as variable frequency drives for ef­ Power Electronic Measurements
ficient high-power electric motors at reduced overall
system cost; automotive power electronics with re­ Tel: +44 (0)115 946 9657 www.pemuk.com
duced losses and relaxed cooling requirements;
novel data center topologies with reduced cooling
loads and higher efficiencies; more electric aero­
space with weight, volume, and cooling system re­ Battery technology
ductions contributing to energy savings; and more
efficient, flexible, and reliable grid applications with calls for the
a reduced system footprint.
Veliadis’ talk suggested that two factors impeding best protection
WBG adoption include high cost from limited manu­
facturing volumes in dedicated foundries and the
lack of a workforce with experience in wide-bandgap
there is
(WBG) technologies. PowerAmerica is addressing
these challenges by funding projects in multiple ar­
eas of the WBG supply chain that will synergistically
culminate in large-scale WBG power electronics adop­
tion. Concurrently, he added, it is also training the
workforce in WBG technology while the organiza­ SIBA
tion is working on creating jobs in the field. Toward
that goal, PowerAmerica is also offering short cours­
es and workshops for power electronics engineers
in the industry, noted Veliadis, who is also an IEEE
Perfect solutions for your
Fellow and a professor in electrical and computer en­ energy storage units –
gineering at North Carolina State University, Raleigh. SIBA fuses in battery installations
Texas Instruments’ Technology Innovation Ar­
Protect your investments, and those of your clients.
chitect Stephanie Watts Butler was the second Protect your business. With us. The world over.
keynote speaker. In her talk, “Power Semiconduc­ SIBA is active around the world and has eleven subsidiaries:
tors: Enabling a Powerful Decade of Changes,” she USA, China, Russia, UK, Denmark, Netherlands, Poland,
Austria, Czech Republic, South Africa and Singapore.
suggested that advances in power semiconductors
in the last decade have driven power electronics to SIBA GmbH (Headquarters and Production) Our Protection.
Borker Str. 20-22 • D-44534 Lünen
T.: +49 (23 06) 70 01-0 • www.siba.de Your Benefit.
Inquiries: inq01@siba.de
new heights, resulting in new levels
of power density, efficiency, and form
factors. As a result, according to But­
ler, it has enabled an explosive growth
in electronification in industrial, con­
sumer, and automotive markets.
Sean James, Microsoft’s director of
energy research, presented “Data Cen­
ters: Disruptive Facility Architectures
with Fuel Cells and Load Side Integra­
tion.” He discussed the use of fuel cells
for feeding power-hungry data centers,
which consume 2% of the U.S.’s elec­
tricity. According to James’ presenta­
tion, Microsoft has been researching
and testing a new architecture that in­
tegrates a simple solid oxide fuel cell
(SOFC) system with a server. Besides
cost savings and high efficiency, the
SOFC system also simplifies design,
stated James. Unplugging the power
grid from data centers will translate
into big cost savings while significantly
reducing carbon emissions.
The next keynote was delivered
by Jason Busch, executive director of
POET. He spoke about creating clean
energy from the motion of the ocean.
According to Busch, POET has been
engaged in various aspects of marine
renewable energy for more than a
decade. With a mission to promote
the responsible development of ma­
rine renewable energy, POET has
worked to advance marine renewable
energy technologies toward commer­
cialization. To that goal and with over
US$14 million of funding from the state
of Oregon, the company has fund­­­e d
technology research and ­development,
environmental studies, stakeholder
outreach, education, and policy devel­
opment. As a result, Busch indicated
that “worldwide marine renewables
are quickly tracking toward commer­
cial viability.” By comparison, he add­
ed, tidal energy is more advanced than
wave energy.
The last speaker of the session was
Director of Power Electronics Jiaqi
Liang of Hyperloop One. In his talk,
“Hyperloop: Creating the Future of
Transportation,” Liang shared Hy­
perloop One’s vision of making high-
speed transportation effortless and
affordable. Because advanced elec­
tromagnetic and electrical energy con­
version systems are some of the key
enabling components for Hyperloop,
the company has leveraged the latest
computational and optimization tools
to predict system performance, iterate
its designs with less time, and develop
prototypes at a much lower cost. Liang
said that Hyperloop One’s full-scale
DevLoop testing facility in Nevada is
the only one of its kind in the world,
allowing the company to test and vali­
date its prototypes at scale and to in­
tegrate various complex components
and subsystems—all in the controlled
environment of the tube.

by Gerard Hurley

IEEE Power Electronics Society

Celebrates Its 30th Anniversary

T
he IEEE Power Electronics
Society (PELS) celebrated the
30th anniversary of its foun­
dation at a town-hall meeting in Port­
land, Oregon, where the annual IEEE
Energy Conversion Congress and
Expo (ECCE) took place 23–27 Sep­
tember 2018 (Figure 1). The town
hall meeting was also live streamed
to the worldwide membership of
PELS. It was opened by PELS Presi­
dent Alan Mantooth, who introduced
the Committee Board members and
thanked them for their tireless ef­­
forts, dedication, and service to
the membership.
FIG 1 Pictured at the town hall meeting are (from left) PELS History Chair Gerard Hur-
Digital Object Identifier 10.1109/MPEL.2018.2874756 ley, PELS President-Elect Frede Blaabjerg, PELS Founding President John Kassakian,
Date of publication: 19 December 2018 and PELS President Alan Mantooth.

68 IEEE Power Electronics Magazine z December 2018

 Industry Leading Industry Leading

ALLOY POWDER CORES ALLOY POWDER CORES

While there are
Design Software Design Software
numerous symposiums
and workshops
sponsored by PELS
across the globe, its
flagship conferences,
• Trusted the IEEE Applied
Power Electronics
• Innovative
Conference and Powder Core Solutions for
• Collabrotive
Exposition and ECCE, Lighting Applications
• Experienced
continue to grow
Po
internationally. AC Input Differential-Mode Choke, Resonant Inductor, Au
Flyback Inductor, DC-DC Convertor Inductor, and DC
Power Factor Correction Boost Inductor. AC Inp
DC-DC Converter Inductors & Output Filter Inductors
Power electronics drives the 21st-century en­
ergy revolution. As it grows to become universal
as the enabler of nearly everything that processes
or uses energy, the opportunities and future are
bright. PELS has grown by leaps and bounds in the
last 30 years to become a global organization with
Micrometals
more than 9,000Alloy
members Powder Core
worldwide. Division
Today, it has Micrometals Alloy Powder Core Division Micro
over Anaheim CA, USA • Shenzhen China in
150 Chapters, including Student Chapters, Anaheim CA, USA • Shenzhen China
www.micrometalsarnoldpowdercores.com
all ten Regions of the IEEE around the world—and www.micrometalsarnoldpowdercores.com www
that count is growing each year. While there are nu­
merous symposiums and workshops sponsored by
PELS across the globe, its flagship conferences, the
IEEE Applied Power Electronics Conference and
Exposition (APEC) and ECCE, continue to grow
internationally. This year, APEC 2018 flaunted a
total registration of over 5,000 attendees represent­
ing some 45 countries, while ECCE has become a
premier conference for energy conversion profes­
sionals and researchers in many countries across IEEE POWER ELECTRONICS
the world. MAGAZINE APEC ISSUE
Vibrant and growing, PELS is now a resource for
power electronics professionals and researchers
nationally and internationally. To effectively serve PELS SOCIETY a sponsor
its global membership and the ever-evolving power of APEC 2019 will publish
electronics field, it has numerous key publications,
including IEEE Transactions on Power Electronics,
their annual APEC Issue
which has become a top-referenced journal among
all IEEE publications. PELS organizes global com­ This first Quarter Issue contains
petitions like Empower a Billion Lives (EBL) to in­
novate scalable solutions that will bring electricity
a PREVIEW of APEC,
to more than a billion people living in energy-pover­ and extra copies of PELS magazine
ished regions of the planet. will be given out to the attendees
PELS History Chair Gerard Hurley gave a brief
background of the Society and described its early
evolution from an IEEE council to Society. PELS For full details, email
has its origins in the Power Electronics Coun­
cil that was formed in 1983. Many people who had
a clear vision of the need and benefit of a PELS
Walter Chalupa
worked tirelessly to make it happen. To quote Trey PELS Magazine NSM
e-mail:chalupapels@aol.com
Burns, the first council president:
“We had to start life as a council, a
group of member Societies rather
than individual IEEE Members, be­
cause not everyone shared our vision
at that time. Some felt that ‘power
electronics’ was covered adequately
in other Societies and that there was
no need for yet another IEEE Soci­
ety.” The individuals involved in these
efforts had been involved in the early
conferences related to power elec­
tronics. The Power Conditioning Spe­
cialists Conference was sponsored by
the IEEE Aerospace and Electronics
Systems Group and held in April 1970,
and in 1973, it became the Power Elec­
tronics Specialists Conference. That
conference later evolved into ECCE,
which, along with APEC, is one of the
flagship conferences of PELS with
ECCE-USA, ECCE-Europe, and EC­
CE-Asia. All of this activity eventu­
ally led to the establishment of PELS,
under the leadership of its founding
President John Kassakian.
PELS President Alan Mantooth
summarized the recent accomplish­
ments and activity within the Society.
He said, “We sustain a better than 5%
growth rate and enjoy a high member-
retention rate as well. As diversity and
inclusion are one of the cornerstone
goals of our strategic plan, two of
our fastest-growing segments include
Young P rofe s sion a l s and Women
in Engineering.”
He spoke about the leadership role
PELS has taken in helping to establish
and nurture the IEEE Transportation
Electrification Community (TEC).
TEC was created to be the central
hub of focus for transportation elec­
trification activities across 11 IEEE
Societies and the IEEE Standards As­
sociation, acting as the key connec­
tion point for external stakeholders.
TEC embodies a global community of
engineers, academics, industry part­
ners, and passionate practitioners
70 IEEE Power Electronics Magazine z December 2018
for rail, aerospace, marine, and auto­ and opportunities related to cyber­
motive connectivity, autonomous arti­ physical security as it pertains to pow­
ficial intelligence, manufacturing best- er electronics. Cyberphysical security
practices, standards development, and is a vital element in energy delivery,
research information transportation, in­
sharing for coordina­ dustr ial a nd com­
ted applications. The TEC was created merci a l s y s t em s,
PELS President- to be the central and other applica­
Elect Frede Blaab­ t ion a rea s where
hub of focus for
jerg spoke about the power electronics is
transportation relied upon for op­
future of PELS, its
strategic goals, and electrification eration. The main
major foundational activities across 11 focus of the PELS
i n it iat ive s. He ex­ IEEE Societies and the Cyber-Physical Se­
plained that the over­ IEEE Standards curity Initiative is to
arching goal of PELS define design-phase
Association, acting as
is to ensure that it power elec t ron ic
the key connection systems hardening
stays relevant to its
members, in today’s point for external and to communicate
progressive and dy­­ stakeholders. the cyber physical
namic global econ­ requirements within
omy. PELS has three power electronics.
major ideas: EBL, the PELS Cyber-­ I T RW w a s launched to ­identify
Physica l Secur ity Initiatives, and and actively support the future re­
the International Technology Road­ search and technology developments
map for Wide-Bandgap Power Semi­ of wide-bandgap power semicon­
conductors (ITRW). ductors by prov iding r e fe r ence,
EBL is the humanitarian and phil­ guidance, and services to their use
anthropic program and directly ad­ and application and, thereby, pro­­
dresses the socioeconomic challenge vide a reliable and comprehensive
of ending energy poverty by encourag­ v iew on t he St r a t eg ic R e s e a r c h
ing and funding the accelerated devel­ Agenda and Technology Roadmap.
opment of in-field renewable energy Earlier this year, the ITRW initiative
generation, access, and deployment took the first step to establishing the
of scalable technological solutions— definitive road map for this techno­
including the physical, technical, so­ logical area.
cial, and business elements needed to Participants were able to ask ques­
operate the system sustainably. Three tions, many of which were regard­
billion people live in energy poverty, ing how PELS could a ssist mem­
including 1.1 billion people with ab­ bers in networking and exchanging
solutely no access to electricity. PELS ideas. The various vice presidents of
EBL is a long-term commitment by the Society emphasized the impor­
our Society to drive a recurring global tance of volunteering for the various
competition to crowdsource inno­ activities such as conference orga­
vation and develop viable solutions nization and participation, member­
to energy access with radical scal­ ship of local Chapters, and getting
ing potential. involved in the multitude of standard
The PELS Cyber-Physical Secu­ committees—many of which are or­
rity Initiative was undertaken to call ganized remotely by Webex confer­
attention to the issues, challenges, ence calls.
 by Kai Sun

IFEC 2018 Competition Held Successfully

at Tsinghua University

O
n 19 and 20 July, the final
competition of the 2018 IEEE
International Future Energy
Challenge (IFEC 2018) was held at
Tsinghua University, Beijing, China
(Figure 1). This international com­
petition was initiated by the IEEE
and marks the highest level of stu­
dent competitions in power elec­
tronics. IFEC 2018 was sponsored by
the IEEE Power Electronics Society
(PELS), IEEE Industry Application
Society, IEEE Power & Energy Soci­
ety, and Power Supply Manufacturer
Association. This competition, started
in 2001 by the IEEE and the U.S.
Department of Energy, invites under­
graduate and graduate students from
around the world to work together on
challenging power electronic topics.
It aims to inspire creative thinking
and cultivate elite future profession­
als in the field of power electronics.
Participating teams are required to
propose an innovative solution to a
specific technical problem and build a
prototype to verify their design. A
grand prize and special awards are
given based on the reviews from aca­
demic and industrial experts for the
proposals and experimental tests for
the prototypes.
IFEC 2018 was hosted by the De­
partment of Electrical Engineering,
Tsinghua University, with the topic
of high-efficiency and high-density
isolated bidirectional dc–dc con­
verter for residential energy stor­
age systems.
This event received support from
the PELS Beijing Chapter, the China
Digital Object Identifier 10.1109/MPEL.2018.2874757
Date of publication: 19 December 2018
FIG 1 Participants at IFEC 2018. Ten teams from six countries participated in this
competition. (Photo courtesy of Tsinghua University.)
Electrotechnical society, the Tsinghua
University Energy Internet Research
Institute, and the State Key Labora­
tory of Control and Simulation of
Power System and Generation Eq­
uipment. Industry sponsors for the
competition included Diamond part­
ner Chroma (Shenzhen) Corporation
and Golden partner Mitsubishi Elec­
tric Corporation. The test platform
for the final competition was provid­
ed by Chroma (Shenzhen) Corporation.
Prof. Zhengming Zhao from Tsinghua
University served as the general chair
of the competition. Dr. Kai Sun from
Tsinghua University and Dr. Qiang
Li from Virginia Tech, Blacksburg,
served as the general cochairs. Prof.
Yaow-Ming Chen from National Tai­
wan Uni­­versity, Taipei City, served as
the steering committee chair. The judg­
ing panel for the final competition con­
sisted of Prof. Philip Krein from UIUC
(member of the National Academy
of Engineering and an IEEE Fellow),
Prof. Jih-sheng Lai from Virginia Tech
(founding chair of IFEC and an IEEE
Fellow), Prof. Braham Ferreira from
Delft University of Technology, The
Netherlands (past president of PELS
and IEEE Fellow), Prof. Toshihisa
Shimizu from Tokyo Metropolitan
University (IEEE Fellow), Dr. Alpha
Zhang from the Delta Group, and Dr.
Dachong Gu from Unique Technical
Services, LLC. Ten student teams from
six countries and regions participated
in the final competition.
After the exciting two-day competi­
tion in oral presentation and hardware
testing, student teams from Xi’an
Jiaotong University, Tsinghua Uni­
versity, and Beijing Jiaotong University
won the Grand Prize Award, Outstand­
ing Performance Award, and Innova­
tion Award with US$10,000, US$5,000,
and US$3,000 rewards, res­­pectively.
The U­ niversity of Belgrade, National
Taiwan University of Science and
Technology, and The Ohio State Uni­
versity recei­v ed the Educat i o n a l
Impact Award, Outstanding Tech­
nical Report Award, and Outstanding
Presentation Award, respectively.

December 2018 z IEEE Power Electronics Magazine 71


International Workshop Fosters
Energy Harvesting Ecosystem
T
o capitalize on the
growing momen­
tum and adoption
of e n e r g y h a r v e s t i n g
(EH), the E n e r g y Har­
vesting Committee of the
P o w e r S o u r c e s Manu­
facturers A s s o c i a t i o n
(P S M A ), in partnership
w i t h the I E E E P o w e r
Electronics Society, the
European Center for
Power Electronics, the
CONNECT Science Foun­
dation Ireland Research FIG 1 Most of the EnerHarv 2018 delegation. (Photo courtesy of Tyndall National Institute.)
Center, and the China Pow­­
er Supply Society, launched the first
International Energy Harvesting
Workshop in Cork, Ireland, on 29–31
May 2018, called EnerHarv 2018 at
the Tyndall National Institute. The
media partner was How2Power.
Roughly 81 attendees from around
the world participated in this work­
shop (Figure 1). The prog ra m was
divided fairly evenly among lecture
sessions, functional demonstrations,
and interactive panel discussions
with plenty of time reserved for net­
working and team-building prospects
as well as a poster session as a bonus.
There were 20 lecture presentations
(including two keynotes and three
panel introductions), 16 functional
demonstrations, and 19 posters. The
program opened with two keynotes,
the first from Yogesh Ramadass from
Texas Instruments’ Kilby Labs, with
an overview of EH technologies and
applications from the past to where
we are headed in the future. This
was followed by Eric Yeatman from
Imperial College London, with a fo­
cus on wireless sensor networks and
Digital Object Identifier 10.1109/MPEL.2018.2874758
Date of publication: 19 December 2018
72 IEEE Power Electronics Magazine z December 2018
outlining how their deployment, en­
abled by EH, opens the doors to many
of today’s applications.
Then followed a prolific lineup
of speakers for technical, applica­
tion, and panel sessions from Ana­
log Devices (Massachusetts), Carn­
egie Mellon University (Pittsburgh,
Pennsylva nia), ST Micro (Ita ly),
Ilika (United Kingdom), ARM (Uni­­
ted Kingdom), UNIST ­(Korea), Cam­­
bridge ­Display Technologies (Unit­­ed which they were previously unaware
Kingdom), Fraunhofer, IMTEK-Uni­
versity of Freiburg (Germany), Tyn­
dall (Ireland), MCCI (Ireland), North
Carolina State University, Boston
Scientific (Massachusetts), Cap-XX
(Australia), the University of South­
ampton (United Kingdom), and Unit­­
ed Technologies (Connecticut). The
speakers were divided among major
topic areas of transducers, low-power
loads, storage, power management,
and system integration. The three
panel sessions drew engagement from
the full delegation and generated
lively conversations around applica­
tions, energy optimization, and fu­
ture directions.
Many great things came out of this cells, even within a common lot, that
workshop; some were expected, and
by Brian Zahnstecher and Mike Hayes
some were surprises. Given the em­
phasis that was put on developing syn­
ergies across the siloed contributors
of this industry, it was great to see
the excitement and willingness of
the participants to learn about each
other’s a rea of ex per tise a nd or­
ganically seek collaboration where
a p p r o p r i a t e . Many ­p a r t i c i p a n t s
were surprised to learn of key con­
tr ibutions to t he EH eco s y s t em,
of, that they did not rea lize were
relevant, let alone important, addi­
tions to and enablers of the In­­ternet
of Things (IoT). One example of this
was a presentation given by Brandon
Lucia of Carnegie Mellon University,
in which he described the need for
software code that has the intelli­
gence to use an intermittent execution
model that can handle unexpected
power interruptions and can sim­
ply pick up seamlessly from where
it was operating when power was
lost. Alex Weddell of the collabora­
tion between ARM and the University
of Southampton gave some perspec­
tive on the variation of photovoltaic
can drastically change the design
 calculations required to reliably de­
liver an integrated IoT device.
One of the most important enablers
for perpetuating the EH ecosystem
is demonstrating how many of these
technologies are viable today for
real-world applications, particularly
for the IoT. This is why the inclu­
sion of functional demonstrations
(29–30  May) was a mandatory and
complementary aspect of the work­
shop program. A poster session al­
ways creates structure and stimulates
discussions and networking, and this
also went extremely well. All of the
demonstration and poster sessions
were a nice mixture of contributions
from both industry and academia. FIG 2 As this experimental event ended, it was important to gauge participant feed-
back and determine follow-up action with a series of poll questions. Luckily, the
While this diverse group of contrib­ responses were overwhelmingly positive! (Photo courtesy of Tyndall National Institute.)
utors and audience provided a wide
area of breadth and depth of coverage,
there is still much more to consider. If implementer perspective. This meant a critical mass of people in the power­
one truly wants to internalize the tech­ an obvious bias by those (particu­ ing IoT ecosystem. To this end, it was
nological and market requirements for larly in Europe) providing EH ma­ a major success and a solid foundation
enabling EH-powered applications, terials and components as opposed for future activities. Taking advantage
then it is important to transcend the to those on the application side. of this momentum, the committee de­
typical coverage areas (e.g., transduc­ For i n st a nce, if cided to push to the
ers, power management, storage, and this event was held next level and offi­
loads) to consider things like soft­ in Silicon Valley (San One of the most cially recognize this
ware, test and measurement, user in­ Francisco Bay area), important enablers emerging technol­
teraction, and industrial design. It is then it would be far ogy area for what it
for perpetuating the
also important to recognize that many more likely to include is, a true ecosystem
applications will not have an ambient more contributions EH ecosystem is (Figure 2). Most de­­
EH source, but nonetheless, the micro­ from the implementer/ demonstrating how legates arrived at
power management technologies be­ application perspec­ many of these EnerHarv 2018 with
ing developed offer the opportunity to tive and, therefore, technologies are viable a passion for EH and
significantly extend the battery life for may have a more ap­­ today for real-world a motivation to figure
IoT devices, thereby reducing mainte­ prehensive, cautious, out how to identify
applications,
nance costs and improving reliability and risk-averse ap­ and work with part­
and functionality. proach to implement­ particularly for ners to perpetuate it.
Postevent surveys were distrib­ ing EH technologies. the IoT. Those same people
uted both at the workshop and in One of the most fre­­ walked away with
electronic form following the event quent feedback com­ knowledge, collab­
with responses received from over ments was not in regard to whether orative relationships, and the realiza­
40% of the attendees. Feedback was there should be another EnerHarv but tion that they were witness to and
quite posit ive, but this must also a question of when and that, for ex­ members of a proper EH ecosystem.
be interpreted in the proper context. ample, a biennial event is not frequent If you are interested in learning
EnerHarv 2018 was designed primar­ enough to keep up with the pace of this more about EnerHarv, the PSMA EH
ily for the contri­butors of the EH eco­ emerging technology space or the de­ Committee, or anything else refer­
system to develop synergies and learn mand of the ecosystem for knowledge enced in this article, please feel free
from one another, which also means and insight. to contact Brian Zahnstecher (bz@
that there was a significant deficit This effort was an experiment and powerrox.com) or Mike Hayes (michael
of representatives from the user or an effort to build, network, and steer .hayes@tyndall.ie).

December 2018 z IEEE Power Electronics Magazine 73

 by Bruno Lequesne and Uthane Supatti

IEEE ITEC Asia-Pacific 2018

Focuses on E-Mobility

T
he 2018 IEEE International
Transportation Electrifica­
tion Conference and Expo
Asia-Pacific (ITEC AP) was held in
Bangkok, Thailand, 6–9 June 2018.
The focus of the conference was
on transportation electrification
across the world. ITEC AP was co­­
sponsored by the IEEE Industry
Applications Society (IAS), IEEE
Power E l e c t r o n i c s Society, and
IEEE Power & E n e r g y S o c i e t y
(PES) in cooperation with the
Electric Vehicle Association of
Thailand (EVAT). Te c h n i c a l co­­­­
sponsors of this event included the
IEEE Joint IAS/Industrial Electron­
ics/Power Electronics T h a i l a n d
Chapter, the IEEE PES T h a i l a n d
FIG 1 The panel discussion session “How Does Electric Charging Infrastructure and Its
Chapter, and the Electrical Engi­ Interoperability Unlock the Barriers to EV Adoption?” The panelists included (from
n e e r i n g / E l e c t r o n i c s , Computer, left) Christian Wiedmann, president of BMW Group Thailand; Dr. Nuwong Chollacoop,
Telecommunications and Informa­ committee member, EVAT; Carolin Capone, head of sustainable transport projects in
tion Technology Association. It was Thailand and with ASEAN-GIZ; and Keith Hardy, director of the EV-Smart Grid Interop-
erability Center, Argonne National Laboratory. (Photo courtesy of EVAT.)
held in conjunction with EVAT’s
third International Electric Vehicle
Technology Conference and Exhi­ vehicle chargers to cars and elec­ Hori (Japan), Prof. Volker Pick­
bition or (iEVTech 2018) and the tric tuk-tuks. Test drives of some of ert (United Kingdom), Colin Mck­
Association of Southeast A s i a n the vehicles were e r r a c h e r (United
Nations Sustainable Energy Week. a l s o i n c l u d e d in Kingdom), and Dr.
This cooperation enabled the orga­ the event. Overall, Tichakorn Wongpi­
Overall, ITEC AP had
nizers to offer a comprehensive and ITEC AP had 90 par­­­ r o m s a r n (S i n g a ­
varied program to an international t i ­c ipants for the 90 participants for the p o r e ) speaking on
audience, with a particular focus main track with main track with the future of trans­
on the Asia-Pacific region. Held at 50  technical p r e ­ 50 technical portation, f r o m
the Bangkok International Trade sentations from presentations from p o w e r t r a i n s to
& E x h i b i t i o n Center, the joint 12  countries, and 12 countries, and autonomous driv­
event w e l c o m e d a s i g n i f i c a n t s o m e 200 p e o p l e ing and infrastruc­
some 200 people for
exhibit of electric vehicles, with for the tutorials. ture (Figure 1).
some 35 ex­­hibitors presenting a The expo itself saw the tutorials. P repa r a t ion s a re
variety of products, from electric over 25,000 visitors under way for the
from 40 countries. next ITEC AP, to be
Digital Object Identifier 10.1109/MPEL.2018.2874760
The conference also featured key­ held on Jeju Island, South Korea,
Date of publication: 19 December 2018 note speeches from Prof. Yoichi 8–10 May 2019.

74 IEEE Power Electronics Magazine z December 2018

 by K. Biju

An IEEE Distinguished Lecture at the

Indian Institute of Science and Technology

A
n IEEE Distinguished Lecture
was held on 14 May 2018 at
the Indian Institute of Sci­
ence and Technology (IIST), Trivan­
drum. Dr. Akshay Kumar Ra­­thore,
associate professor, Concordia Uni­
versity, Montréal, Canada, delivered
the lecture “Impulse Commutated
Frequency Modulated Soft-Switching
Current-Fed Converters” (Figure 1).
The IEEE Industry Applications
Society (IAS), IEEE Industrial Elec­
tronics Society (IES), and IEEE Power
Electronics Society (PELS) Kerala
Joint Chapter Secretary Prof. K. Biju
and IEEE IAS/IES/PELS Kerala Joint
FIG 1 Dr. Akshay Kumar Rathore delivering the Distinguished Lecture at the IIST,
Chapter Conference Chair Dr. Jai­ Trivandrum, Kerala. (Photo courtesy of the IIST.)
son Mathew were present during the
event. The IIST IEEE Student Branch
counselor and head of the Avionics De­ Low-voltage, high-current applica­ clamping of switches through a high-
partment, IIST, Dr. B.S. Manoj, talk­ tions was one of the major topics of frequency resonant tank. It is a simple,
ed about the student branch activities. the Distinguished cost-effective, and
Dr. Sudarshan Kaarthik, IAS Student Lecture. It covered easy way of solving
Branch Chapter advisor and assis­ the various anoma­ the traditional turn-
The program also
tant professor in the Avionics Depart­ lies due to the volt­ off voltage spike
ment, IIST, chaired the program. Dr. a ge -fed re son a nt
included an interactive across the device.
Rathore started the Distinguished converter and also session after the talk, A n i nt er a c t ive
Lecture by giving a brief introduc­ discussed the impor­ during which the session with the au­
t ion about dc-dc converters. The tance of the current- participants asked dience followed the
program also included an interactive fed converters. questions regarding Distinguished Lec­
session after the talk, during which The presentation ture. Thirty-two par­
the topic covered in
the participants asked questions re­ started with a discus­ ticipants, including
garding the topic covered in the dis­ sion of low-voltage, the discussion. students and prof­ ­
cussion. Dr. P.P. Rajeevan, associate high-current appli­­­ es­s o r s , a t t e n d e d
professor in the Avionics Department, cations. The various the program. They
IIST, gave a welcome speech. applications were discussed along discussed the various challenges
with a description of impulse com­ with soft-switching, voltage-fed con­
mutated current-fed converters. This verters. Participants also had the
unidirectional class of current-fed opportunity to vist the Power Elec­
Digital Object Identifier 10.1109/MPEL.2018.2874774
converters attains soft-switching, ze­ tronics Laboratory after the Dis­
Date of publication: 19 December 2018 ro-current commutation and voltage tinguished Lecture.

December 2018 z IEEE Power Electronics Magazine 75


IEEE Power Electronics Society
A
fter several years of inactivity,
the IEEE Power Electronics
Society (PELS) Swiss Chap­
ter was reactivated in May 2018 and is
currently ramping up its networking
activities. Prof. Drazen Dujic, head of
the Power Electronics Laboratory at
Ecole Polytechnique Fédérale de Lau­
sanne (EPFL), currently serves as the
chair of the Chapter (Figure 1).
The first technical lectures of the
PELS Swiss Chapter were organized at
two Swiss Federal Institutes of Tech­
nologies: EPFL and Eidgenössische
Technische Hochschule, Zürich. Prof.
Johann Kolar hosted Prof. ­Chunting
Mi from San Diego University, La
Digital Object Identifier 10.1109/MPEL.2018.2874775
Date of publication: 19 December 2018
PELS Chapter Is a Prime Mover
C
apturing an o s c i l l o s c o p e
screenshot once a converter
reaches full power operation
Digital Object Identifier 10.1109/MPEL.2018.2874776
Date of publication: 19 December 2018
76 IEEE Power Electronics Magazine z December 2018
Swiss Chapter Reactivated
FIG 1 IEEE PELS Swiss Chapter Chair
Prof. Drazen Dujic.
by Thomas Foulkes, Andrew Stillwell, Nathaniel Renner,
The University of Illinois
Outside of the Laboratory
is definitely a memorable moment
during graduate school, but the
IEEE Power & Energy Society, IEEE
Power Electronics Society (PELS),
and IEEE Industry Applications
Society Student Branch Chapter at
by Drazen Dujic
Jolla, California, who gave the lecture
“High-Efficiency Wireless Charging
of Electric Vehicles.” In September
at EPFL, Prof. Drazen Dujic hosted a
lecture by Dr. Ki-Bum Park from ABB
Corporate Research Center in Swit­
zerland, “Minimization of Grid-Tied
Power Electronics Converters.” Both
lectures were well attended and pro­
vided opportunities for exchange
of technical ideas among partici­
pants. Materials from these lectures
are available on the Chapter’s website:
www.ieee.ch/chapters/pel/.
In the coming months, the Swiss
Chapter will continue with organization
of technical and educational seminars
and workshops, providing a platform
for networking and exchange among
IEEE Members.
Enver Candan, and Joyce Mast
the University of Illinois at Urbana–
Champaign (UIUC) works diligently
to help members also make memo­
ries outside of the laboratory. Dur­
ing the 2018 spring semester, the
Chapter hosted board game nights,
 (a) (b)

(c) (d) (e) (f)

FIG 1 UIUC PELS members build lasting friendships outside the laboratory through social activities like (a) bowling and
(b) the Illinois Marathon as well as (c)–(f) STEM outreach events such as the Girl Scouts of Central Illinois Girl Power activity
day. Funding to build these STEM exhibits was received in part from the Grainger Center for Electric Machinery and
­Electromechanics as well as the Power Optimization of Electro-Thermal Systems National Science Foundation Engineering
Research Center.

cookouts, a bowling tournament [Fig­ each step of the power generation pro­ power different types of light bulbs
ure  1(a)], and fielded two teams who cess works. After the tour, participants [Figure 1(f)]. The fourth annual Girl
successfully completed the 2018 Illi­ explored other aspects Power e v e n t w i l l
nois Marathon [Figure 1(b)]. of power and energy take place in April
Chapter members also helped en­ with demonstrations During the 2018 2019. If you or some­
sure the success of a science, tech­ desig ned by UIUC one you know in the
spring semester, the
nology, engineering, and mathematics PELS Chapter mem­ PE L S c om mu n it y
(STEM) outreach event with the Girl bers. Popular exhib­ Chapter hosted board is interested in par­
Scouts of Central Illinois, aptly named its this year included game nights, cookouts, ticipating in the next
Girl Power, which was organized by Dr. a shock i ng lesson a bowling tournament, Girl Power event,
Ashlynn Stillwell, assistant professor about charge from and fielded two teams contact eit her the
of civil and environmental engineering an electrostatic di­­­ UIUC PELS Chapter
who successfully
and an affiliate in electrical and com­ rod [Figure 1(d)], mix­ or Dr. Stillwell (ash
completed the 2018
puter engineering at UIUC. At the start ing h e a t t r a n s fe r lynn@illinois.edu).
of the day, each Girl Scout r­ eceiv­­ed w it h a r t b y u s i n g Illinois Marathon. All in all, h o s t i n g
one of the special hard hats shown in he a t pipes to make both social and ser­
Figure 1(c) and toured the Abbott Power ice scul­­ptures [Fig­ vice events has been
Plant (an 85-MW combined heat and ure  1(e)], and gaining an apprecia­ the key to growing both memories and
power facility) at UIUC to learn how tion for how much work is required to membership of the local Chapter.

December 2018 z IEEE Power Electronics Magazine 77

 by Laili Wang

First IEEE Workshop on

Wide-Bandgap Power Devices
and Applications in Asia

T
he IEEE Workshop on Wide- of PELS, was the honorary chair. Laili wide-bandgap power electronics.
Bandgap Power Devices and Wang (Xi’an Jiaotong University) There were four sessions in the techni­
Applications in Asia (WiPDA served as the technical program chair. cal program, which included pack­
Asia) took place on 17–19 May 2018 More than ten of the world’s top ex­ aging and gate drive, high-efficiency
in Xi’an, China. It was the first WiPDA perts were invited to give keynote converters, silicon carbide and gallium
held in Asia, and it aimed to provide speeches at WiPDA nitride characteriza­
a forum for device scientists, circuit Asia, including eight tion, and applications
designers, and applications engi­ IEEE Fellows such WiPDA Asia contained of power electronics.
neers to share technology updates, as Fred C. Lee, an aca­ keynote speeches, Related to the afore­
re­­search findings, development expe­ demician of the U.S. mentioned themes,
tutorials, a Young
riences, and application knowledge. National Academy of 170 papers were sub­
WiPDA Asia was sponsored by Engineering; Frede Professionals forum, mitted, and 112 pa­
the IEEE Power Electronics Soci­ Blaabjerg, president- plus technical and pers were accepted
ety (PELS). It was financially sup­ elect/vice president poster sessions. after review.
ported by Platinum sponsor Infineon, of products at PELS; More than 550 par­­
Gold sponsor TYSiC, and many Sil­ Yan-Fei Liu, vice pre­­ ticipants attended
ver sponsors. sident of technical operations at PELS; the conference, including 160 in­
Yue Hao (XiDian University) and and Dehong Xu, who is the director of dustry professionals, 150 professors
Jinjun Liu (Xi’an Jiaotong University) the China Power Society. from u n iver sit ies, a nd over 30 re­
served as the general chairs of the WiPDA Asia contained keynote searchers from well-known research
conference. Alan Mantooth, president speeches, tutorials, a Young Profes­ institutions (e.g., University of No­­
sionals forum, plus technical and t­­tingham, University of Cambridge,
Digital Object Identifier 10.1109/MPEL.2018.2874791
poster sessions. This workshop cov­ University of Arkansas, and Virginia
Date of publication: 19 December 2018 ered the most important aspects of Tech) (Figure 1).

FIG 1 There were more than 550 WiPDA Asia participants in 2018.

78 IEEE Power Electronics Magazine z December 2018

 by Mahmoud Saleh
City College of New York
IEEE Student Branch Chapter Chair
Receives Best Paper Award
M
ahmoud Saleh, chair of
the IEEE Power & Energy
Society, IEEE Industrial
Applications Society, and IEEE
Power Electronics Society City Col­
lege of New York (CCNY) Student
Branch Chapter, is the recipient of
the IEEE International Conference
on Renewable Energy Research and
Applications Best Student Paper
Award (Figure 1). This prestigious
event was held in San Diego, Califor­
nia, in December 2017. He received
the award for his paper “Hardware-
Based Testing of Communication-
Digital Object Identifier 10.1109/MPEL.2018.2874777
Date of publication: 19 December 2018
by Slavko Mocevic, Rebecca Rye, and Joseph Kozak
Virginia Tech IEEE Student Branch
Chapter Organizes Company and
Laboratory Tours
R
ecently, the I E E E P o w e r
Electronics Society (PELS)
Student Chapter at Virginia
Tech, Blacksburg, organized a series
of company and laboratory tours for
Digital Object Identifier 10.1109/MPEL.2018.2874778
Date of publication: 19 December 2018
FIG 1 Mahmoud Saleh. (Photo courtesy
of Mahmoud Saleh.)
Based Control for DC Microgrid” and
for his exceptional work regarding
smart grids and microgrids control
and automation. IEEE Chapter Mem­
more than 40 undergraduate and
graduate students, with the purpose
of interacting with industry mem­
bers and researchers in the power
electronics field.
The Chapter toured ABB’s trans­
former factory in Bland, Virginia
(Figure 1). On this tour, students were
December 2018 z IEEE Power Electronics Magazine 79
ber Yusef Esa coauthored this exem­
plary work.
Saleh currently teaches circuits
at Bronx Community College to help
students and share his expertise. He
is also introducing students to IEEE
activities and Societies so that they
can interact with professionals in the
field. In 2014, Saleh was enrolled in the
graduate program at the Smart Grid
Laboratory, Department of Electrical
Engineering, CCNY. He was awarded
several prizes and scholarships while
pursuing his Ph.D. degree. He is con­
sidered an expert in the field of smart
grids and microgrids, and he has
published and coauthored more than
14 technical papers.
guided throughout the factory to see
the entire process of obtaining, as­
sembling, testing, and packaging dif­
ferent types of transformers such as
laminated iron core, which are first
encased by a low-voltage winding and
then a high-voltage winding. Further­
more, ways of protecting transformer

Fig 1 Virginia Tech students at the ABB transformer factory in Bland, Virginia. (Photo
courtesy of Muhammad Altaf.)
core and windings from damage due
to vibrations, foreign matter, and high
temperature were explained. The tour
turned out to be extremely educational
and presented a valuable experience
for all attendees to learn the funda­
mentals and manufacturing of power-
system transformers.
The next tour organized by t he
Chapter wa s at a loca l compa ny
called VPT, Inc. in Blacksburg. Dan
Federal University of Campina
T
he Federal University of Campi­
na Grande (UFCG) IEEE Power
Electronics Society (PELS)
and IEEE Industry Applications Soci­
ety Student Branch Chapter (SBC)
organized several activities in 2018,
some of which were mentioned in this
Digital Object Identifier 10.1109/MPEL.2018.2874779
Date of publication: 19 December 2018
80 IEEE Power Electronics Magazine z December 2018
Sable, chief executive officer and
Virginia Tech professor, illustrated
the history of the company to the 20
student participants and led the tour.
During this visit, students learned
about state-of-the-art, reliable power
electronics solutions for space, mili­
tary, and industry applications and
had an opportunity to interact with
one of the best engineers and re­
searchers in power electronics.
by Phelipe L.S. Rodrigues and Montiê A. Vitorino
Grande IEEE Student Branch
Chapter Hosts Events
column in the September 2018 issue of
IEEE Power Electronics Magazine.
Continuing its activities in the second
semester of 2018, the SBC developed
lectures and two workshops.
The third edition of the Photovoltaic
Systems Workshop took place on 18
June 2018. This edition was a partner­
ship with the IEEE Women in Engineer­
ing Affinity Group in the UFCG, the
university in which the workshop took
F ina lly, the Chapter orga nized
a tour for Virginia Tech students at
the Center for Power Electronics
Systems (CPES), Blacksburg. The
tour introduced research themes in
power electronics spanning across
low-voltage power supplies, novel
control methodologies, and high-volt­
age semiconductor devices. The tour
was organized during the dialog ses­
sion of the CPES conference, and at­
tendees network­­e d with graduate
students, professors, and industry
leaders, where they learn­­­ed about
current problems facing power elec­
tronics industries.
In the past two years, the PELS
Student Chapter at Virginia Tech has
successfully hosted several other
events including hands-on workshops
focusing on magnetic components
and wireless power transfer. In ad­
dition, the Chapter hosted Distin­
guished Lecture seminars (Dr. Pat
Wheeler and Dr. Huang-Jen Chiu), in­­
dustry information sessions, and PELS
webinar showings.
place and was presented to the girls
of Itan Pereira College. On 20 June
2018, Prof. Osvaldo Ronald Saavedra
Mendez, from the Federal University
of Maranhão (UFMA), gave the lec­
ture “Ocean Energies.” The lecture’s
objective was to present the projects
that are carried out by Prof. Saavedra
Mendez and his research group at
UFMA involving ocean energies. He
presented an overview on the use
 Prof. Osvaldo Mendez (holding the IEEE PELS flag) with the UFCG professors and students after his lecture “Ocean Energies.”
of this type of renewable energy in
the world.
On 26 July 2018, Prof. Marcelo Lobo
Heldwein, from the Federal University
of Santa Catarina–Brazil, gave a talk
on static converter systems for un­
derwater electrification in oil and gas
exploration. The SBC also developed
and started a new activity: the Meth­
ods, Efficiency, Time, Agility, and
Solutions (METAS) Workshop.
by Gerald Christopher Raj Irudayaraj
IEEE PELS PSNA College of Engineering
and Technology Student
Chapter Workshops
O
n 7 September 2018, the
IEEE Power Electronics
S o c i e t y (PELS) S t u d e n t
Digital Object Identifier 10.1109/MPEL.2018.2877519
Date of publication: 19 December 2018
The METAS Workshop presents
tools that can be used by students
of electrical engineering through­
out their academic and professional
lives. These words summarize the
purpose of the METAS Workshop,
which is to use methods and solu­
tions so that students can manage
their time in a certain activity with
efficiency and agility. On 30 August
2018, the first topic addressed by
Chapter of PSNA College of Engi­
neering and Technology (CET), Din­
digul, Tamilnadu, India, hosted a one-
day national workshop “Hands-On
Training on Low-Cost Solar Lamps”
for more than 70 undergraduate stu­
December 2018 z IEEE Power Electronics Magazine 81
Chapter President Phelipe Leal Se­
rafim Rodrigues was about the use
of the Prezi tool to create presenta­
tions. The workshop was a great suc­
cess and had 26 attendees.
IEEE Hour was held in partnership
with the UFCG SB on 13 September
2018. The event lasted for one hour,
during which the SB, SBCs, and af­
finity groups of the UFCG presented
their developed activities.
dents (Figure 1). The event was orga­
nized in association with the IEEE
Special Interest Group on Humanitar­
ian Technology (SIGHT) and Liter of
Lights, Bengaluru, India. Dr. Rajasek­
aran Vairamani, head of the Electrical
FIG 1 The IEEE PELS student members with event coordinators at the end of a training session on low-cost solar lamps. (Photo cour-
tesy of PELS Student Chapter of PSNA CET.)

and Electronics Engineering Depart­ funded the materials along with the tant professor in the Electrical and
ment, delivered the inaugural address. IEEE SIGHT. Electronics Department. In the after­
Over the course of the day, students noon session, the fabrication of solar
heard a lecture by Dr. Sithambaram Foundation Course on cell modules was demonstrated. On
Muthukumaran, IEEE Student Branch Solar Photovoltaic Design the second day, students received
counselor and professor in the Elec­ The Chapter also organized a founda­ practical training in charging con­
trical and Electronics Department, tional course on solar photovoltaic troller units, battery charging, and
followed by a demonstration on the design from 20 to 21 July 2018. Dr. mounting the structure in various
use of reusable materials for cost R a j , an a s s o c i a t e environments from
reduction and reduction of green­ professor in the Elec­ Dr. Mark Arul Pra­­
house emissions by PELS Chapter trical and Electron­ Approximately 60 fully sanna, a s s o c i a t e
faculty advisor Dr. Gerald Christopher ics D e p a r t m e n t , assembled lights are professor, and Arun
Raj Irudayaraj. organized this two- ready for installation, Prasad, a s s i s t a n t
The students built and tested their day course, w h i c h professor in the
with plans to erect
own lights by assembling the solar was a t t e n d e d by Electrical and Elec­
panel, controller, battery, and light- more than 100 under­ them in remote tronics Department.
emitting diode lamp pipes and test­ graduate electrical areas where access The s t u d e n t s
ing their operation. Students gained engineering students to electricity is not gain­­e d theoretical
exposure to the latest technology from various insti­ available. knowledge and pra­­
and the use of reusable materials. tutes. Dr. Vairamani c­­tical experien­­ce from
Approximately 60 fully assembled delivered the wel­ this two-day work­­
lights are ready for installation, come address. The students learned shop. This course encouraged stu­
with plans to erect them in remote about photovoltaic systems and solar- dents to design and build their own
areas where access to ­e lect r icit y cell module development during the photovoltaic systems and undertake
i s not available. Pankaj Dixit, co ­ first day’s morning session lectures by new projects.
founder of Liter of Lights, Bengaluru, Dr. Raj and Dr. Soundar Rajan, assis­ 

82 IEEE Power Electronics Magazine z December 2018

Introducing IEEE Collabratec™
The premier networking and collaboration site for technology
professionals around the world.

IEEE Collabratec is a new, integrated online community where IEEE members,

Network.
researchers, authors, and technology professionals with similar fields of interest
can network and collaborate, as well as create and manage content. Collaborate.
Featuring a suite of powerful online networking and collaboration tools, Create.
IEEE Collabratec allows you to connect according to geographic location,
technical interests, or career pursuits.

You can also create and share a professional identity that showcases key
accomplishments and participate in groups focused around mutual interests,
actively learning from and contributing to knowledgeable communities.
All in one place!

Learn about IEEE Collabratec at

ieee-collabratec.ieee.org
 In Memoriam
The Power Electronics Community
Loses an Accomplished Professional
T
uesday, 9 October 2018, was
a sad day for Dr. Milan M.
Jovanović’s family and the
power electronics community. On
this day, at the age of 66 years, Milan
passed away peacefully in Durham,
North Carolina, surrounded by fa­­
mily. However, he leaves behind a le­­
gacy of innovation, research, hard
work, and humor. For all those who
knew him, his special sense of humor
made life memorable.
Calling Milan a dear friend and
colleague, Delta Electronics founder
and honorary chair Bruce Cheng
recalled his first impression of Milan
when he met him as a research scien­
tist at Virginia Power Electronics
Center while working with Prof. Fred
Lee. “I was impressed with his work
and his deep knowledge and under­
standing of power electronics. Short­
ly after 1991, Milan joined us at Delta
to lead our advanced electronics
research and development at our
facility located in Research Triangle
Park, Raleigh, North Carolina,” stat­
ed Cheng. He added, “Everyone who
knew or met Milan quickly under­
stood his intelligence, his quick wit,
and his concern for people and fel­
low employees.”
Cheng said, “Milan’s legacy and
work will continue at Delta through
his colleagues here in Raleigh and at
our other research labs around the
world. Delta plans on honoring Milan
Digital Object Identifier 10.1109/MPEL.2018.2877473
Date of publication: 19 December 2018
84 IEEE Power Electronics Magazine z December 2018
Dr. Milan M. Jovanović, 1952–2018.
by naming the Raleigh Research Lab
in his honor.” Continuing to honor his
memory, he stated, “I will miss Milan’s
valued contributions and advice in
furthering the development of power
electronics for the benefit of Delta and
the world. Milan, you will always have
a dear place in my heart, my friend.”
Remembering Milan, Prof. Dushan
Boroyevich of Virginia Tech stated
“Amazing Milan, dragi naš Ćuća! What
a person, what an engineer, what a
friend! A lifetime of friendship.” Prof.
Boroyevich met Milan during his
freshman year while studying applied
physics in the College of Electrical
Engineering at the University of Bel­
grade, Yugoslavia. He added, “After
graduation, we started working togeth­
er as instructors of electronics in the
newly founded department for energy
and electronics at the University of
Novi Sad. We shared an office and our
master’s degree advisor, Prof. Vojin
Cvekić. There at the Department,
Milan also met his wife for life, Mirja­
na. We came to Blacksburg, Virginia,
because Prof. Fred Lee accepted us to
be his Ph.D. students at Virginia Tech.”
Prof. Boroyevich continued, “Milan
did not have a long life, but a life lived
to the fullest. Last time when we talk­
ed, he said that he accomplished all he
was hoping for. We are all honored
and proud to have been a part of such
a life, and our love and respect for him
will stay forever.”
Prof. Lee, who had the privilege of
having Milan as his student for five
years and then as a colleague and dear
friend for the remaining 30 years, rem­
inisced Milan as a smart, gentle, and
kind man. According to Prof. Lee,
“While he was a student, he was so
involved in his research that I had to
remind him several times that he has
done more than enough to earn a
Ph.D. degree. His reply was always ‘I
enjoy doing what I am doing and I am
still learning, why hurry?’ I have never
heard another student make a similar
remark ever since.” He further added,
“Milan graduated in 1988 and stayed
on for a couple more years as research
faculty before joining Delta in 1991.
“After that, our relationship took a dif­
ferent turn, Milan not only became a
closer friend but also became my
sponsor and boss, who directed my
research for the next 30 years. Milan
is much loved by so many people that
he will always be held in high esteem.
I am so fortunate and so proud to have
a friend like him. I miss him so dearly,”
Prof. Lee said.
For four decades, Jovanović inno­
vated efficient ways of deliver ing
power supplies to devices and sys­
tems, ranging from smartphones to
data centers, in smaller packages with
lighter weight and higher reliability.
As a senior vice president for research
and development at Delta Electronics,
one of the world’s largest manufactur­
ers of power supplies, he led research
teams in developing numerous design
techniques that have been implement­
ed in millions of power supplies,
enabling further miniaturization of
electronics equipment and resulting in
enormous cumulative energy savings
and waste reduction. All of this con­
tributed to making this world greener.
Jovanović believed in sharing his
tremendous knowledge and experi­
Are You Moving?
Update your contact information so you don’t miss
an issue of this magazine!
Change your address
E-MAIL: address-change@ieee.org
PHONE: +1 800 678 4333 in the United States
or +1 732 981 0060 outside
the United States
If you require additional assistance
regarding your IEEE mailings,
visit the IEEE Support Center
at supportcenter.ieee.org.
IEEE publication labels are printed six to eight weeks in advance
of the shipment date, so please allow sufficient time for your publications
to arrive at your new address.
ence. Consequently, he was very active
in the IEEE Power Electronics Society
(PELS) and continuously encouraged
his coworkers and all power electron­
ics engineers to grow professionally
through free and critical information
exchange. The following are just some
of his roles:
■■tutorial chair at the 2012 IEEE
Power Conversion and Motion Con­
trol Conference
■■secretary of the Meetings Commit­
tee of the IEEE PELS (1991–1993)
■■program cochair of the Power Elec­
tronics Specialists’ Conference
(1994)
■■member-at-large of the PELS Admin­
istrative Committee (1995–1997)
■■IEEE PELS representative of the
Applied Power Electronics Confer­
ence and Exposition (APEC) Steer­
ing Committee (1997–1998)
■■member of the IEEE APEC Pro­
gram Committee (1990–2004).
December 2018 z IEEE Power Electronics Magazine 85
In 1976, he received his Dipl.-Ing.
degree in electrical engineering from
the University of Belgrade, Yugoslavia.
From 1976 to 1983, he was on the fac­
ulty of the Institute for Power and
Electronics at the University of Novi
Sad, Yugoslavia. Milan came to the
United States in 1983 as a doctoral
student at the Virginia Polytechnic
Institute and State University (better
known as Virginia Tech), Blacksburg.
He received his Ph.D. degree in 1988.
His Ph.D. advisor at Virginia Tech
was Prof. Fred Lee, the fifth president
of PELS.
An IEEE Fellow since 2001, Milan
was inducted into the U.S. National
Academy of Engineering in 2015. He
authored more than 270 tech­n ical
articles and held more than 50  U.S.
patents. He is survived by his wife
Mirjana; daughter, Maria; and son,
Max. He will be truly missed.

IMAGE LICENSED BY INGRAM PUBLISHING
 Book Review
Electric Powertrain: Energy Systems,
Power Electronics and Drives for
Hybrid, Electric and Fuel Cell Vehicles
I
find Electric Powertrain: Energy
Systems, Power Electronics and
Drives for Hybrid, Electric and
Fuel Cell Vehicles to be timely and
relevant for automotive engineers
and engineering students seeking a
deeper appreciation of what vehicle
electrification means. The book was
years in preparation and reflects the
authors’ own wealth of industrial and
academic experience.
It has been almost three decades
since General Motors (GM) displayed
the Impact concept car at the Los
Angeles Auto Show in January 1990.
The Impact inspired the design of
the GM EV1, the first modern produc-
tion electric car, and for California to
subsequently issue its zero-emissions
mandate. The EV1 became available
to lease in 1996 and demonstrated to
the public that electric cars were fea-
sible. Automakers at the time recog-
nized that it would require very con-
siderable education for the electric
car to go mainstream, and in fact, it
still does.
The automobile is a complex ma-
chine that has undergone continuous
refinement ever since Henry Ford
rolled out the first Model T in 1908
a s a n a ffordable ma ss-produced
vehicle. By the 1990s, the motoring
public was accustomed to the auto-
mobile’s refined propulsion and highly
tun­­ed ride and handling. Replacing the
Digital Object Identifier 10.1109/MPEL.2018.2874780
Date of publication: 19 December 2018
86 IEEE Power Electronics Magazine z December 2018
public, will find this book a treasure
trove of knowledge on modern au-
tomotive technology.
The book is divided into four parts
in 16 chapters, starting with “Part I:
­Vehicles and Energy Sources.” This
section is actually a stand-alone,
concise description of what a more
highly electrified automobile is. The
reader is taken on a path that tra-
verses our present dependence on
fossil fuels and carbon-dioxide emis-
sions of the ICE into the world of fuel
cells and battery energy storage as
the electric fuel for the traction mo-
tor. What is most refreshing in all the
chapters of this book are the worked
examples, exercises that expand on
these, and computer simulations of
the more complex topics. The spe-
ci f ic exa mples a nd problems get
internal combustion engine (ICE) and further refined as the reader moves
gasoline tank with an electric traction into each new section of the book.
drive and a relatively massive battery The next section, “Part II: Electric
pack constituted a quantum jump in Machines,” is a painstakingly deep
engineering for manufacturers to just dive into the four main players in elec-
stay on par and for potential buyers tric traction motors: brushed dc, in-
to comprehend just what all-electric duction, surface permanent magnet,
means. What buyers do know is that, and interior permanent magnet types.
regardless of hybrid, fuel cell, or bat- Engineers familiar with electric ma-
tery electric, these vehicles must match chines will find these chapters an
or exceed their expectations. ideal refresher, students will find the
The authors, Dr. John G. Hayes and topics fascinating, and the informed
Dr. G. Abas Goodarzi, both veterans lay person will garner a deeper ap-
of the GM EV1 program, have focused preciation of how that stored electric
their considerable talent and experi- fuel gets converted into mechanical
ence on teaching the inner workings energy for vehicle propulsion.
of the electric car. Readers, whether “Part III: Power Electronics” is a
engineers, students, or the interested more esoteric topic as many readers
 wishing for a deeper understand-
ing of the electric car will view these
chapters as really getting under the
hood. This is the world of power con-
verters, inverters, and how they are
controlled. Most will recognize these
five chapters as a real strong suit of
the authors because of their abil-
ity to describe with such clarity the dc
transformer and ac synthesizer. The
reader is led by an expert hand through
switch-mode power conversion and is
shown the intricacies of the comple-
mentary behavior of active (transis-
tor) and passive (diode) switching de-
vices. Examples, again, reinforce their
knowledge of not only computing av-
erage and root mean square quanti-
ties of signals but of realistic power
dissipation and efficiency computa-
tions. All are extremely beneficial
to the practicing design engineer
and fundamental to understanding by
Digital Object Identifier 10.1109/MPEL.2018.2875656
the student and inquiring reader. My
favorite chapter was on battery charg-
ing because of the need to not only
understand the role of power factor
correction to meet utility grid power
quality regulations but also to mini-
mize ripple exposure of the battery.
Some readers may need to brush up on
their control theory as the examples
in Chapter 15 get into tuning the in-
ner current/torque loop and the outer
voltage/speed loop for the electric ve-
hicle drivetrain.
Finally, “Part IV: Basics” is another
stand-alone section, but this time it is
a single chapter on electromagnetism
and energy conversion topics that
are, in fact, fundamental. The sec-
tion includes topics that are basic to
understanding not only electric ma-
chines but the electrified automobile
in general. My recommendation is
that readers may find some topics in
December 2018 z IEEE Power Electronics Magazine 87
Parts II and III more understandable
by consulting Chapter 16 first, while
others may actually wish to read over
Chapter 16 following Part I.
My overall impression is that this
book was well worth the investment,
and one that I will proudly display
on my shelf. For those wishing to dig
even deeper, or for appreciation of
sources, the authors provide excellent
references after each chapter. My con-
clusion—what a great book!
About the Reviewer
John M. Miller (jmmiller35@aol.com)
is an industry consultant with more
than 42 years of experience in electri-
cal engineering across various indus-
tries. He is an IEEE Life Fellow, a
­fellow of the SAE, and a registered
Professional Engineer in Michigan
(1980) and in Texas (2014).

 Event Calendar

2019 31 May–2 June

Quanzhou, China
7–8 February
IEEE International Symposium on Predictive Control of
College Station, Texas, United States Electrical Drives and Power Electronics (PRECEDE)
IEEE Texas Power and Energy Conference (TPEC)
19–21 June
26 February–1 March Novi, Michigan, United States
Bhubaneswar, India IEEE Transportation Electrification Conference and Expo (ITEC)
IEEE International Conference on Sustainable
Energy Technologies and Systems (ICSETS) 25–29 June
Svalbard
17–21 March
IEEE Future of Electronic Power Processing
Anaheim, California, United States and Conversion (FEPPCON)
IEEE Applied Power Electronics Conference
and Exposition (APEC) 6–9 August
Washington, D.C., United States
25–29 March
IEEE Electric Ship Technologies Symposium (ESTS)
Sochi, Russia
International Conference on Industrial Engineering, 2–4 September
Applications, and Manufacturing (ICIEAM) Istanbul, Turkey
International Aegean Conference on Electrical Machines
24–26 April
and Power Electronics (ACEMP) and the 2019 International
Toulouse, France Conference on Optimization of Electrical and Electronic
IEEE International Workshop on Integrated Power Equipment (OPTIM)
Packaging (IWIPP)
2–5 September
11–15 May Genova, Italy
San Diego, California, United States IEEE 21st European Conference on Power Electronics and
IEEE International Electric Machines and Drives Applications (EPE’19 ECCE Europe)
Conference (IEMDC)
29 September–3 October
20–23 May Baltimore, Maryland, United States
Matsue, Shimane, Japan IEEE Energy Conversion Congress and Exposition (ECCE)
IEEE Third International Conference on DC Microgrids (ICDCM)
1–4 December
27–31 May Santos, Brazil
Busan, South Korea IEEE 15th Brazilian Power Electronics Conference and
Tenth International Conference on Power Electronics IEEE Fifth Southern Power Electronics Conference (COBEP/SPEC)
and the IEEE Energy Conversion Congress and Exposition, Asia
(ICPE 2019–ECCE Asia)

Digital Object Identifier 10.1109/MPEL.2018.2874793

Date of publication: 19 December 2018 

88 IEEE Power Electronics Magazine z December 2018

White Hot (continued from page 92)
Some adjustment in the external cir-
cuits had to be made so that the IGBTs
could be driven with a negative voltage
at turn-off, but these problems were
quickly solved.
Later there, we experimented with
resonant gate drives for MOSFETs. The
resonant gate drive circuit added an
inductor between the driver output and
the MOSFET gate. The idea was to res-
onate the inductor with the MOSFET
gate-source capacitance to quickly
charge that capacitance with minimal
loss. While the research papers showed
interesting results, this idea never
caught on in the market.
When considering how to improve
the gate drive of a MOSFET, one needs
to consider two different inductances.
One is the inductance between the driv-
er output and the MOSFET gate. This
inductance slows down the switching,
increasing loss, and can create noise
when it resonates with the MOSFET
gate capacitance.
The other inductance is the com-
mon source inductance (CSI or CSL).
Consider the MOSFET source current
as it leaves the die. This current typi-
cally travels through a wire bond to a
package lead to a trace on a printed cir-
cuit board. Each of these conductors
adds inductance. During turn-on and
turn-off, both the drain-to-source cur-
rent and the gate current flow through
We want
to hear
from you!
IMAGE LICENSED BY GRAPHIC STOCK
this inductance (and thus the name
common source inductance). During
turn-on, the voltage on the CSI must be
positive to ramp up the current as the
MOSFET turns on. This positive volt-
age subtracts from the voltage from the
gate drive output and slows down the
turn-on. During turn-off, the voltage
must be negative to ramp the current
down as the MOSFET turns off. This
negative voltage on the CSL (more neg-
ative at the MOSFET source than at the
connection to the circuit board) now
acts to apply a positive voltage to the
gate-source capacitance and slows
down the turn-off.
The effect of the CSI on the switch-
ing of silicon MOSFETs is a problem
but is not overwhelming. However,
when we consider the wide bandgap
devices like the gallium nitride (GaN)
high electron mobility transistors
(HEMTs) and silicon carbide (SiC) MOS-
FETs, the CLI becomes a critical issue.
Why? Because the switching speed of
these devices is an order of magnitude
(or more) faster than a silicon MOSFET
(and even faster compared to a silicon
IGBT). With this high rate of change of
the current (di/dt), the CSL can severely
degrade the performance of these
devices. The applications engineering
team at Efficient Power Conversion
(EPC) has published some very good
material on this problem.
Do you like what you’re reading?
Your feedback is important.
Let us know—send the editor-in-chief an e-mail!
December 2018 z IEEE Power Electronics Magazine 89
I will continue this discussion in
Part 2 of this article in the March
2019 issue of IEEE Power Electron-
ics Magazine. There are interesting
things that can be done with the GaN
HEMTs that I think will help speed
their adoption in the same way that
the ease of driving silicon MOSFETs
hastened the end of the era of the
bipolar power transistor.
About the Author
Robert V. White (bob.white@ieee
.org) has more than 30 years of indus-
try experience as a power electronics
engineer. He has worked in product
design, systems and applications
engineering, and technology develop-
ment. He has been an active volun-
teer with the IEEE Power Electronics
Society, serving several years on the
Administrative Committee, two terms
as technical vice president, and as a
Chapter chair. He earned a B.S.E.E.
degree from the Massachusetts Insti-
tute of Technology and an M.S.E.E.
degree from Worcester Polytechnic
Institute. He is currently pursuing a
Ph.D. degree in power electronics at
the University of Colorado, Boulder.
Presently, he is the chief engineer of
Embedded Power Labs, a power elec-
tronics consulting company. He is a
Fellow of the IEEE.

 CALL FOR PROPOSALS
POWER ELECTRONICS FOR DISTRIBUTED
GENERATION SYSTEMS

The IEEE Power Electronics Society (PELS) invites you to submit a proposal to host IEEE PEDG
2022 (Power Electronics for Distributed Generation Systems).

Do you want to be a host? Please contact Prof. Martin Ordonez (UBC) at mordonez@ieee.org for
more details. The deadline for proposal submissions is March 3rd, 2019.

VANCOUVER, CANADA - JUNE 27-30, 2016

Dubrovnik, Croatia

?
Charlotte, NC, USA

Chair:
Martin Ordonez
Chicago, USA
Steering Committee members:
Juan Carlos Balda Frede Blaabjerg
Vladimir Blasko Liuchen Chang
Deepak Divan Rik W. De Doncker
Johan Enslin Gerard Hurley
? Jinjun (XJTU) Liu
Sudip Mazumder
Denizar Martins
Mark Dehong Xu

Digital Object Identifier 10.1109/MPEL.2018.2877110

 Advertisers Index

The Advertisers Index contained in this issue is compiled as a service to our readers and advertisers: the publisher is not lia-
ble for errors or omissions, although every effort is made to ensure its accuracy. Be sure to let our advertisers know you
found them through IEEE Power Electronics Magazine.

sales contact Page Advertiser URL Phone

Walter Chalupa Associates
11 Acopian Technical Company www.acopian.com +1 800 523 9478
Walter Chalupa
National Sales Manager 6 AGILE Magnetics www.agilemagco.com +1 800 805 8991
Kerstin Chalupa
14 Applied Power Systems, Inc. www.appliedps.com +1 516 935 2230
International Sales Manager
Telephone: +1 973 835 7015 CVR 4 Chroma Systems Solutions www.chromausa.com +1 949 600 6400
chalupapels@aol.com
3 Coilcraft www.coilcraft.com/AGP

23 CKE/Dean Technology www.deantechnology.com

10 ICE Components, Inc. www.icecomponents.com

22, 65 ITG Electronics www.ITG-Electronics.com

62 Kendeil S.r.l. www.kendeil.com +39 0331 786966

CVR 2 Magna Power Electronics www.magna-power.com

16 Magnetic Metals Corp. www.MagneticMetals.com +1 888 892 8392

65 Mersen ep.mersen.com

69 Micrometals Alloy Powder Core Division www.micrometalsarnoldpowercores.com

5 Nagamori Foundation www.nagamori-f.org/en +81 75 935 7731

12 Omicron Lab www.omicron-lab.com

21 Payton Planar Magnetics www.PaytonGroup.com +1 954 428 3326

67 PEM Ltd. www.pemuk.com +44 0 115 946 9657

CVR 3 Plexim GmbH www.plexim.com

7 Power Integrations www.power.com

9 Powersim, Inc. www.powersimtech.com/pels2018 +1 301 841 7445

19 PREEN AC Power Corp. www.acpower.net +1 949 988 7799

67 SIBA GmbH www.siba.de +49 23 06 70 01 0

17 Teledyne Lecroy teledynelecroy.com/static-dynamic-complete

15 Triad Magnetics www.triadmagnetics.com +1 951 277 0757

16 WEMS Electronics www.wems.com +1 310 644 0251

Digital Object Identifier 10.1109/MPEL.2017.2770684

December 2018 z IEEE Power Electronics Magazine 91

 White Hot
Evolution of Gate Drivers—Part 1
W
hen I graduated in 1980, I
started in the small sys-
tems power supply group
at the Digital Equipment Corporation
(DEC). My first assignment was to
design a 208 W three output power
supply for DEC’s first desktop com-
puter. The topology was a two-transis-
tor forward converter that used bipo-
lar switching transistors (BUW45s, if I
recall correctly). The bipolar power
transistors of that time were rugged
devices but had a fairly low current
gain (beta) of 4–10. This meant that if
you were switching 5 A, one had to
supply about 1 A in base current.
One popular way of doing that was
a proportional base drive. A propor-
tional base drive used a current trans-
former to couple a fixed fraction of the
emitter current to the base. In my
design, the base drive transformer had
one turn in the emitter and four turns
driving the base of each BUW45. This
drove the transistors with a forced
beta of 4. The BUW45s has a typical
current gain of 5–10, so this assured
that the BUW45s would always be sat-
urated during the on-time.
Actually, while being well saturat-
ed was good for conduction loss, it
was not good for switching time and
switching loss. As a bipolar device,
there was a lot of charge stored in
the base-collector junction when the
transistor was saturated. To turn the
transistor off, this charge had to be
removed through a large negative
base current, which caused a signifi-
Digital Object Identifier 10.1109/MPEL.2018.2873993
Date of publication: 19 December 2018
92 IEEE Power Electronics Magazine z December 2018
cant turn-off delay. To minimize the
stored charge and turn-off delay, a
Baker clamp was used. This Baker
clamp used two diodes in series with
the base to assure that the collector-
emitter voltage was always greater
than the base-emitter voltage and
that the transistor never fully saturat-
ed. This traded off additional conduc-
tion loss for reduced turn-off delay
and reduced turn-off loss.
The proportional base drive trans-
former was great at keeping a bipolar
transistor conducting, but it needed
additional circuitry to turn the transis-
tor on and off. At DEC, we used a base
drive circuit invented by Jim Gregorich,
the senior power supply designer at the
company. The Gregorich base drive
peak charged a capacitor from a sec-
ondary winding. To turn the main tran-
sistor off this capacitor, an auxiliary
switching transistor was turned on to
apply a large negative voltage and cur-
rent to the transistor being turned off.
After the capacitor discharged, a cur-
rent was taken from an output through
a resistor to keep this winding shorted.
To turn the main transistor on, the aux-
iliary switching transistor was turned
off. The base drive transformer now
acted as a flyback, and the current that
was flowing in the auxiliary switch now
became a positive current into the base
of the main transistor, turning it on.
By using the Gregorich base drive
circuit and the Baker clamp, I was
able to efficiently switch the BUW45
transistors at the outrageously high
frequency of 50 kHz.
I tell this long tale to contrast with
the simplicity of driving silicon MOS-
by Robert V. White
FETs. Although the first true power
MOSFETs were commercially avail-
able in the 1977–1978 t i me f r a me
(I used a MOSFET in the design of a
dc–dc converter for the GE Power
Systems Business unit as a summer
engineer in 1979), widespread adop-
tion did not take place until about
1983–1984. Once the power supply
design community started using MOS-
FETs, the MOSFETs very quickly dis-
placed bipolar transistors in commer-
cial power supplies.
One of the key reasons was the sim-
plicity of the gate drive of a MOSFET
versus the complexity of the base drive
of a bipolar transistor. The MOSFET,
although a charge controlled device, is
driven from a voltage source. No longer
was there a need to supply amperes of
base current to keep a bipolar transis-
tor on. Once the gate-source capaci-
tance of the MOSFET was charged, no
additional current was needed.
For topologies with a single switch
referenced to the common return, like
the boost, single-transistor flyback,
or single-transistor forward convert-
ers, the MOSF ET could be d r iven
directly from the control IC. For con-
verters such as the two-transistor for-
ward or the bridge converters, the
MOSFETs could be driven with a sim-
ple gate drive transformer that only
needed to supply pulses of current for
turn-on and turn-off.
Also in the mid-1980s, the IGBT
came along. IGBTs are MOS gated
devices like the MOSFET, so the save
driver IC technology could be reused.
(continued on page 89)
 electrical engineering soft ware

FROM OFFLINE SIMULATION

TO REAL-TIME TESTING

PLECS SOFTWARE RT BOX PLATFORM

Fast simulation of complex systems Hardware-in-the-loop (HIL)

Comprehensive component library Rapid control prototyping
Analysis tools Full integration with PLECS
Standalone and Simulink versions Deploy large models on multiple boxes
Model-based code generation Competitively priced

n at
Get more informatio
w w w.plexim.com
 Simulate AC Line Conditions
Measure Critical Product Characteristics

NEW

15Hz-
2KHz

Model 61509 - 6kVA

New Additions to the 61500 Series

Compact Single- or Three-Phase
Programmable AC Power Sources
Chroma’s new compact programmable AC sources are programmable to 350VL-N
and deliver single- or three-phase output without derating under SCIP program
control. Frequency includes DC and 2kHz with 5kHz option. Models generate pure
sine waveform output with distortions levels <0.3% at 50/60Hz.
3kVA | 4.5kVA | 6kVA
Voltage range: 0-175V/0-350V/Auto
Frequency: DC, 15Hz-2KHz (5KHz opt.)
Single- or three-phase output selectable
Distortion Waveform, Harmonics, and Interharmonics

To learn more about power conversion testing, visit chromausa.com

Testing on-grid devices? Ask about our Regenerative Grid Simulators

Full 4 quadrant, regenerative, grid simulation with advanced features
for compliance, safety and product verification testing - up to 300kVA

chromausa.com | 949-600-6400 | sales@chromausa.com

© 2018 Chroma Systems Solutions, Inc. All rights reserved.