issn: 1863-5598 zkz 64717 11-06 bodo´sbodo´spower ... and simulation ... rosu, ph.d., group leader...

Bodo´s Power SystemsBodo´s Power SystemsSystems Design Motion and Conversion November 2006

ZKZ 6471711-06

ISSN: 1863-5598

Bodo´s Power SystemsBodo´s Power SystemsViewpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2

Mobility, Automotive and Milestones

Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2

News . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-8

Product of the Month . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10

Point of Load Reference Design for Digital PowerApplications

Guest Editorial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12

Advancements in MOSFET Technology for High Switching Frequency

Applications

By Tom Loder, Vice President of Sales & Marketing, MicrosemiPower Products Group

Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14

The Lennox Report; By Robert Lennox

Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16-17

DC-DC Converter Module Market Slows;

By Jeremiah P. Bryant, Managing Research Analyst, Darnell Group

VIP Interview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

Interview on Power Semiconductor Technology with Izak Bencuya,

VP GM Fairchild; By Bodo Arlt, Editor BPSD

Cover Story . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19-22

Sensorless Motor Control Algorithm;

By Aengus Murray, International Rectifier

Design and Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24-27

Selection, Power Loss Calculation and Temperature Rise Analysis;

By Prasad Bhalerao and Eugen Stumpf, Mitsubishi Electric Europe,Germany

MOSFET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28-31

A Fast Body Diode UniFET

By Sampat Shekhawat, Praveen Shenoy, Mark Rinehimer and BobBrockway, Fairchild Semiconductor, Mountaintop, USA

Power Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32-34

Best Layout Practices for Switching Power Supplies;

By L. Haachitaba Mweene, Applications Manager, National Semiconductor

DC/DC Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .36-38

Eliminating the Inductor in DC/DC Regulators;

By Ralf Muenster, Director of Marketing, Power Products, Micrel, Inc.

Power Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40-43

Large Form Factor Video Displays;

By Michael Day, Power Management Application Supervisor forPortable Power, Texas Instruments

Design and Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .44-49

Efficient Virtual Prototyping of Electric Drive Systems; ByMariusRosu, Ph.D., Group Leader Simplorer Modeling, AnsoftKoichi Shigematsu, Ph.D., Application Engineer, Ansoft and ThomasLiratsch, Country Manager, Germany, Ansoft

Transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50-51

Optimising the Power Switch in High Voltage Applications; By PeterBlair, Discrete Product Development Manager, Zetex Semiconductors

Motion Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .52-55

Current Control for the Switched Reluctance Machine with Enhanced

Performance

By Joanna Bekiesch and Günter Schröder, University of Siegen

High Power Switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .56-59

The Internally Commutated Thyrisor (ICT)

By Peter Köllensperger and Rik W. De Doncker; Institute for PowerElectronics and Electrical Drives RWTH AachenUniversity

New Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .60-64

Advertising Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64

C O N T E N T S

1www.bodospower.com

It happens always today, little boys learn firstto say “Auto” instead of Mama and Papa –the fascination is to drive around and to beindependent. Now the buzzword is “Hybrid”.What does this mean for European automo-biles, already highly developed with relative-ly small engines?

I have been in contact, over the years, withUniversities and manufactures involved inthe design of Hybrid vehicles. During mytime in engineering and applications I havedriven various electric powered vehicles.About 15 years ago, I tried several electriccars at the University in Aachen. The thing Iremember most was the large space occu-pied by battery packs.But the accelerationprovided by an electric motor alwaysimpressed me.

So electric vehicles would be great if theycould be powered from overheads like thetrolley buses, or get energy at the stops bycapacitor storage or inductive charging.Good solutions for busses maybe - but notfor individual private cars. Private cars needa good battery to have a reasonable rangeto drive and return home again. The USPresident once had an electric car - it is ondisplay in the Peterson Auto museum onWiltshire Blvd in LA. Also an old hybrid truckwas on show at the London TransportMuseum. Both systems were not successfulin the market, perhaps as they lackedtoday’s inverter drive technology. A verynice development was the hybrid Smart Carfrom Swatch and Daimler that never made itto production – very compact with 4-wheeldrive that ran extreme well - I rememberdriving it around in Biel. It was 10 yearsahead of its time. The combustion motorremains the most popular engine for today’sautos.

Today, switches using MOS technologymake it possible to build efficient inverterand energy storage systems. It does notmatter at this point where the energy comesfrom – a battery, a super capacitor, or a fuelcell can be the power source. What isimportant is that the car should have anacceptable level of comfort and space.These goals were missing in the early days.There were no MOSFET or IGBT switchesfor high performance drives, but combined

with modern driver and control circuits, wenow have the elements to make improvedefficiency happen – and inevitable increasesin fuel prices will continue the pressure fordevelopment.

Some Japanese carmakers are producingnice hybrid cars, acceptable in the market.Last month, I met Eric Lidow, who showedme he still looks forward to innovations - hegot himself the Toyota Highlander Hybrid. Inmy August issue, Intersil had it’s articleabout powering backlight LED arrays, whileEpcos showed how custom made capacitorsfit into hybrid automotive designs. In theSeptember issue, articles from Danfossabout “Shower Power” liquid cooled mod-ules, and from Maxwell on the Ultra capaci-tor, were focused on hybrid automotive appli-cations. A highlight in this issue is the articlefrom Ansoft, which encompasses the wholespectrum of application and design tasks.

I am looking forward to see what the H2expoand Electronica will show us for innovationsin Automotive. And I look forward to seeingyou in Munich or later on in November at theNuremberg SPS.

Best Regards

Mobility, Automotiveand Milestones

Bodo´s Power SystemsBodo´s Power SystemsA Media

Katzbek 17aD-24235 Laboe, GermanyPhone: +49 4343 42 17 90Fax: +49 4343 42 17 [email protected]

Publishing EditorBodo Arlt, [email protected]

Creative Direction & ProductionRepro Studio [email protected]

Free Subscription to qualified readers

Bodo´s Power Systems magazine is available for the following subscription charges:Annual charge (12 issues) is 150 €world wideSingle issue is 18 €[email protected]

Printing by: Central-Druck Trost GmbH & CoHeusenstamm, Germany

A Media and Bodos Power magazineassume and hereby disclaim any liability to any person for any loss ordamage by errors or omissions in thematerial contained herein regardless ofwhether such errors result from negligence accident or any other causewhatsoever.

Events

ELECTRONICA 2006

Nov. 14 – 17, Munichwww.electronica.de

SPS/IPC/DRIVES 2006

Nov. 28 – 30, Nurembergwww.mesago.de

ED Bloom Design, Dec. 4, Portsmouth UK,

www.ejbloom.com

APEC 2007

Feb. 25-March 1, Anaheim CAwww.apec-conf.com

EMV 2007

March 6-8, Stuttgart, www.e-emv.com

ELECTRONICA China 2007

March 21-23, Shanghai, www.global-electronics.net

PCIM China 2007

March 21-23, Shanghaiwww.pcimchina.com

V I E W P O I N T

2 www.bodospower.com

Already part of your vision.LEM.

www.lem.com At the heart of power electronics.

Whatever you invent, imagine or develop, LEM’s transducers are at the heart of your power electronics applications from the very start. LEM’s products, R&D, and people provide knowledge intensive solutions to keep up with your changing industry, allowing your visions to come to life.

N E W S


To furtherstrengthen itsDemandCreation busi-ness model inthe SouthernEuropeanregion, AvnetMemec hasacquired thesemiconductor

specialist distribution company ESCOItaliana, based in Monza, Italy. 34-year-oldESCO Italiana is one of the most successfuldistributors in Italy, with an annual salesturnover of approximately 30 Million Euro.

ESCO employs about 60 people and oper-ates from 5 offices across Italy. In addition, ithas distribution activities in Greece andTurkey.Highly dedicated to premier technical sup-port and Demand Creation, ESCO has beenable to develop excellent supplier relation-ships, among them US Analog, Mixed-Signaland Microcontroller manufacturerMaxim/Dallas. Maxim/Dallas, who is distrib-uted by Avnet Memec in other regions ofSouthern Europe (France, Spain andPortugal), is a major contributor to ESCO’srevenue stream in Italy, Greece and Turkeyand is set to enhance Avnet Memec’s futurevalue proposition in the analog sector.

Upon closing of the acquisition last week,Steve Haynes, President of Avnet Memec,said: “ESCO Italiana will be an excellentenhancement for our company. Apart froman ideally matching linecard, we foundESCO to have an extremely successful busi-ness model, well regarded by customersand suppliers alike. I expect this to be amatch of winning spirits between the ESCOteam and our Avnet Memec team inSouthern Europe, resulting in a fast return ofour investment.”

www.escoitaliana.it

www.avnet-memec.eu

Avnet Memec to acquire ESCO

Power Integrations (OTC: POWI), the leaderin high-voltage analogue integrated circuitsfor power conversion, today announced thatit has won a verdict in its patent-infringementlawsuit against Fairchild Semiconductor. Ajury found that Fairchild has willfully infringedall four Power Integrations patents assertedin the case, and has awarded PowerIntegrations damages of approximately $34million. A second trial, scheduled to begin onDecember 4, will address Fairchild’s chal-

lenges to the validity of the infringed PowerIntegrations patents. The patents are pre-sumed to be valid. “Power Integrations respects the intellectualproperty of others, and we expect our com-petitors to do the same,” said BaluBalakrishnan, president and CEO of PowerIntegrations. “We are the leader in our mar-ket thanks in large part to the intellectualproperty that we have worked hard to devel-op over nearly two decades. This is our third

successful effortto protect thisintellectual prop-erty againstunlawful infringe-ment by ourcompetitors, andwe will continuemaking everyeffort to protect itgoing forward.”

Jury Finds in Favour of Power Integrations

LinearTechnologyCorporation, aleading supplierof high-perform-ance analogintegrated cir-cuits, celebrated25 years sinceits founding in1981. Linear isa Silicon Valley

success story, growing beyond $1 billion inrevenue, with innovative products and indus-try-leading financial performance. Linear hasset the standard for product quality and on-time delivery. Its products continue to lead intechnical innovation and are designed intoproducts throughout the world.Robert Swanson, founder and ExecutiveChairman stated, “When Linear was foundedin 1981, to succeed it had to overcome a lotof what was then conventional wisdom.Investors didn’t think there was still much of

an opportunity to back a new chip venture,much less an analog chip company duringthe dawn of the ‘Digital Revolution.’ Thevision of the founding team turned out to becorrect. 25 years later we can say with pride,we did it our way and the results speak forthemselves.”

www.linear.com

25 Years Linear Technology

have achieved critical milestones in theadvancement of their joint design programaimed at accelerating innovation in the auto-motive industry. Since announcing their ini-tiative seven months ago, the two compa-nies have staffed joint design facilities,designed a next-generation microcontrollercore, defined product roadmaps, and alignedprocess technologies.

The two companies have opened jointdesign centers to provide broad access toglobal design talent in silicon, software, andautomotive applications. The two companiesanticipate reaching a headcount of 120 engi-neers by the end of the year.As part of their far-reaching collaboration,Freescale and ST have standardized onPower Architecture technology as the

instruction set architecture for jointly devel-oped microcontroller (MCU) products. Thetwo companies also are focusing productdevelopment efforts on a wide range of auto-motive applications, including powertrain,chassis, motor control, and body systems.

www.st.com

STMicro and Freescale


N E W S

IntellonCorporation, theleading globalprovider of inte-grated circuits(ICs) for power-line communica-tions,announced thatRick E. Furtneyhas been

named President and Chief OperatingOfficer of the company. Furtney will overseesales, marketing, business development,engineering, research and development andoperations for the rapidly growing semicon-

ductor company. He will report to CharlieHarris, who will continue his role asChairman and Chief Executive Officer.

Furtney has 23 years of experience in thesemiconductor industry with a history of driv-ing customer-focused market strategy andrapid, profitable growth. He joins Intellonfrom International Rectifier, where he wasVice President, High Reliability Products.Earlier, he served as Vice President andGeneral Manager of Analog Products atIntersil Corporation. Furtney holds a BS inElectrical Engineering from the University ofIllinois and a MBA in Management fromFlorida Institute of Technology.

“Intellon is one of the most exciting compa-nies in the semiconductor space,” saidFurtney. “With more than 12 million power-line ICs shipped to date and sales growingat a rapid pace, Intellon has establisheditself as the global leader in powerline tech-nology, IC sales and product enablement. Iam delighted to be joining the talentedIntellon team as the company ramps volumeproduction of the world’s first powerline ICbased on the HomePlug AV standard.”

www.intellon.com

Rick Furtney President and COO

Powersemannounces theformation of an100% EOU(Export OrientedUnit) inBangalore, India.The main pro-duction there willbe Bridge

Rectifier Modules and Thyristor Modules.Powersem GmbH, Schwabach Germany willproduce the same products as well. By the end of year 2008 Powersem will

start_ Eco-Top production in India, which willinclude both IGBT and MOSFET Technology._Eco-Pac Production ( with DIODEs,THYRISTORs, MOSFETs and IGBTs) willremain at the headquarters in Schwabach,Germany.Mr. Ashok Chadda, CEO Powersem, isconvinced that _ cost-effective production inIndia will help _ Powersem to compete glob-ally in the market, especially with the newcompetitors from countries like Taiwan,Korea and China.Mr. Chadda. states: ”The main goal forPowersem is to supply Power Modules from

India with the same quality as the productsmade in Schwabach Germany”.Powersem emphasises user-friendly solu-tions supporting customers success on theirend-applications. With success in India, Powersem also plansto expand its Production Unit in Schwabach,Germany. By mid - 2007 Powersem, will build a newfacility for its quality control departementnext to the production facility in SchwabachGermany -built in the year 2000.

www.powersem.de

Powersem in Bangalore, India

With the kick-off event at the NurembergAirport Conference Centre held on 12October 2006, the Bavarian PowerElectronics Cluster presented its organisa-tion, objectives and programme. Togetherwith sensors it is one of the 19 Clustersselected and financially supported by theBavarian “Staatsregierung”.The management of the Bavarian PowerElectronics Cluster is done by the Europeannetwork ECPE European Center for PowerElectronics with Thomas Harder acting ascluster manager.In the kick-off meeting with about 100 partici-

pants from industry, academia and politics,the planned work programme of the clusterhas been presented and discussed with thepower electronics community. Expectationsfrom industry and university were presentedin statements given by Dr. H. Heilbronner(Semikron), Prof. Dr. B. Piepenbreier(Friedrich-Alexander-University Erlangen-Nuremberg) and Dr. R. Schmidt (chamber ofcommerce and industry). Dr. H. Heilbronnermentioned that our “Global WarmingIndustry” has to change into a “GlobalCooling Industry” that meansindustry has to make serious

efforts to strengthen energy efficiency.Another highlight of the kick-off event wasthe presentation of the research competencein power electronics in Bavaria. Ten insti-tutes from university, university of appliedsciences and research organisations gavean overview of their research activities in thefield of power electronics. The followingtable-top exhibition with posters and demon-strators provided an opportunity for detaileddiscussion with the experts.

Power Electronics Cluster in Bavaria

www.cluster-bayern-leistungselektronik.de

Stackable.Scalable. Flexible.

High Performance. Analog.Texas Instruments.

For datasheet, evaluation module

and samples visit:

www.ti.com/tps40140-e

The TPS40140 turns power supplies in data center and telecommunication equipment into fully

scalable, stackable power systems with greater load-handling capability and maximum efficiency.

This unique PWM buck controller offers the simplicity of a stand-alone dual or two-phase controller

with the ability to “stack” multiple devices together, creating a high-density power supply.

Generating from 10 A to 320 A of output current, true interleaved operation enables maximum

efficiency up to 16 phases.

DC/DC Controller Boosts Efficiency

TPS40140

Vdd1

Vdd2

CH1

CH2

PHASESELECT

MASTERCLOCK

COMP

StackableUp to 16Phases

Independent or InterleavedOutputs

Technology for Innovators and the red/black banner are trademarks of Texas Instruments. 1638A1 © 2006 TI

N E W S


PARNEXUS wasformed by ateam of enter-prising entrepre-neurs with yearsof associationwith Indo-German relations

in trade and industry. The team understandsboth the European and Indian mindset.Suparna Majumdar (Director) has vast expe-rience having worked both in Europe andIndia. PARNEXUS is not your regular event

and marketing company going about thebusiness of making and keeping regular tar-gets and deadlines. India today is the desti-nation for most European nations. With asteadily rising GDP ( Gross DomesticProduct ), increasing world-class infrastruc-ture, a vast pool of well-qualified and skilledlabour, India has become the hub for mostindustries. There is opportunity for almosteveryone to establish and market their prod-uct, specially a new and innovative one.PARNEXUS already represents variousGerman companies in India to create a mar-

ket and offers assistance from marketresearch, data base maintenance, advertis-ing, travel and accommodation, organizingclient meetings, conferences, trade fairs par-ticipation and maintaining the publicity andpromotional activities all over India. For moreinformation please contact Ms. Majumdar.

[email protected]

European Companies Enter Indian Market

During the week of December 4 this year, e/jBLOOM associates Inc. will be sponsoring a2-part 4-day course in both structured ana-log design methods and modern power mag-netics design. This will be a ONE-TIME pub-lic presentation featuring both Dr. R.D.Middlebrook and Ed Bloom, and will takeplace in Portsmouth, England in the UK.Because I am sure you want to know moredetails about this course, I have prepared a

PDF file containing 4 pages of informationincluding registration contacts, instructorsbackgrounds, cost, and the hotel location inPortsmouth for attendees and for the coursepresentations.I would like once again to thank you and allof our customers and friends for their contin-ued support of our services since 1981.Because of your valued support, we are nowcelebrating our 25th business year, provid-

ing educational products and services topower electronics industries all over theworld.So, don't miss out on this unique educationalopportunity.....plan now to attend.To download the file go:

http://www.ejbloom.com/EJB 25THANNIVERSARY COURSE.pdf

25th Anniversary Course in the UK

Silicon Laboratories a leader in high-per-formance, analog-intensive, mixed-signal ICscelebrates ten years of mixed-signal innova-tion. By fundamentally changing semicon-ductor architectures, Silicon Laboratorieshas achieved multiple industry firsts in thecommunications, wireless, networking,power and microcontroller markets. SiliconLaboratories has built a portfolio of marketleading products leveraging mixed-signal

trade secrets, a world-class engineeringteam and more than 700 issued or patentpending innovations. Silicon Laboratories was founded by NavSooch, Dave Welland and Jeff Scott in 1996in Austin, Texas to develop world-classmixed-signal ICs. The company’s initial prod-uct was an analog modem for personal com-puters (PCs). Ten years later, the companyis now a global enterprise with operations,

sales and design activities worldwide. Thecompany became profitable just two yearsafter its inception and completed a success-ful initial public offering (IPO) in 2000.Today, Silicon Laboratories is a publicly trad-ed, approximately $500 million company,leading mixed-signal innovation across abroad set of products.

www.silabs.com

Ten Years of Innovation

Microsemi Corporation, a leading manufac-turer of high performance analog/mixed sig-nal integrated circuits and high reliabilitysemiconductors, has announced thatCongress has appropriated $1.8 million toallow Microsemi’s Power Products Group(formerly Bend, Oregon-based AdvancedPower Technology) to develop technologyrelated to the use of silicon carbide semicon-ductor components in military avionics appli-cations. It is expected that the program willbe administered by AFRL (Air Force

Research Laboratory)This commitment by the Air Force andCongress to further development of the newsilicon carbide technology supports futuredesigns of lighter and more efficient jet fight-er communications systems, and will enablesubstantial growth of Microsemi’s operationsin Bend.The appropriation comes on the heels of acontract with Northrop Grumman earlier inthe year wherein Microsemi will provideleading edge silicon carbide products to this

leading defense contractor.News of the Congressional funding came onFriday in an announcement in Washingtonby Oregon U.S. Senators Ron Wyden andGordon Smith. According to Senator Wyden,“This commitment reaffirms the status ofCentral Oregon as a haven for high technol-ogy research, development, and production,particularly as it relates to aerospace appli-cations.”

www.microsemi.com

$1.8m R&D Contract on Silicon Carbide

P R O D U C T O F T H E M O N T H


Win one of three POL design kits

Point of Load Reference Designfor Digital Power Applications

Complete solution eases digital power design

Silicon Laboratories, a leader in high-per-formance, analog-intensive, mixed-signalICs, is offering reader’s of Bodo’s power sys-tems the chance to win one of three single-phase point of load (POL) reference designkits. The kits, valued at $149, enable engi-neers to easily and quickly implement digital-ly-controlled POLs for many end applicationsincluding servers, telecom, datacomm andstorage systems, medical equipment andaeronautical systems.

This reference design is a 20 A POL suitablefor packaging as a stand-alone power supplymodule or can be implemented directly onthe end application circuit board by theOEM. Included in the POL reference designis a pre-configured software kernel that fea-tures active dynamic dead time control formaximum operating efficiency, non-linearcontrol response for fast transient response,and SMBus port capable of supportingindustry-standard communication protocols.

The reference design kit is shipped with acomplete development tool suite enablingusers to modify the POL application code asdesired. This tool suite consists of a power-

ful GUI-based application builder that initial-izes the switch timing, loop compensation fil-ter and processor set-up without writingapplication software. Also included is anIntegrated Development Environment (IDE)that contains an editor, macro assembler,demo C compiler and a special online

debugger which allows for manualinspection and adjustment of sys-tem parameters during powersupply operation.

The POL reference design kitincludes full schematics and lay-out (Gerber) files that facilitatefast time to market and greatlyreduce design time and effort.The total area occupied by thePOL is only 615 mm2 and deliv-ers a maximum of 20 A for a com-plete 20 A/100 W power convert-er. The POL reference design uti-lizes a 4-layer PCB and is avail-able with all necessary connec-tors and switches for completeevaluation of the product. Alsoincluded in the POL is a USBdebug adapter, USB to SMBus

communications adapter, USB cable and 2 Ùload resistor. The only external hardwarerequired to evaluate the kit is an externalinput power supply.

The POL reference design is based on theSi8252 digital power controller in a 32-pinQFP package.

How to win? Fill out the free subscription form by fax oron the web for Bodo’s Power Systems.Any of the present subscribers already onmy list just mark it reconfirmation SILABS.All responses until 22nd of November will gointo the final drawing .December issue will have the winnersannounced.

www.bodospower.com

www.silabs.com

Intersil – An industry leader in Switching Regulators and Amplifiers.©2006 Intersil Americas Inc. All rights reserved. The following are trademarks or services marks owned by Intersil Corporation or one of its subsidiaries, and may be registered in the USA and/or other countries: Intersil (and design) and i (and design).

Intersil has a long-standing reputation as a trusted supplier of high-reliability ICs for military, space and specialty markets. This background suits us ideally for the mandatory quality processes in the automotive market. Intersil plans to become TS 16949 compliant by the end of 2006.

Intersil understands the challenges faced in infotainment and driver assistance applications. We have developed cost-effective, robust ICs that can be applied in standard configurations to meet electrostatic discharge requirements at connector pins.

Intersil display solutions support high-quality video in all shapes, sizes and environments. Intersil’s innovative light-to-digital sensors can simplify automatic lighting systems throughout the entire vehicle.

Intersil is committed to bringing its extensive analog expertise and portfolio to the following automotive applications:

Video DistributionDisplay Power ManagementHeads Up DisplaysCluster DisplaysDriver AssistanceBody ElectronicsLED LightingLight-to-Digital SensorsPressure, Acceleration andRain Sensors

High Performance Analog

Intersil Automotive Solutions

Win A Video iPodTM

at Electronica

Name

Job Title

Company

Email Address

Complete and return this form to Intersil at Electronica booth A4.207 to register for a chance to win!

Power metal-oxide semiconductor field-effecttransistors (MOSFETs) are known for superi-or switching speed, and low gate drivepower. In these respects, power MOSFETsapproach the characteristics of an “idealswitch.” Because of these features, powerMOSFETs are the best choice for high volt-age, high power, high performance switchmode power supply applications includingpower factor correction, server and telecompower systems, solar inverters, arc welding,plasma cutting, battery chargers, medical,semiconductor capital equipment and induc-tion heating.

In N-channel MOSFETs, only electrons flowduring forward conduction – there are nominority carriers. Switching speed is deter-mined by how fast the internal capacitancescan be charged and discharged. MOSFETswith low capacitances are therefore capableof very high switching speeds and corre-spondingly low switching losses. This is whyMOSFETs are the device of choice in highswitching frequency, switch mode powersupplies and inverters operating in excess of100 KHz.

The drawback of fast switching is that theMOSFET can interact with parasitic circuitelements, causing oscillations and genera-tion of excessive electromagnetic interfer-ence (EMI). This effect can cause problemsranging from increased switching losses tocatastrophic system failures. Finding theproper balance between high switchingspeed and oscillation immunity is the keydesign tradeoff in the latest generation ofMOSFET technology.

Companies like Microsemi, with its recentacquisition of Advanced Power Technology,

continue to develop new power managementtechnology, assembly and manufacturingtechniques to address the needs of high-voltage devices and applications.

The acquisition combined two strong highperformance analog companies offering bothdifferentiated RF product into niche end mar-kets as well as a strong focus on the high-power, high-speed segment of the powersemiconductor market.

The latest MOSFET technology advance-ment by Microsemi has led to the introduc-tion of devices with low switching losses andimproved oscillation immunity. The keydevice parameters are the value and ratio ofinput and reverse transfer (Miller) capaci-tances as well as the intrinsic gate resist-ance of the MOSFET. This optimizationresults in the elimination of gate oscillation,

reduced EMI (“quiet” switching), simple gatedrive circuitry and increased drain-gate noiseimmunity, resulting in high dv/dt ruggedness.

As with previous generations of products,Microsemi continues to focus on high volt-age (>500V) devices that are capable ofhandling high power at high switching fre-quencies. This family of devices also featurelow on-resistance, which results in low con-duction losses. Combined with low switchinglosses the result is an overall efficiencyimprovement and increased system powerdensities. Advanced manufacturing process-es have lowered thermal resistance andenabled higher current ratings for each diesize and package type compared to earliergenerations.

These advancements also apply to FRED-FET devices. All MOSFETs have an intrinsicbody diode that allows reverse current flow.This body diode has a slow reverse recov-ery. Slow body diode recovery leads to relia-bility problems in zero-voltage switching(ZVS) circuits. A FREDFET is a MOSFETwith additional processing to speed up thebody diode recovery. A FREDFET has identi-cal switching characteristics to a MOSFET,with the fundamental difference being theimproved reverse recovery speed of thebody diode. The key benefit of FREDFETs isimproved reliability in ZVS bridge circuitsand higher commutation dv/dt ruggedness.

Companies like Microsemi are paving theway for the next generation of high reliability,high efficiency, power conversion systems atever decreasing costs.

www.microsemi.com

G U E S T E D I T O R I A L


Advancements in MOSFET Technology

for High Switching FrequencyApplications

By Tom Loder, Vice President of Sales & Marketing, Microsemi Power Products Group

Intersil – An industry leader in Switching Regulators and Amplifiers.©2006 Intersil Americas Inc. All rights reserved. The following are trademarks or services marks owned by Intersil Corporation or one of its subsidiaries, and may be registered in the USA and/or other countries: Intersil (and design) and i (and design).

Intersil offers a wide selection of power management and analog signal processing

solutions for the consumer, industrial, communication and computing markets. You won’t

believe what our innovation, combined with your design imagination, can achieve.

Intersil is a global technology leader

specializing in the design and

manufacture of high performance analog

semiconductors.

Intersil continues to innovate, developing

superior analog signal processing

devices such as digital light sensors and

composite video drivers available in ultra

small chipscale packages. In power

management, Intersil’s recognized

leadership in computing switching

regulation is only a fraction of the story.

Intersil has developed a diverse portfolio

of PWMs and battery management

devices for everything from handhelds,

displays, and medical/industrial products.

Industry Leadership Positions:

Amplifiers

Switching Regulators

Interface Products

Switches/MUXes

Video Crosspoints

Voltage References

Digitally Controlled

Potentiometers

Real Time Clocks

Hot Swap Controllers

High Performance Analog

Intersil Analog Solutions

Win A Video iPodTM

at Electronica

Name

Job Title

Company

Email Address

Complete and return this form to

Intersil at Electronica booth A4.207

to register for a chance to win!

GENERAL

The recent automobile exposition in Parisreminds us once again of major changes inthe car and truck industries in the directionof more alliances on a worldwide basisdeeply affecting the electronics industry as amajor supplier. This adds to the potentialimpact of Chinese automobile exports.

SEMICONDUCTORS

According to the WSTS global semiconduc-tor sales in August grew 10.5% over prioryear and 2.1% sequentially to $ 20.5 B, arecord high and a year-to-date growth of8.7%. The Americas led again with an 18.3%increase over prior year followed by AsiaPacific 10.5%, Japan 8.8% and Europe 4.4%but only 1.3% sequentially and a year-to-date negative 0.3%. In Euros Europe was up0.1% over prior year and 1.1% sequentially,+2% year-to-date. The ZVEI reports August semiconductorsales in Germany (in Euros) down 8% com-pared to prior year after –5% in July and–4% for the eight months. Thus the ZVEInow sees a sales decline this year unlessmicroprocessor demand strongly recovers atappropriate price levels. Excluding micro-processors total semiconductor sales shouldend up with a 1% increase. Analysts are speculating thatSTMicroelectronics could receive a leveragebuyout offer following competitors Freescaleand NXP. ST had 2005 sales of $ 8.9 B,ahead of those of Freescale ($ 5.8 B) andNXP ($ 6.9 B). The firm has increased sales

and profits but is in a major move to improveoverall productivity. The French operationsare to be divided into four companies part ofa Montrouge-based holding, namely theindustrial sites of Tours (power), Rousset(secure communications solutions), Crolles 2(cooperation with NXP and Freescale) andGrenoble (part of competition poleMinalogic). NXP, the former Philips Semiconductor unit,has named Pascal Langlois Sr. VP Globalsales and Guido Schlegemilch now headsthe transistor business unit and is succeed-ed by Michel van Crombrugge as sensor unitmanager. The firm claims strength in inter-face ICs, discretes and standard analog ICs,sees MCUs and power management asfocus for increased efforts, so the company’sJan Willem Vogel. Asia now accounts forover 60% of sales, the USA 10%. NXPinherits over 25 000 patents from Philips,has an R&D budget of € 1 B per year and atotal of 6700 engineers in R&D.

OPTOELECTRONICS

The increasing demand for solar cells is put-ting strains on the producers of polycrys-talline silicon and could lead to shortages forsemiconductor production. Industry statisticsinclude a second quarter 2006 51% growthof the OLED market to 19.3 M units withLGE industry leader at a 23.2% share fol-lowed by SDI 21.2% and RiTdisplay 19.2%,so Displaybank while LGE is also the topPlasma-Display panel producer with a 31%unit market share in the first quarter accord-ing to iSuppli. News from Philips includes a high powerLED production facility in Singapore doublingcapacity by end of 2007, claims N° 1 posi-tion in this field for its Lighting Division whichalso cooperates with Novaled in whiteOLEDs including the recent Olla project(Organic LEDs for Lighting Applications).Affiliate Lumileds now offers most productsRoHS compatible.

PASSIVE COMPONENTS

DECISION made a 100+ page presentationat a recent CARTS Europe passive compo-nents marketing seminar entitled “Electronicindustries and the world market for passivecomponents” including magnetic compo-

nents, capacitors, resistors, quartz/oscillatorsand filters. World passive component con-sumption last year was worth € 26 B withChina $ 6.4 B or 25%, Europe and NorthAmerica $ 4.9 B and $ 5 B respectively,Japan and other Asia Pacific $ 4.3 B eachand ROW $ 1 B. Capacitors accounted for39%, magnetics 33%, resistors 11%,quartz/oscillators 8% and filters 9%. ContactDECISION, tel. 33/1 45057013 for moreinformation. Murata’s last annual revenue of $ 4.194 Bcame 73% from exports while MLCCs con-tributed 35.5% with monthly production of 30B units. Communication and Computer &Peripherals accounted for 42% and 21%respectively with automotive in Germanyalready 40% of sales. Sy Chip bought lastApril is a wireless chip module specialistopening entry into integrated passive devicetechnology as the firm moves torward cus-tomized module solutions, anticipates 10%growth this fiscal year. China’s Jianghai, an electrolytic capacitorspecialist, with 2005 production of 2.3 Bunits claims European sales in the double-digit million range, 80% in industrial but withautomotive a key target market.

OTHER COMPONENTS

Bosch is going through a transformationunder CEO Franz Fehrenbach aiming atdiversification from the automotive sector,now two-thirds of total sales, mainly throughacquisitions and focus on services. Boschhas overtaken struggling Delphi to becomethe world’s largest car parts maker, with € 41B sales and 249 000 employees. The AC/DC power supply world market is togrow from $ 11.8 B this year to $ 14.8 B by2011 with 301-500 W devices growingfastest, so Darnell. TDK-Lambda recently consolidated thepower supply activities under the Densei-Lambda name reporting to Takeo Suzuki.Germany, European headquarters and logis-tics center, contributes about 33% toEuropean sales with 80% of products sup-plied from there.

[email protected]

THE LENNOX REPORTELECTRONIC COMPONENTS INDUSTRY

M A R K E T


Fuji Electric Device Technology Europe GmbHGoethering 58 · 63067 Offenbach am Main · GermanyFon +49 (0)69 - 66 90 29 0 · Fax +49 (0)69 - 66 90 29 [email protected]

6-Pack IGBT600V : 15A - 150A

1200V : 10A - 150A1700V :100A &150A

PIM IGBT600V : 30A - 100A

1200V : 10A - 75A

RectifierModules

800V : 30A - 250A1600V : 30A - 125A

Discrete IGBT600V : 5A - 75A

1200V : 3A - 25A

IPM-IGBT600V : 15A - 300A

1200V : 15A - 150A

1200V :225A - 450A

www.fujielectric.de

Your Quality Partner forPower Solutions

Your Quality Partner forPower Solutions

New DualPACKs

New High Power IGBT

Special version availablefor rough environments

1200V :225A - 450A

High Power 6-Pack1200V : 225A - 450A1700V : 225A - 450A 2-Pack IGBT

600V : 50A - 600A1200V : 50A - 450A1700V :150A - 400A

1-Pack IGBT600V : 600A

1200V : 200A - 800A

with Spring Contacts

with Soldering Pins1-Pack

1200V : 1200A - 3600A1700V : 1200A - 3600A

2-Pack1200V : 800A &1200A1700V : 600A &1200A


M A R K E T

The dc-dc converter module market has been subject to a number offactors over the past couple of years that have brought down DarnellGroup’s projections for dollar sales between 2006 and 2011. In pro-jections made during early 2005, the worldwide dollar market for dc-dc converter modules was expected to be approximately $3.6 billionin 2006. Currently, that number has been lowered to approximately$3.2 billion for 2006. This is a 12.5% drop, and it is due to severalbusiness and economic factors in the dc-dc converter market. Thesefactors differ, depending on whether you’re looking at the brick mar-ket or the nonisolated market.

Unit sales for dc-dc converter modules have been relatively flat. Infact, we only reduced our previous 2006 worldwide forecast by 2% inour current update. Last year, we expected 2006 unit sales of 104.3million, now we’ve slightly reduced this figure to 102.0 million units.So the 12.5% revenue drop cannot be explained by declining unitsales.

What we could not have predicted was that Delta Electronics hasgrown about 40% a year for two consecutive years. Delta has beenincreasing market share in the dc-dc converter market and is now amarket share leader. Its Power Systems Business Group II, whichincludes sales of Delta’s board mounted dc-dc converters, has grownfrom about 9% to 16% of Delta’s business.

Delta’s primary business in the power electronics industry has beenhigh-volume, low-margin production of ac-dc power supplies for thecomputer, consumer and communications markets. Now this industrypowerhouse is bringing its competitive, ac-dc power supply pricingmodel to the dc-dc converter module world. Mergers, such asEmerson’s acquisition of Artesyn or Power-One’s pending acquisitionof Magnetek’s Power Electronics Group, will simply reinforce thesepricing pressures.

As a result of this and other pricing pressures, dollars-per-Watt anddollars-per-amp declined sharply over the last two years. In the high-er-wattage brick segments (not including bus converters), downwardpressure has pushed prices well below the $0.28/W reported in our2005 report. Several companies, including Delta, have releasedbricks, not bus converters, that have pushed pricing to at or justbelow $0.20/W. Additionally, competition from semiconductor housesis increasing pricing pressure on nonisolated Point-of-Load (POL)converters. These products can regularly be found near $0.50/A.

While pricing pressure has reduced dc-dc converter module revenuefor 2006 compared to previous projections, the dc-dc converter mar-ket is still expected to grow over the five years. While our earlier fore-casts projected the dc-dc converter modules to experience a 10.5%five-year compounded annual growth rate for revenue, this has beenreduced nearly two percentage points to 8.6%. As seen in figure 1,the dc-dc converter market is expected to grow from $3.2 billion in2006 to $4.8 billion in 2011.

Further affecting the revenue landscape is the maturity of theIntermediate Bus Architecture (IBA). The IBA architecture grew quick-ly, with new product introductions peaking in 2004, according toDarnell’s PowerPulse daily newsletter. In 2005, however, new productannouncements started tapering off, and that trend is continuing in2006. The IBA market appears to be maturing and reaching “satura-tion” in terms of new product adoption. While this will slow IBA unitsales growth, sales of bus converters and POLs are still slightly out-pacing the market due to their use in faster-growing applications suchas blade servers, storage equipment, MicroTCA and AdvancedTCA.

Although the IBA once drove sales of dc-dc converters, IBA adoptionhas flattened out as a result of normal market dynamics. Since anycontinued adoption of the IBA will not be large enough to ramp up dc-dc converter sales on its own, business trends will have to be the pri-mary driver of the dc-dc converter industry over the next two years,which will result in a challenging market environment for makers ofdc-dc converter modules.

DC-DC Converter ModuleMarket Slows

Competition, Consolidation and Maturity Hamper Growth

Within the power electronics industry, dc-dc converter modules have traditionally beenviewed as a bastion of growth; however, a number of forces have recently converged,

which will reduce growth opportunities for the foreseeable future.

By Jeremiah P. Bryant, Managing Research Analyst, Darnell Group

Figure 1: Worldwide DC-DC Converter Module Market

0

1,000

2,000

3,000

4,000

5,000

Mill

ions

of U

S$

2006 2007 2008 2009 2010 2011

Turning to the nonisolated portion of the dc-dcmarket, competition is coming from semicon-ductor makers. Power management ICs and“hybrid” products have reached a higher-cur-rent “node” and are now encroaching on terri-tory that had, until recently, been the exclusivedomain of modules. In particular, the >3Aranges are where ICs are taking market sharefrom the module makers, and the >5A to #10Arange is experiencing the greatest dynamicsand growth. By comparison, these sameamperage ranges in nonisolated dc-dc con-verter modules are experiencing either declin-ing growth or slow growth. Combined, the>5A-#10A and >10A segments will go fromjust 8% of the Converter/Regulator IC marketto 11% of the market between 2006 and 2011.Given this new competition, module makershave had to trim margins in order to remaincompetitive.

Another important factor affecting dc-dc con-verter module sales is the changing competi-tive environment. For years considered a frag-mented industry, with hundreds of participantscompeting for market share, the dc-dc convert-er market is undergoing a broad consolidation.Many of the larger manufacturers are turningto a variety of business strategies, includingpartnerships, consolidations, and mergers andacquisitions in order to compete successfullyin the market. In 2005, the top 10 marketshare leaders combined for about 35% of theworldwide dollar market. This year, the top tencompanies make up over 44% of the market.In fact, in 2006, the top two companies, DeltaElectronics and Emerson (including new acqui-sition Artesyn and existing power conversioncompany Astec), make up almost 17% of theworldwide dollar market, and that number isexpected to grow.

Application dynamics are also playing a keyrole in the evolution of the dc-dc converter

module market. For example, theComputer market is expected to over-take Communications as the largestworldwide application by 2011. In 2006,Communications is projected to beabout 46% of dc-dc converter moduleunit sales, with Computers accountingfor about 44%. By 2011, Computers willincrease to 50% of the market, whileCommunications declines to approxi-mately 42% of the market.

These current business factors, com-bined with broader economic indicators,are expected to contribute to lower dc-dc converter revenues over the nextfew years. This is supported by applica-tions such as servers, which have expe-rienced eight consecutive quarters ofdeclining growth rates. Regionally, only

North Americaand Asia(excludingJapan) sawserver salesgrowth duringthe second quar-ter of 2006, andthis was only3.6% and 2.6%,respectively.Additionally,high-priced con-verters used inMilitary/Aerospace andIndustrial appli-cations are alsoseeing slower

growth as spending is being divertedelsewhere.

Dc-dc converter module makers arenow facing a challenging market envi-ronment. Despite falling prices and theconcurrent decline in revenue forecasts,however, opportunities still exist formodules. The business and technologyclimates may be changing, but if powersupply makers are aware of theseshifts, they can take advantage of theopportunities that such shifts present.

www.darnell.com

www.bodospower.com

M A R K E T

Figure 2: Comparison of Nonisolated POL Module andConverter/Regulator IC Growth Rates by Amperage Range

-5%

0%

5%

10%

15%

20%

Com

poun

d A

nnua

l Gro

wth

Rat

e

<5A 5-10A >10AConverter/Regulator ICs Nonisolated Modules

If you’re designing high-performance

or high-reliability power supplies, you

can count on the magnetics experts at

Datatronics. From aerospace to instru-

mentation to consumer products, and

automotive too, Datatronics has the

right experience, technology, quality

and cost-effective approach.

The Power SupplyMagnetics Experts

We offer one of the industry’sbroadest lines of custom andstandard magnetics:

Gate Drive, Current Sense, RF,Shielded SMD Power Inductors,Forward-Converter, Flyback, Pulse,LAN, Isolation, Networking

QS9000 Certified

ISO 9001/9002

UL, BABT, CSA, VDE Available

MIL-PRF-27, MIL-PRF-21038,MIL-STD-981

NASA Space Station Approved

> Free white paper and catalogVisit us at datatronics.com for

a free Catalog and “Measuring

Transformer Distributed

Capacitance” White Paper. For

sales assistance, contact us at

Abacus Deltron Components,

Munich, Germany, Telephone

+49 (0) 89 745 19 454.

See us atElectronica, Munich,Nov 14-17, Hall A4,

Booth 247

17


V I P I N T E R V I E W

Bodo Arlt: What end markets will drivemodule technology?Izak Bencuya: Industrial markets are thefirst for modules followed by consumer appli-ance with variable speed inverter drives.This is mainly seen in Japan and China.Computing/server applications needing highefficiency will follow.

Bodo Arlt: What is Fairchild’s positionbesides the wide range of discretes and theselected modules?Izak Bencuya: We have a strong systempower expertise that will define and deliverthe most cost effective power solution foreach application.

BodoArlt: What are the technologies thatcan offer innovation for leadership?Izak Bencuya: Application specificIntegrated Smart Power Modules (SPMs)are up for innovation with multi-chip packag-ing technologies.

Bodo Arlt: Is it more in silicon, or is it part ofpackaging technology?Izak Bencuya: Both capabilities are requiredto give the system cost advantage. Designstaking advantage of modular integration areconsumer appliances, display panels,servers and desk top applications.

Bodo Arlt: What makes Fairchild differentfrom traditional module suppliers?Izak Bencuya: Fairchild offers solutions inpower ranges from <1W to kW enabled byprocess and packaging. This combined withhigh volume production capability offers themost efficient design options for our cus-tomers.

Bodo Arlt: What makes Fairchild differentfrom traditional discrete suppliers?Izak Bencuya: The balance between dis-crete and module solutions.

Bodo Arlt: How much is Fairchild involvedin the end customer’s application?Izak Bencuya: Fairchild approaches the

customer base witha strong technicalteam with experi-ence in both deviceand system engi-neering. We defineour products toalign with our cus-tomer’s designneeds for the rightmarket window.

Bodo Arlt: Howmuch is Fairchildinvolved in motionapplications usingthe advantage ofIGBTs?Izak Bencuya:

Variable speed motor drives need an inte-grated solution of driver and IGBTs to func-tion at line voltages. The 600V and 1200VIGBT product families are specificallydesigned for this application.

Bodo Arlt: What will be the target to intro-duce new module products?Izak Bencuya: Fairchild targets applicationsthat will benefit from turn-key power solu-tions. This shorten time to market and sim-plifies end product deigns.

Bodo Arlt: What will be the future for mod-ules with line voltage and driver technology?Izak Bencuya: The fundamentals consider-ing physics and cost will determine the bestapproach between multi-chip or monolithic.

Bodo Arlt: Do we expect monolithic solu-tions for modules?Izak Bencuya: For lower power levels, later-al switches can be a good solution. Fairchildhas a family of lateral switches that employ amonolithic approach. We call this family ourIntelliMAX™ products.

Bodo Arlt: Do we expect to see high voltageIC technology in the line voltage range?Izak Bencuya: High Voltage Driver ICs

(HVICs) complement the application.Fairchild has invested to have a 600 voltHVIC process and is looking into a 1200 voltprocess.

Bodo Arlt: Who are your competitors youbelieve will stimulate the race for leadership?Izak Bencuya: Infineon has good IGBTscombined with packaging resources.Voltera is doing the low voltages. Companiesin Far East turning from meet to now to inno-vative approaches.

Bodo Arlt: Are you ready for Electronica inNovember?Izak Bencuya: Fairchild will be present andshow the latest developments. We are high-lighting our SPM family demonstrating effi-cient power solutions in everyday appli-ances.

Bodo Arlt: Thank you Izak for the time andwe look forward to a successful future forpower.

Izak Bencuya, VP GM Fairchild

www.fairchildsemi.com

Interview on Power Semiconductor

Technology with Izak Bencuya,VP GM Fairchild

By Bodo Arlt, Editor BPSD

Dr. Izak Bencuya

Executive Vice President and GeneralManager, Functional Power

BackgroundDr. Izak Bencuya is responsible for growingFairchild’s Functional Power group throughthe integration and innovation of new prod-uct and process technology, and through thedevelopment of highly efficient advanced

packaging.

With more than 24 years of industry experience, he began hiscareer at Yale University researching ultra thin oxide MOS devices.Dr. Bencuya later worked at GTE Laboratories and Siliconix in vari-ous research and management roles to develop and market lead-ing edge Power Discrete devices, such as MOSFETs, IGBTs andSITs. He joined Fairchild in 1994 to start the Low Voltage MOSFETbusiness which has grown to be one of the major revenue andearnings generating lines at Fairchild.

Dr. Bencuya has a B.S.E.E. from Bosphorous University inIstanbul, Turkey, an M.S. and Ph.D. in Engineering and AppliedScience from Yale University and an M.B.A. from the University ofCalifornia-Berkeley. He was the recipient of the IBM Fellowship,the Thomas Alva Edison Fellowship and the Charles Deere WimanFellowship and is a member of the IEEE Electron Device Societyand Electrochemical Society. Dr. Bencuya holds 15 patents andhas published extensively in the electronics field.

C O V E R S T O R Y


Electric motors are the primary power usersin home appliances such as refrigerators,fans and air conditioners. Unfortunately, thesingle-phase shaded pole or permanentcapacitor induction motors typically used inmany fans or pumps have efficiencies as lowas 25%. And while the larger single-phaseinduction motors used in air conditioner andrefrigerator compressors are better with typi-cal efficiencies greater than 65%, a greatdeal of energy is still lost during start-up offixed speed compressors with low dutycycles.

In recent years, manufacturers haveimproved appliance efficiencies with variablespeed permanent magnet motors featuringcontrol systems that use electrical powerinverters to vary motor speed by changingthe motor winding voltage frequency. Insome appliances, such as fans, hall sensorsdetect rotor position to synchronize it withwinding switching in order to maximize effi-ciency and simplify start up. The advantageof this approach is in that the controlrequires very simple electronic circuits.However, a sealed compressor cannotaccommodate hall sensors and so a sensor-less algorithm is required.

One popular sensorless algorithm used insix step permanent magnet motor drive sys-tems detects the zero crossing of the wind-ing back emf. The control algorithm typicallyuses an 8-bit microcontroller to manage thephase advance and start up sequence. Onedisadvantage of the six-step system is thatthere is a torque glitch as the motor currentswitches (commutates) between windings. Inmany fan and pump applications theseglitches produce an annoying acoustic noiseespecially at low speeds when the fan bladenoise is almost zero. The ideal solution is to

drive the motor with sinusoidal currents,which completely eliminates the glitch. Thistype of control also enables the use of analternative motor design with interior perma-nent magnets (IPM). IPM motors can pro-duce 15% more torque than permanentmagnet motors, which allows for further effi-ciency improvement.

Sensorless Control

Advances in hardware and control technolo-gy have made it possible to build cost effec-tive drives for IPM motors based on a fieldoriented control (FOC) algorithm that drivesthe motor with sinusoidal currents for maxi-mum efficiency and low acoustic noise. Now,International Rectifier is taking theseadvances further with a set of integrateddesign platforms for appliance motor control,in which a mixed signal control IC imple-ments a sensorless FOC algorithm in hard-ware eliminating the need for software devel-opment. The sensorless algorithm detectsthe rotor position based on the motor currentmeasurement only, while the control hard-ware allows motor winding currents to bemeasured without expensive isolation cir-

cuits. These technologies provide designerswith a route to using IPM motors for drivingcompressors, as well as eliminating the needfor hall sensors in fans with surface magnetor IPM motors.

Figure 1 illustrates the algorithm functionsused to drive either an IPM or surface mag-net motor with sinusoidal currents and with-out sensors. A key element of the algorithmis the FOC structure that uses a vector rota-tion block (e-jè) to transform ac motor wind-ing currents into two dc current componentsone producing torque and the other control-ling the flux. The current inputs to the rota-tion block are first transformed from three-phase to two-phase equivalent values usingthe Clarke transformation block. The rotorflux angle enables the splitting of the currentinto the D component, which aligns with theflux, and the Q component that producestorque. The tuning of the two current controlPI compensators matches to the motor wind-ing RL time constant and does not have tochange with the frequency of the ac windingcurrents. The forward vector rotation block(ejè) transforms the dc voltage outputs of the

Sensorless Motor ControlAlgorithm

Delivery of Improved Appliance Efficiency

Conservation of energy is one of the most important challenges facing us in the 21st cen-tury. Recent advances in sensorless permanent magnet motor control technology can helpto address this challenge by significantly reducing the energy used in appliances such as

refrigerators, washers, fans and air conditioners.

By Aengus Murray, International Rectifier

Figure 1 Sensorless PM control algorithm

THE POWER MANAGEMENT LEADER

SIMPLIFY LAMP DRIVESLASH COMPONENT COUNT

For more information call +44 (0)1737 227215 or +49 (0) 6102 884 311 or visit us at

www.irf.com/lighting

with IR’s Robust, Reliable PFC, Buck and Full-Bridge ICs for HID Ballast Designs

Make your HID ballast design robust andreliable with International Rectifier'slighting ICs:

• High voltage technology for robust drive

• Simplifies lamp drive

• Reduces component count

• No mercury migration shorteninglamp life

PFC Stage

• IR1150 µPFC™ continuous conduction modePFC controller

Buck Stage

• IRS2117 single high-side driver

• IRS21844 half-bridge driver

Full-Bridge Stage

• IRS2453D self-oscillating full-bridge driver

BOOST PFC

Part Number Package VCC IO +/- Frequency Tamb

IR1150(I)PBF DIP-8, SO-8 13V-22V 1.5A 50-200 kHz 0 to 70˚C

BUCK

Part Number Package Voltage IO +/- VOUT T on/off (typ)

IRS2117(S)PBF DIP-8, SO-8 600V 200mA/420 mA 10V - 20V 125ns, 105ns

Part Number Package Voltage IO +/- VOUT Input Logic

IRS21844(S)PBF DIP-14, SO-14 600V 1.4A / 1.8A 10V - 20V 3.3V

FULL BRIDGE OUTPUT

Part Number Package Voltage IO +/- Duty InternalCycle Deadtime (typ.)

IRS2453D(S)PBF DIP-14, SO-14 600V 180mA / 260mA 50% 1µs

µPFC™ is a trademark of International Rectifier

See us on stand: A5.576.

C O V E R S T O R Y


PI compensators into ac voltages matchedto the rotor frequency. The Space VectorPWM unit calculates the power inverter tran-sistor switching times needed to apply thecalculated ac voltages. Space Vector modu-lation produces sinusoidal voltage modula-tion with automatic third harmonic injectionto maximize the use of the dc bus voltage. Italso includes a two phase modulation func-tion that will minimize the power inverterswitching losses.

Maximizing the motor torque output perampere achieves maximum motor efficiency.In the case of a surface magnet motor, thecontroller maintains the flux component ofcurrent (ID) at zero to maximize efficiency.However, the construction of an IPM motorproduces an additional torque componentknown as reluctance torque. When drivingan IPM motor, the IPM control block increas-es the ID current from zero as a function ofthe IQ target in order to operate at maximumefficiency. In either case, the speed loopcompensator calculates the required torquecurrent needed to maintain the speed at thetarget value. There are some applications,such as washing machines, that require anexpanded speed range. Here, the fieldweakening control function inserts negativeflux current (ID) to reduce the effective backemf of the motor and allow the motor to runat a higher speed before the back emfreaches the dc bus voltage limit.

A unique feature of this algorithm is its abilityto derive rotor position and velocity from themotor winding currents without a physicalsensor. The sensorless algorithm derives therotor flux functions from the motor circuitmodel based on the following equations:

The controller drives the stator voltages andthe current reconstruction circuits measurethe resultant motor currents. A simplereordering of the equation terms and mathe-matical integration yields the sine and cosineterms. A phase locked loop tracking algo-rithm, derives both angle and velocity.

The second unique feature of the algorithmis the phase current reconstruction unit thatderives the motor phase currents from thecurrent flowing in the inverter dc link. Theprinciple, illustrated in figure 2, is that for anyactive inverter state there will always be onewinding connected to one bus rail and twowindings connected to the other bus rail.This means that in every PWM cycle thereare two motor winding current values avail-able. The reconstruction unit includes a sam-

ple timing generator based on the SVPWMinputs, a sampling A/D converter and themathematical unit to calculate the thirdphase current. The significant advantage ofthe approach is that it eliminates the require-ment for isolated current sensing and somakes the current sensorless algorithm costeffective in appliance applications.

Additional Functionality

In addition to the algorithm implementation,as the illustration in Figure 3 shows, the IRcontrol IC integrates the A/D converter andbuffer amplifiers needed for current meas-urement, and additional hardware functionsfor error handling and start up sequencing.The IC also integrates an 8-bit microcon-troller core with its own memory to allow theappliance engineer to implement additional,application-specific functionality.

www.irf.com

( )( )( )( )rrdt

ddt

diSS

rrdtd

dtdi

SS

LiRv

LiRv

θ

θβ

α

ββ

αα

sin...

cos...

Ψ−++=

Ψ−++=

Figure 3 – Block diagram or similar of controller IC functions

Figure 2 There phase current reconstruction

I N F I N E O N T E C H N O L O G I E S offers a broad range of leading-edge power

semiconductors for standardized and application-specific industrial applications such

as industrial drives, renewable energies, transportation, power supplies and medical

instruments. Our proven chip expertise combined with many years’ package know-how

enable our customers to select the right solutions for their applications.

You are most welcome to join us at SPS/ IPC / DRIVES 2006, 28 – 30 Nov, in Nuremberg.

We can be found in hall 1, booth 1-658.

www.infineon.com

Efficient POWER for your applications

In 2004, first time Mitsubishi customers ben-efited from a direct link to its new calculationtool. With the constant improvement, in2006, that program has been enhanced andexpanded to include even more operatingmodes and more detailed output data. Apower designer can analyse the perform-ance of Mitsubishi’s power devices with thissoftware efficiently, optimizing the productivi-ty by reducing the development time. Thisarticle deals with each feature of MEL-COSIM and its function in detailed. Itdescribes also the guidelines for the effec-tive use to customers with sample outputresults of 100A/1200V dual IGBT module(CM100DY-24A).

As a power designer, it is necessary to cal-culate and analyze the power module per-formance on the basis of its losses (switch-ing, and conduction losses) and junctiontemperature rise for the design of the systemunder given application conditions. In addi-tion to the delivery of high performance andreliable power modules to customers, inorder to increase the efficiency of the designand the development Mitsubishi offers anenhanced, compatible online tool for theappropriate module selection and powerloss analysis. The power loss analysiscan be done in DC-AC inverter and DC-DC chopper configuration under theassumption of sinusoidal and constantoutput currents at inductive loads.Compared to the similar available toolsMELCOSIM offers a several advantagessuch as fully compatible to Windows oper-ating systems, independent on other soft-ware platform, dedicated design to imple-ment it for the Mitsubishi modules andeasy to use features. MELCOSIM pro-vides calculation results as graphical and

text output data at fast speed.

Operational features of MELCOSIM

Customers can access MELCOSIM throughthe Mitsubishi web site at www.mit-subishichips.com. This software is compati-ble with Windows98/NT/2000/XP. To providean updated info time to time for theMitsubishis customers, a quick registrationprocess is required to access this software.After registration, download of MELCOSIMon own hard disc is possible. Once thezipped file is downloaded, only a quickinstallation process is needed. For thedetailed instruction please open the “ReadMe” file first. After the successful installationuser will be ready to use this software imme-diately. Selection of modules and input of applicationconditionsThe first step in the use of MELCOMSIM isto consider worst-case application operatingconditions. User can select the operatingconditions parameters like DC-link voltage(VCC), output current (IO), switching fre-quency (fSW), power factor (PF) and alsomodulation parameters like modulation ratio

(D) or duty cycle (for chopper mode) to real-ize the calculation conditions comparable toreal application conditions.

Figure 1 depicts the main window. The leftpart of this window is reserved for applica-tion conditions input: DC link voltage (VCC)and switching frequency (fSW) are importantfor switching loss calculations, while modula-tion ratio (D) and power factor (PF) influ-ences the distribution of conduction lossesbetween the IGBT and FwDi. The output cur-rent IO is necessary for both conduction andswitching losses. Additionally module type(CM100DY-24A) and its thermal resistancesare shown in the above part of main window.

Based on the worst case application condi-tions module pre-selection must be done.The main criteria for the module type pre-selection are:

The rated current (IC) of the module shouldbe more than the half of the peak output cur-rent (ICP)

The blocking voltage VCES of the deviceshould be more than 1.5 times the appliedDC link voltage (VCC)

Once the appropriate VCES and IC for thedevice is chosen, the module can be select-ed from the available in-built product line-upvia a module section button from tool menubar. The present version includes an updat-ed device data file with 45 new devicesincluding HVIGBT IGBTs up to 2400A/1700Vand 600A/6500V and the new 600V DIP-IPMVersion 4 (Super Mini-DIP-IPMs). Now MEL-COSIM can be used for the checkingwhether pre-selected device is suitable fromthermal point of view.

D E S I G N A N D S I M U L A T I O N


Selection, Power LossCalculation and Temperature

Rise Analysis All-in-one efficient online tool for the power modules

MELCOSIM offers a variety of possibilities to calculate the power loss and temperaturerise for several different circuits and load conditions in order to give application support

for the available Mitsubishi power modules.

By Prasad Bhalerao and Eugen Stumpf, Mitsubishi Electric Europe, Germany

Figure 1. Main window of MELCOSIM

For a better environment

[email protected] · www.mitsubishichips.com

IPM Modules IGBT Modules DIP-CIB Modules

Mitsubishi's service-proven product ranges with the latestchip and production technologies for best efficiency, highestreliability and, needless to say, good environmental compatibility.

Mitsubishi offers a large variety of RoHS compliant series foryour industrial application. And compatibility with previousproduct generations is always assured.

All the power you need...

The selected device is marked as a targetdevice in the input and output window. Fig.1shows an example of loss calculation inputand output parameters for better understand-ing of the each features. The modulationstrategy such as 3-arm modulation (contin-ues) 2-Arm modulation (discontinuousspace-vector, in which PWM pattern is var-ied such that each device is subject toswitching for two-thirds of a cycle) and DCchopper modes (typically used in DC to DCconverter circuits) can be selected depend-ing on the customer’s requirements. Figure 2shows the typical circuit configuration for 3arm PWM modulation scheme with the rele-vant input parameters for loss calculations.The current version of MELCOSIM (version3.0) is able to calculate the average junctiontemperature Tjav in transistor (IGBT) andfree wheel diode (FWDi). The differencebetween maximum junction temperatureTjmax and Tjav can be neglected in case ofoutput frequency fout is above 30 Hz. For thejunction temperature calculation in case ofoutput frequency is below fout=30 Hz, Tjmax

must be considered.

The next MELCOSIM version 4 will contentthe possibility of fout input and Tjmax calcula-tion. MITSUBISHI plan to introduce version 4by end of year 2006.

MELCOSIM is checking all input parameter(application conditions) for the selected mod-ule for limits provided in specification. Incase an input parameter is not in a specifica-tion range, the warning window will beappeared.Loss calculationThe total average power loss of an IGBTwith sinusoidal output current is the sum ofIGBT static loss and IGBT switching losses.Similarly, the total loss of free-wheelingdiode is a sum of switching loss and static

loss. The static loss of IGBT is given inequation (1) and equation (2) calculates theswitching loss of IGBT. The equation (3) andequation (4) gives the steady state loss andrecovery loss of the free-wheeling dioderespectively.

Eq. (1)

Eq. (2)

Eq. (3)

Eq. (4)

where, ICP : Peak value of sinusoidal outputcurrent, PF : power factor, VCE(sat) : IGBTsaturation voltage drop at ICP and Tj =125°C,D : modulation index, ESW(on) & ESW(off) :IGBT’s turn-on and turn-off switching energyper pulse at peak collector current ICp andTj= 125°C, Irr : Diode peak reverse recoverycurrent, trr: Diode reverse recovery time,VCE(peak) : Peak voltage across the diode atrecovery, VEC : Diode forward voltage drop.

Calculation results display the total loss inone module, total loss in an IGBT/Diode andthe separate IGBT/Diode loss: switching andconduction loss components. The results ofthe power loss can be plotted graphicallysuch as: Av. Power loss (P) vs. peak output current(ICP) as shown in Figure 3.

Av. Power loss (P) vs. switching frequency(fsw)

On the right part of graphical output the fixedapplication conditions are given. The red lineprovides IGBTs result and the blue lineFWDi results. The variable application condi-tion Icp (Figure 3) is limited by two times ofmodule rated current: Icp(max)=200A forCM100DY-24A. This limit is corresponded tothe RBSOA specification.

Figure 4 shows the part of the main window,where the power loss and junction tempera-ture of IGBT and diode in the moduleCM100DY-24A are summarized.

Temperature rise calculation.

The calculation of the temperature rise isnecessary for checking whether pre-selectedmodule is suitable for the given application,for the heat sink design and also for the con-sideration of over temperature protection.

Figure 5 shows the equivalent thermal modelof IGBT and Free-Wheeling Diode (FWDi)for the calculation of the temperature rise.

The crosstalkbetween transis-tor and diodeloss areoccurred alreadyin the modulebaseplate. Forthe calculation inMELCOSIM,thermal resist-ance values of

the IGBT and FWDi are measured just underthe chip. MELCOSIMs interface to user isheatsink temperature Tf must be measureddirect under the IGBT/Diode chip. In additionto the magnitude junction temperature MEL-COSIM provides the average temperaturerise between junction (j) and module base-plate (c) as well as module baseplate (c) and

heatsink (f) of IGBT andDiodes All temperaturecalculation results aresummarized in the out-put part of main windowshown in Figure 4.

Additional to above men-tioned results calculatedfor one operation condi-tion, MELCOSIM is ableto provide graphical out-put with changeableparameters Icp and fsw:Junction temperature (Tj)vs. peak output current(ICP)

swpeakCErrrrFWDrr fVtIP ....125.0 )(=−

−=− PFDVIP ECCPstaticFWD π38

1..

dxSinXfEEP SWoffSWonSWswIGBT ∫+=−

π

π0

)()(2

1.).(

dxDPFXXSinVIP satCECPstaticIGBT2

).sin(1.

2

1..

0

2

)(

++= ∫−

π

π



Figure 2. Basic 3 arm PWM modulation anddefinitions

T fsw=1/T

Ht Hs

Modulation Ratio ma = Hs/Ht

Chopper or sinus

Figure 3. MELCOSIM Graphical output p vs. ICP

Figure 4 shows the window, where the power loss of IGBT anddiode is summarized.


27www.bodospower.com

Junction temperature (Tj) vs. switching frequency (fsw) as shown in Figure 6.

The line with red colours shows IGBT temperature and line with bluecolours shows FWDi temperature. The striped rectangular on Fig. 6shows not recommended area above Tj=125 degree. The maxspecified junction temperature is Tj=150 degree.

The graph of peak output current (ICP) vs. switching frequency (fsw) atfixed junction temperature is interesting for selection of the properpower module. This graph is shown in Figure 7. The target junctiontemperature can be adjusted in a separate window for Tj input. Thetarget junction temperature in Figure 7 is Tj=125 degree. For a partic-ular switching frequency, the graphical analysis shows directly thepeak value of the output current at selected junction temperature.Figure 7 shows that for module CM100DY-24A by using of switchingfrequency fsw=5kHz the maximal collector current Icp=150A can bereached provided that junction temperature will not exceed Tj=125degree.

The end result data can be stored or transmitted graphically and intext format. Multiple calculations with varying conditions can be com-pared by saving one calculation window and creating a new windowby clicking on the “new window” button on your toolbar.

Summary

In short, MELCOSIM helps to select the suitable Mitsubishi powermodule for specific design considered the customers worst-caseapplication conditions. Additionally MELCOSIM provides a quick andeasy way for junction temperatures and power loss calculation bysimply entering module types of interest and then adjusting thedefaulted specifications to match customers application conditions.The program will instantaneously calculate junction temperatures andpower loss of IGBT and FWD under the given application conditionsin a short time.

Points to be aware of while using the MELCOSIM:The module data based on the respective IGBT module data sheet.The data used in MELCOSIM may subject to changes, improve-ments or correction without prior notice to user.Calculations are based on the linear approximations.Due to slight deviation in the operating condition parameters, theprogram may not replace a detailed reflection of the customer appli-cation with all its operating conditions.

www.mitsubishichips.com

Figure 5. Thermal Model of IGBT and Free-wheel Diode

Figure 6. Temperature rise graphical analysis by MELCOSIM

more than you expect

www.we-online.com

Rapid prototyping with WE-FLEX tranformer:

25 different sizes with 300 transformerand choke variants. Power level 1W-50W.Available ex stock. Consulting service included.

DELIVERYEX STOCK

DESIGN KITS

SAMPLES &BELOW MOQ

EMC Components

Inductors

Transformers

RF-Components

Press-Fit Technology

WERI Connectors

Fastener Technology

Transformer forDC/DC-Converter

electronica Munich

November 14 -17th

B6 stand 424

Figure 7. MELCOSIMGraphical output ICPvs. fsw at target juntiontemperature Tj

M O S F E T


Abstract

The power supply market is dominated by majority carrier (highspeed) switching devices. Fairchild Semiconductor has introduced anew highly optimized high speed fast body diode planar MOSFET(UniFET) for the SMPS market by reducing overall loss over regular500V MOSFET. By using this MOSFET, the power supply efficiencycan be improved or other trade off such as cost reduction, sizereduction or increase in power density can be realized. This MOS-FET reduces the switching loss while keeping EMI in control. ThisMOSFET rated at 500V can compete with some of the super-junctionMOSFETs. The new highly optimized MOSFET with fast body diodeis designed to address the diode reverse recovery problem. This arti-cle also analyzes how diode behaves in third quadrant. The advan-tages of this fast body diode MOSFET body diode compared tosuper-junction fast body diode MOSFET are lower Qrr & Trr, bettersoftness and high dv/dt. A special lifetime control further reduces Qrrand Trr and increases dv/dt ruggedness.

UNIFET

Fairchild has introduced the new optimized fast body diode cell struc-ture DMOS technology known as UniFET for and SMPS industry.High efficiency requirement of SMPS are driving the demand ofimproved power devices. Server Power Supply, Telecom, MedicalEquipments, UPS & Solar Inverters and House Hold appliances hasbecome reality. Solar and UPS inverter also requires fast body diodeMOSFET. At present, generally high speed switches, such as con-ventional DMOS or SuperFET is the natural choice for these applica-tions. The parasitic capacitances for the UniFET have been opti-mized and the Miller capacitance (Cgd) has been reduced in order toimprove the performance. The output capacitance has also beenreduced this has reduced the turn-off energy loss and at the sametime the ringing in turn-off current has been reduced compared toother MOS technologies in order to reduce EMI. The gate ESR hasbeen reduced to improve the dynamic switching performance of thedevice. The dynamic changes of parasitic capacitances are gradualand uniform which reduces ringing during turn-on and turn-off there-by reducing EMI. The specific Rdson of the UniFET has also beenreduced by designing new optimized cell structure.

Zero Voltage Switching (ZVS)

High switching frequency allows reduction in power supply volumeand weight. The key of increasing the switching frequency is to usesoft switching topologies. ZVS topology operating at high frequencycan improve the efficiency as well as reduce the size of the power

supply. ZVS also reduces the stress on the semiconductor switch,improving the reliability. These advantages have made the high volt-age high power Phase-Shifted Full Bridge ZVS PWM converter avery popular topology. The conventional power MOSFET body diodecan cause device failure during switching due to un-recovered minori-ty carriers. The Fairchild fast body diode UniFET has been targetedfor this application. The diode recovery plays vital role not only inabove topology but also UPS & motor drive inverters as well as solarinverters. The reflected load to these inverters can also be none lin-ear and this causes third quadrent current in the MOSFET. Thebehavior of MOSFET body diode has been analyzed in this condition.

ZVS Topology Description

The Phase shifted full-bridge zero-voltage switched (ZVS) pulsewidth modulated (PWM) converter (1) operating in excess of 100 kHzis shown in figure 2 is a simplified schematic of the power circuit. Theconventional full-bridge topology is switched off under hard switchingconditions where switch voltage stress is also high. The conventionalfull bridge topology has been modified in two ways to achieve ZVS.

First modulation is done by phase shifting two overlapping constantfrequency square waves by using leading-leg and lagging-leg.Second, the soft switching (ZVS) is achieved to minimize or reducethe switching losses.

In contrast to turning on the diagonally opposite switches of thebridge simultaneously (i.e. Q1 & Q4, Q2 & Q3), a phase shift is intro-duced between the switches in the left leg (leading-leg Q1 & Q2) andthose in the right leg (lagging-leg Q3 & Q4) as shown in Figure 1 & 2.This phase shift determines the operating duty cycle of the converter.

A Fast Body Diode UniFET

Analyzed and Tested under Zero Voltage Switching

The power supply market is dominated by majority carrier (high speed) switchingdevices. Fairchild has introduced a new highly optimized high speed fast body diode

planar MOSFET (UniFET) for the SMPS market by reducing overall loss over regular500V MOSFET.

By Sampat Shekhawat, Praveen Shenoy, Mark Rinehimer and Bob Brockway, Fairchild Semiconductor, Mountaintop, USA.

Figure 1: The difference between regular Full Bridge and PH-FullBridge ZVS PWM DC/DC converter topologies control switching

Q1-ON 180’ Q1-ON 180’ Q1-ON 180’

Q2-ON 180’ Q2-ON 180’

Q3-ON 180’ Q3-ON 180’

Q4-ON 180’ Q4-ON 180’

Transformer PrimaryWinding Voltage

M O S F E T

www.bodospower.com30

ZVS process

Zero-voltage turn-on is achieved by using the energy stored in theleakage and series inductance of the transformer to discharge theoutput capacitance of the switches through resonant action. The res-onance forces the body diode into forward conduction prior to gatingon the switch. Two different mechanisms exist which provide ZVS [3]for the lagging-leg and leading-leg.

1) During light loads, very little energy is stored in the primary sideinductance L1. This causes the lagging-leg to turn on under a hardswitching condition. As the load increases the energy stored ininductor L1 increases. This energy is used to charge the outputcapacitance of the devices that are turning off, and to discharge theoutput capacitance of the complementary device, thus forward bias-ing the free-wheeling diode. A switch with low output capacitancehelps to achieve ZVS process at light load improving the efficiency.The UniFET output capacitance has been reduced for this very samerequirement.

2) For the leading-leg switches (Q1& Q2), a different process pro-vides the ZVS as explained below. Before Q1 turns off, the currentin the primary reaches its peak value of the reflected filter inductor(L2) current. When Q1 is turned off, the energy available to chargethe output capacitance of Q1 and to discharge the output capaci-tance of Q2 is the sum of the energy stored in the output filter induc-tor L2 and the primary side inductor L1. The energy stored in filterinductor L2 is available because the filter inductor current does notfreewheel through the diode until the voltage across the secondaryhas fallen to zero. In this mode, even at lighter loads, much morestored energy is available to turn-on and turn-off the leading-legswitches than is available for the lagging-leg switches. Therefore,the body diode in the leading-leg, turn-on before the UniFET is gatedon. This will reduce the turn-off loss of these switches.

In case of solar and UPS inverters also output filter is used toreduce THD. Under certain conditions similar situation can occur asexplained for phase shifted full bridge PWM-ZVS application. In allthese situations once the body diode conducts the UniFET can begated-on. Once the UniFET is turns-on it carries current in the thirdquadrant.

Importance of free wheeling diode for this topology

The role of MOSFET body diode must not be taken lightly in the ZVStopology. Even with the reduced switching stress on the MOSFET,failure may occur because of a poor diode recovery. The failureshave been reported [1, 2] at no or light load conditions. These fail-ures result from the lagging-leg loosing ZVS at turn-on and turn-offforcing a hard switching condition. One potential failure mechanism is

due the CGD*dVDS/dt current. The resulting CGD*dVDS/dt currentcan cause VGS to charge above VTH (re-applied dVDS/dt). A shootthrough condition will exist where both of the MOSFETs in the lag-ging-leg can turn-on and cause the leg to fail.

In a ZVS topology, the intrinsic body diode of the MOSFET is forcedto conduct current before the channel of the device is turned on(Figure 3a). As current flows through the body diode, minority carri-ers are generated in both the N- epi and P body regions of the device(Figure 4b). When the gate of the MOSFET is turned on, most of thetotal current is diverted in the third-quadrant mode through the seriespath including the MOS channel, the parasitic JFET, and the driftregion resistance.

In this mode, source-to-drain conduction occurs through both thebody diode and the associated channel regions of the device (Figure3b). The addition of this parallel current path begins to reduce injec-tion of minority carries into the N- region of the body diode.

As the current in the primary transformer changes direction, currentflowing through the associated channel regions also changes direc-tion to a first-quadrant mode of conduction. This forces a smallamount of reverse current in the body diode (Figure 3c), the smallmagnitude of the current being proportional to the low reverse volt-age across the body diode. Since the injecting P body / N- junctionis in parallel with the low resistance channel regions, the effectivereverse voltage across the diode is only on the order of several hun-

Figure 2: Phase shifted PWM full bridge ZVS DC/DC topology

Controller

Q1

Q2

Q3

Q4

L1

L2

L2

Figure 3: (a) Forward current through body diode, (b) Forward currentthrough body diode & channel, (c) Reverse current through bodydiode and channel, (d) Reverse current through channel and (e)Reverse current through body diode

Drain Drain Drain Drain Drain

Source Source Source Source Source

JFET JFET JFET

Rch Rch Rch

Repi Repi Repi Repi Repi

Rc,s Rc,s Rc,s Rc,s Rc,s

+ + +

- - -

(a) (b) (c) (d) (e)

Drain Drain Drain Drain Drain

Source Source Source Source Source

JFET JFET JFET

Rch Rch Rch

Repi Repi Repi Repi Repi

Rc,s Rc,s Rc,s Rc,s Rc,s

+ + +

- - -

(a) (b) (c) (d) (e)

Figure 4: Current conduction in the 3rd quadrant (drain voltage is–ve): (a) Conduction of current when gate is ON - through the MOSchannel, series JFET & drift regions. (b) Conduction of current whengate is OFF - through integrated body diode.

M O S F E T


dred milli volts. This results in minimal removal of minority carriersfrom the heavily modulated N- region of the diode structure. In theabsence of sufficient voltage to force significant carrier removal, onlya relatively small reduction in minority carrier concentration due tolifetime-controlled recombination mechanisms occurs (Figure 3d).

When the MOSFET is turned off, the high reverse voltage results inthe rapid removal of excess carries as the drift region begins to sup-port the applied bias. Minority carries in the drift region at this timeare swept across the P body / N- junction (Figure 3e). This rapiddepletion results in a significant current density flowing through the Pbody region, a portion of which extends beneath the source of theMOSFET structure. The resistance of P doped region directlybeneath the source of a conventional MOSFET structure is repre-sented as Rb in Figure 4. If the magnitude of the current passingthrough Rb is sufficiently large to cause injection across the P-body /N+ source junction, the parasitic bipolar transistor may becomeactive. This uncontrolled state of operation usually results in thedestruction of the device.

FRFET for ZVS Topology

The Qrr and Trr of the body diode should be low enough to providecomplete minority carrier removal before the device turns-off. If thisdoes not occur, the turn-off dVDS/dt of this device could turn-on theparasitic NPN transistor and forcing the transistor into secondarybreakdown. The possibility of this failure occurring increases as thefrequency is increased. The intrinsic body diode of the UniFET hasbeen improved by reducing the Qrr and Trr as shown in Figure 5.The Trr has been reduced to 20%, Irrm has been reduced to 38%and Qrr has been reduced to about 15%. Following are the switchrequirements for this topology:

Low body diode QRR and TRRLow QGD & Low QGD to QGS ratioLow CossHigh Threshold (VTH) voltageLow conduction lossesLow internal gate ESR

Conclusion

A new fast body diode UniFET is presented for the switched modepower supply (SMPS), UPS and solar inverters market. By using thefast body diode UniFET, the power supply efficiency can be improvedor other trade off such as cost reduction, size reduction or increase inpower density as presented here. UniFET reduces both the switching

and conduction losses over regular benchmark MOSFET. The turn-offswitching loss was reduced by over 50% compared to fastest IGBTwithout compromising EMI. The conduction loss was also reduceddrastically compared to regular benchmark MOSFET. When UniFETconducts current in third quadrant most of the current is carried bythe MOS channel and very little current is carried by body diode. Thishelps in reducing the minority carriers in the drift region and henceimproves reliability of operation.

References

[1] H. Aigner, et al., “Improving the Full-Bridge Phase –shift ZVTConverter for Failure-free Operation under Extreme Conditions inWelding and Similar Applications” IEEE proceedings of IAS SocietyAnnual Meeting, St. Louis, 1998.[2] L. Saro, et al., “High-Voltage MOSFET Behavior in Soft-SwitchingConverter: Analysis and Reliability Improvements,” International Tel-communication Conference, San Francisco, 1998.[3] J. A. Sabate, et al., “Design considerations for high-voltage highpower full-bridge zero-voltage-switched PWM converter,” in Proc.IEEE APEC, 1990, pp. 275-284 [4] K. Shenai, et al., “Soft-Switched, Phase-Shifted Topology CutsMOSFET Switching Stress in FBCs,” PCIM Magazine, May 2001.

www.fairchildsemi.com

Figure 5: Reverse recovery waveforms comparing fast body diodeUniFET with competitive devices

di/dt = 250Ausec

FDH45N50F

Competition 1

Competition 2

Competition 1

Body diode reverse recovery

TJ = 250C & Vr = 400V20A

P O W E R S U P P LY

www.bodospower.com32

Laying out a power supply board entails positioning and routing thecomponents in such a way that the high-power signals do not corruptthe low-power signals and cause poor performance. A poor layout willlead to the generation of unwanted voltage and current spikes whichwill cause not only noise to appear on DC voltages in the supply, butalso EMI to radiate to adjacent equipment. Thus proper layout tech-niques are critical to achieving optimal performance of a power sup-ply. This article describes the most important of these techniques.

Placing the Power Components

After importing a power supply schematic into a PCB editing environ-ment, deciding where and how to place and route many discretecomponents on the board can be confusing.Most power supplies are laid out on multi-layer boards with four cop-per layers or more. Most of the board space will be occupied by thepower components: input capacitors, MOSFETs, current sense resis-tors or transformers, rectifiers, inductors, and output capacitors.These components will pass large currents and require thick traces toconnect them together. They should be laid out first.First loops of high di/dt, where large switching currents circulate,should be identified and made as tight and compact as possible tominimize stray inductance that will otherwise lead to the generation ofunwelcome voltage spikes. Figure 1 shows how to identify theseloops. In the figure, the small black arrows indicate how the currentcirculates when the MOSFET is on. The big red arrows indicate thecurrent loop for when the diode is on. All the paths which have eithera black or a red arrow (but not both) are the high di/dt paths.Source currents and their return paths should flow one on top of theother or next to each other to minimize the areas of the loops theyform and reduce the generation of magnetic interference. Inputpower should be taken by the switching circuitry from directly across

the input capacitors. Similarly, the load current should be taken fromdirectly across the output capacitor.Circuit nodes should be sized according to the magnitude and natureof the current that passes through them. High impedance nodes withhigh di/dt, such as the switch node (the junction in many topologieswhere the MOSFET, the rectifier, and the inductor meet) should be assmall as possible while being adequately large for the current flowingthrough them. Minimizing the size of such nodes minimizes the EMIgenerating area. Low impedance quiet nodes, such as ground or theoutput, should be made as large as possible.Copper ThicknessThe traces and copper pours carrying current fromone power component to the next should be made adequately wide. An approximate formula for the minimum trace width required to carrya given current which is accurate over a current range of 1 to 20A is

where T = trace width in mils; I = current in Amperes, and CuWt =copper weight in ounces. The formula assumes that the current caus-es a temperature rise of 10 degrees Centigrade in the traces.Using this formula, the minimum trace width for a current of 1A with 1oz copper is 12 mils; for 5A, 1/2 oz copper it is 240 mils; and for 20A,1/2 oz copper it is 1275 mils. If space allows, and especially whereswitching currents flow, these widths should be increased. Designgoals of 30 mils per amp for 1 oz copper and 60 mils per amp for 1/2oz copper should be striven for. Copper pours or floods should beused to connect the high current paths. Pours on multiple layers con-nected together with vias should be used for currents in excess of10A.

Placing the Analog Components

Analog control components should be routed last because they takeup little space and only need thin traces. One way to organize themis to create component subgroups by function and route the sub-groups. For example, all the components that make up the feedbackcompensation network of the supply can be one subgroup. Thebypass capacitors, soft-start capacitor, and frequency-setting resistorof the PWM controller can make up another subgroup.These subgroups typically connect to the PWM controller (or anotherIC). The subgroups should be placed as close to, and routed asdirectly as possible to the pin they connect to on the IC. This is especially true of decoupling capacitors which must be rightnext to the pin that they decouple. The capacitors must connectdirectly to the pins, and not to any ground or power planes that areelectrically part of the pins.

Best Layout Practices forSwitching Power SuppliesMost power supplies are laid out on multi-layer boards

Typical power supplies consists of a mixture of power components handling switchingvoltages and currents of large value and amplitude, and small signal components han-

dling low-level analog signals, all in close proximity.

By L. Haachitaba Mweene, Applications Manager, National Semiconductor

Figure 1 how to identify first loops of high di/dt.

Tiny 1 MHz Synchronous

Buck Controller Simplifies

Voltage Tracking

MSOP-10 Packaged LM3743 Provides Programmable Tracking, Externally Programmable

Current Limit and Short Circuit Protection Features

For samples, datasheets, online

design tools, and more information

on the LM3743, contact us today at:

power.national.com

Phone: +44 (0) 870 240 21 71

E-mail: [email protected]

© National Semiconductor Corporation, 2006. National Semiconductor, , LLP and WEBENCH are registered trademarks of National Semiconductor Corporation. All rights reserved.

Seeus

atour

boothA

4.506

No

v e mb

e r1

4–

17

LM3743 Features

• Input voltage from 3V to 5.5V

• Externally programmable soft-start with

tracking capability

• ±1.75% reference accuracy over full

temperature and input voltage range

• Lossless programmable low-side current limit

without Rdson sensing

• Fixed high-side sensing for supplemental

short-circuit protection

• Switching frequency option: 300 kHz or 1 MHz

• Output voltage adjustable down to 0.8V

• Available in tiny plastic MSOP-10 packaging

Applications: Ideal for use in communications,

high-end consumer, computing, and industrial

applications.

SS/TRACK

COMP/ENFB

GND

LGATE

ILIMSW

HGATE

BOOT

+

+

VIN

VOUTLM3743

VCC

Z2 (s)

Z1 (s)

LM3743

LM3743

5V

1.8V

VOUT1

VOUT2

5V

1.8V

1.8V

VOUT1

VOUT2

Tracking with equal soft-start Tracking with equal slew rates

V01 = 5V

V02 = 1.8V

VSS

VFB

SS/TRACK

Typical Application Circuit

Tracking Multiple Rails

P O W E R S U P P LY


All the big components in the circuit, such as MOSFETs, rectifiers,electrolytic capacitors, inductors, and connectors should be put onthe top side of the board so they do not fall off during reflow solder-ing. The bottom side of the board should contain only small compo-nents, which can stick to the solder flux on the board by surface ten-sion before they are soldered.

Grounding

When routing the circuitry around the controller IC, the analog smallsignal ground and the power ground for switching currents must bekept separate.It is suggested to isolate the control circuitry on a local ground island,which can then be connected to the rest of the system at only onepoint, preferably at the input capacitor. This stratagem helps to keepthe analog ground quiet. If the creation of a ground island is not pos-sible for all or some of these components, the ground pins of thecomponents can be connected together as a daisychain,but theymust still be connected to the main ground at one point.Components which straddle high impedance and low impedancenodes must be placed close to the high impedance nodes. For exam-ple, resistors setting the output voltage will see a low impedance atthe output and ground connections, and a high impedance wherethey connect to the input of the error amplifier. The resistors must beplaced as near as possible to the error amplifier. To achieve the bestpossible load regulation, a separate trace that carries no load currentmust connect one resistor directly to the load terminal of the supply,and the bottom side of the other resistor must hook directly to thechip analog ground.

Segregating Analog and Switching Signals

Power inductors/transformers, MOSFETs, and rectifiers must beplaced away from the traces and circuitry with low level analog sig-nals to minimize the amount of noise from them that the analog cir-cuitry picks up. If power switching and analog components cannot besegregated due to space constraints, they should be placed on oppo-site sides of a multi-layer board and an inner copper ground planeshould be used to shield the two sets of components from eachother. The ground plane must be connected to the rest of the circuitin such a manner that little or no current flows in it, so that it is elec-trically quiet. Only then can it be considered to be a low noise refer-ence node. All high switching currents should be arranged to flow onwide copper pours on the top layer.For a four layer board the layer stack-up should be as follows: all thepower parts should be on the top layer, as well as the copper shapes

carrying the large switching currents. This layer can also have smallsignal components. The second layer should be a quiet ground planewith no large currents flowing through it. Layer three and the bottomlayer can have a mixture of power and signal traces, with only smallcomponents populating the bottom layer. As much of the board areaspossible on all layers should be flooded with copper, to improve thethermal perform-ance.

Vias

Though it is desir-able to have all thehigh current pathson the top layer, thisis not always possi-ble because ofboard size, routing,and componentplacement con-straints. Vias mustthen be used tomake connectionsbetween layers andto parallel the layers to allow more current to be carried betweencomponents on the board. Multiple vias should be used to connecthigh current paths on different layers. Microvias should be designedto pass a current of 1A each; 14 mil diameter or larger vias shouldpass up to 2A; and 40 mil or larger vias should see no more than 5Aeach. Vias should be allowed to fill with solder to spread heat better,and copper alleyways in the direction of current flow should be leftbetween them.

Example Layout

The schematic in Figure 2 is a dual buck converter based on theLM2717. A printed circuit board for this schematic is shown in Figure3 and incorporates the layout practices recommended in this article.Layer 1 contains all the power parts and thick copper pours to passlarge currents. Layer 2 is a ground plane which is connected to therest of the circuit at only one point near the input so it passes no cur-rent. Layer 3 and the bottom contain signal and power traces. All thecomponents on the bottom layer are small. All the unused board areais flooded with copper.More layout recommendations can be found in the references listedbelow, available on National’s website.

Acknowledgement

The author wishes to thank CraigVarga for reviewing this article andproviding critical background materi-al.

References

1 “SIMPLE SWITCHER PCB LayoutGuidelines,”National Semiconductor ApplicationNote AN1229.2 “Layout Guidelines for SwitchingPower Supplies,”National Semiconductor ApplicationNote AN1149.

www.nationalcom

Figure 3. : A Well-Designed Four LayerBoard for the Buck Schematic in Figure 2

Figure 2 is a dual buck converter based on the LM2717.

��

��

��

��

��

�� !��

��

��"��#��

��$��

�% !�&'(�)*

Integration is the common mantra in digitalIC design. For IC suppliers, getting morefunctionality on board is a way to counter theever declining prices for components. Inanalog IC design, integration has been slow-er due to the need of bulky passives such asinductors and capacitors. For example, thetype of integration that can be found in ana-log power management typically means put-ting together several linear regulators into asingle power management IC (PMIC).Integration is limited by is the maximumamount of heat that can be dissipatedthrough the IC package. This tends to befairly limiting since the efficiency of a linearregulator is the ratio of output voltage overinput voltage. For example, generating a1.2V output voltage from a 3.6V input resultsin a 66 percent conversion loss which is dis-sipated as heat. Assuming a typical maxi-mum power dissipation of a package ofaround 2W, means less than 900mA of totalusable output current can be provided.

On the other hand, DC-to-DC switching reg-ulators provide an efficient means of trans-lating between two DC voltages. For theabove example, they can convert the 3.6Vinput voltage to 1.2V with resulting conver-sion efficiencies of greater 90 percent. This

is of great value in battery-operated circuitsor thermally challenging environments suchas densely packed car entertainment sys-tems.

However, in spite of their higher efficiency,cell phones or car entertainment systemstoday still mainly use the less efficient linearregulators to perform DC-to-DC conversions.

D C / D C C O N V E R T E R S


Eliminating the Inductor inDC/DC Regulators

Reduce the inductor size by switching at faster frequencies

Responding to the demand for smaller inductors in space-sensitive applications, leadinganalog IC suppliers have engaged in a speed race over the past few years, coming out

with switching regulators featuring faster and faster frequencies.

By Ralf Muenster, Director of Marketing, Power Products, Micrel, Inc.

Figure 1. Typical Circuits for Adjustable Output DC-to-DC Switching and Linear Regulators.

Linear RegulatorDC/DC Switching Regulator

VIN SW

GND

FBEN

VIN OUT

GND

FBEN

Linear RegulatorDC/DC Switching Regulator

VIN SW

GND

FBEN

VIN SW

GND

FBEN

VIN OUT

GND

FBEN

VIN OUT

GND

FBEN

Figure 2. Advanced 500mA LDO (MIC5319) Versus 500mA 1MHz DC-to-DC Buck RegulatorLayout

This is due to the fact that linear regulators,unlike switching regulators, do not require abulky inductor to facilitate the voltage con-version. Figure 1 gives a comparisonbetween the circuitry of a linear regulatorand a buck switching regulator.

Eliminating the inductor and still achievinghigh conversion efficiencies has been a pri-ority for many power designers. Oneapproach has been to use charge pumps.However, these only work well for simplydoubling or halving voltages at small cur-rents. A more promising approachresearchers have been deploying is toreduce the inductor size by switching atfaster frequencies. Not too long ago, switch-

ing regulators all operated with frequenciesaround 100 kHz. The inductor size of aswitching regulator is inversely proportionalto its operating frequency. Buck regulators,operating at 100 kHz, commonly use induc-tors in the range of 47µH. In comparison,switching regulators, operating at 1 MHz,just need 4.7µH of inductance. Assumingeverything else is equal, this solutionreduces the volume of the required inductorby 90 percent. A low profile 4.7µH inductorthat is capable of continuously delivering 500mA, typically comes in a footprint of around4mm x 4mm and a profile height of around2mm. Adding to the footprint of a switchingregulator IC, which typically comes in a 3mmx 3mm x 1mm package, one can visualize

that this still is a much larger solution com-pared to an advanced 500 mA linear regula-tor such as Micrel’s tiny 2mm x 2mm x0.9mm MIC5319 (refer to Figure 2).

Responding to the demand for smallerinductors in space-sensitive applications,leading analog IC suppliers have engaged ina speed race over the past few years, com-ing out with switching regulators featuringfaster and faster frequencies. TI announceda 3MHz device in 2004. This was followedby Linear Technology and Maxim announc-ing 4MHz capable devices last year. At 3-4MHz, the inductor value can shrink down to1µH thereby reducing the inductor size toaround 3x3x1.2mm.



Figure 3. Solution Size and Height Comparison with Increasing Switching Frequency.


In 2006, Micrel broke the 1µH barrier for500mA switching regulators by offering theindustry’s first 8MHz buck regulator. Thissolution features a tiny 0.47uH chip inductorthat measures a mere 1.25mm x 2mm x0.55 mm; another 95 percent reduction ininductor volume compared to a 1MHz solu-tion (refer to Figure 3). As shown in theillustration, this is also the first time theinductor actually became smaller than theswitching regulator IC package itself. Now,the inductor has shrunk 200 times in volumecompared to the 100 kHz converter.

In addition, Micrel has recently started sam-pling a completely inductorless high efficien-cy buck switching regulator. The device, theMIC3385, is rated for up to 500 mA andcomes in tiny 3mm x 3.5mm x 0.9mm MLFpackage (Figure 4). The efficiency isrespectable with up to 90 percent. Figure 5shows the efficiency curve over load currentfor a 3.6-to-1.8V conversion. Note that a lin-ear regulator would feature just 50 percentconversion efficiency in this case. Thisdevice features a light load mode with only20uA of ground current. Also, output noiseand transient performance are excellent

thanks to a fixed-mode frequencyoperation supportedwith a light loadLDO that assists theswitcher duringdemanding loadtransients. Figure 6shows the MIC3385noise and transientperformance incomparison to anindustry standardbuck regulator witha PFM light loadscheme. As can beseen at loads of<1mA, the industrystandard part pro-duces up to 150mVnoise on the outputwhile the noise on the MIC3385 is virtuallynot measurable. At loads of about 30mA,the part with the PFM mode generates a dif-ferent band of noise with about 190mV ofpeak-to-peak deviation while the MIC3385 isstill quiet. The load transition from PFM-to-PWM mode finally is where the largest devi-ation in output voltage is seen on the indus-try standard part. In the new MIC3385, theLDO mode supports the switcher in this tran-sition and the overall output voltage droop isheld under control. This allows for excellentoverall transient and noise performanceusing small output capacitors thereby furtherintegrating the solution. The industry stan-dard part would need 5 times the outputcapacitance to achieve the same perform-ance.

Conclusion

Ever increasing functionality in mobilephones and infotainment systems continuesto demand ever smaller and more integratedanalog and digital IC solutions. This combi-nation makes it increasingly important toswitch from inefficient linear regulators tohigh efficiency, small solution size DC-to-DCswitching regulator solutions. Micrel hasdemonstrated the viability of a breakthroughinductorless 500 mA switching regulator thatalso allows shrinking the output capacitorswhile maintaining a remarkable efficiency,low noise and exceptional transient perform-ance.

www.micrel.com

Figure 5. Micrel’s MIC3385 Efficiency For 3.6V-to-1.8V ConversionOver Load Current.

Figure 6. Comparison of Load Profiles for an Industry Standard Buck Regulator Using a PFM Light Load Scheme Compared to The NewInductorless MIC3385 Hybrid LDO/DC-DC Converter Offering A LDO Light Load Mode Exhibiting a Stable Output Voltage as Load ProfileChanges.


Figure 4. Micrel MIC3385, Inductorless500mA High Efficiency Buck Regulator, in a3mm x 3.5mm x 1mm MLF Package.

Globalelectronics

March 21–23, 2007Shanghai New International Expo Centre SNIEC

www.global-electronics.net

Contact:Messe München GmbHTel. (+49 89) 9 49-2 03 25Fax (+49 89) 9 49-97 2 03 [email protected]

electronica &ProductronicaChina 2007With PCIM China

Featuring LASER. World of Photonics China

Collocated with Semicon China and CPCA Show

2007

China’s platform for the electronicscommunityComponents and Assemblies & Production Technologies

High-quality displays require precise controlof LED brightness and color. Unfortunately,the inherent variations in LED characteristicsforce video display manufacturers to imple-ment processes and procedures to compen-sate for these variations. This adds cost tothe system. High-end LED driver ICs imple-ment a simple, software-programmable tech-nique called “dot correction” that helps man-ufacturers to achieve high-quality productswhile still keeping system level productioncosts in check. This paper shows how LED-to-LED brightness variations affect video dis-play manufacturers’ costs, how dot correc-tion compensates for LED brightness varia-tions while helping todecrease production costs,and provides a simple dotcorrection example.

The System

Large form factor LEDvideo displays integratehundreds of individual dis-play panels and hundredsof thousands of LEDs tocreate a seamless largescreen video image. Figure1 shows a typical display.Each panel contains a 16 x16 pixel matrix. Each indi-vidual pixel contains multi-ple LEDs, usually two red,one green, and one blueLED.

LED-to-LED Brightness Variations

The key to a high-quality display is precisecontrol of LED brightness. To faithfully recon-struct the original video image, a video dis-play must accurately control LED brightnessto a fraction of a percent. Unfortunately,LED brightness varies from LED to LED,even within a single manufacturer’s produc-tion lot. This “LED production variation”places limitations on a display’s picture qual-ity. The root cause of LED brightness varia-tions stems from the fact that all semicon-ductor-manufacturing processes containinherent and uncontrolled variations. Thesemanufacturing variations result in uncon-

trolled LED brightness variations. Typical off-the-shelf LED products may have a +/- 60percent variation in lumen output for thesame forward current. This excessive varia-tion is not acceptable for higher end videodisplays or in many other applications. LEDmanufacturers use a method called “binning”to help reduce the lumen variation. Binningis a process that separates LEDs into sepa-rate bins or lots of similar brightness. Thecustomer can then buy specific brightnessbins to get LEDs that are similar in bright-ness. For example, the datasheet for a spe-cific LED part number from company Xshows a lumen variation between 71 and

280 lumens (+/- 60 percent) fortwo LEDs with the same forwardcurrent. The manufacturer binsthis LED into six smaller groupsto significantly reduce LED-to-LED variations within eachgroup. Even within these small-er groups, LEDs have significantvariation. The highest brightnessbin still varies +/- 11 percent(224 to 280 lumens). Althoughbinning provides the end userwith higher quality LEDs, itcomes at the expense of highercost. Since the LED manufac-turer cannot control the percent-age of LEDs that end up in spe-cific bins, they incur an effectivereduction in yield when trying tofill specific bins to support cus-tomer orders. Lower yield drivesup manufacturing costs. Testcosts also increase since themanufacturer must individuallytest each LED. Manufacturersalways pass these costs on tothe customer.

P O W E R M A N A G E M E N T


Large Form Factor VideoDisplays

Dot correction reduces system cost

The LED lighting market is growing rapidly with applications ranging from alphanumericinformational displays to high quality, large form-factor video displays. These markets

suffer from two opposing market forces: pressure for higher quality displays, and pressurefor lower cost solutions.

By Michael Day, Power Management Application Supervisor for Portable Power, Texas Instruments

Figure 1. Progression fromindividual pixels to largeform factor LED video dis-play.

M I C R O T E C H N O L O G Y

The new Apex SA56 is the industry’s first 5A monolithic PWM amplifier. With this level of power, combined with 60V operation, designers can now control motors of 1/3HP or more on the same 48V bus. This single chip device allows system level designers to integrate sophisticated motion control into their designs without being experienced analog or power designers. The SA56 is compatible with TTL and CMOS inputs for digital motor control. For analog systems, an onboard PWM oscillator and comparator make it possible for the SA56 to convert analog signals to PWM direction and magnitude.

Industry’s first 5A monolithic PWM amplifier works with digital or analog control to drive bi-directional DC motors up to 1/3HP on 55V supplies.

Drive Larger MotorsWith More Power,Fewer Components

SA56 Digital Mode

SA56 Analog Mode

SA56Operational Control DSP, MCU or AnalogOutput Full BridgeSupply Voltage 12V to 60V Single SupplyOutput Current 5A Continuous, 10A PEAKPower Delivery Up to 250WSwitching Frequency 100KHzProduction Volume Pricing USD $8.90

Visit us online today atwww.apexmicrotech.com/BPS

• Data sheets• Design tools• Application notes• Spice models• Eval kits• Request v12 data book

LED Aging

LEDs age with time and their light outputdecreases. Typical aging specificationsdefine an LED’s end-of-life when the lumenoutput at a specified current drops to half ofits initial value. A typical LED’s lifetimeranges from 50,000 to 100,000 hours,depending on the LED and its operating con-ditions. While LED-to-LED brightness varia-tions affect panel makers at production, LEDaging affects them when the time comes toservice existing panels. A typical video appli-cation may have hundreds or thousands ofindividual panels. If one of these panelsbecomes defective or damaged, it must bereplaced. The new panel will be brighter thanthe rest of the display because the newerLEDs have not aged. The manufacturermust calibrate the new panel to match thebrightness of the aged display. This calibra-tion adds cost to the system, especiallywhen the calibration involves changes inhardware. Fortunately, higher-end LED driv-er ICs provide the panel maker with a simpleto use method for managing differences inLED brightness and aging.

Dot Correction

Dot correction helps manage variations inLED brightness by adjusting the current sup-plied through each individual LED in thearray. When properly implemented, dot cor-rection provides a uniform brightness acrossa panel, even with unmatched LEDs. Its abil-ity to match LED brightness provides signifi-cant improvements in display quality. Theconcept of dot correction is simple.Manufacturers measure the brightness ofindividual LEDs at full-rated current. Thedimmest LED in the system is designated asthe “base” LED to which every other pixel ismatched. They then program a dot correc-tion value for each LED that appropriatelyscales the forward current to set all LEDs tothe same brightness. Dot correction can beimplemented at the hardware or softwarelevel. Devices like the Texas InstrumentsTLC5940 implement dot correction at thehardware level, but with software control.This method provides significant benefitsover other purely hardware or purely soft-ware implementations. The TLC5940’s dotcorrection values are calculated and thenprogrammed into the IC using a softwareinterface. The TLC5940 controls the dot cor-rected LED current at the hardware level.The hardware-software method is superior toa purely hardware method because it elimi-nates the need for expensive and time-con-suming hardware changes such as compo-nent value changes in the production flow. Itis also superior to a purely software method,since the TLC5940 controls the dot correct-

ed LED currents during a panel’s operation.As a result, the system or board level micro-processor is free to perform other tasks. TheTLC5940 also eliminates the need forexpensive look-up tables and complex soft-ware multiplication routines for each LED inevery video refresh cycle.

Implementing Dot Correction in

Production

The mechanics to implement dot correctionare relatively easy and are best explainedwith a simple example. If a panel’s specifica-tion requires its green LEDs to have a maxi-mum luminosity of 80 mcd (millicandela), themanufacturer must set the LED driver’s max-imum current to produce at least 80 mcd inall LEDs. This current setting must take LEDbrightness variations into account. After eachindividual panel is built, the microprocessorprograms all LEDs to their maximum bright-

ness. The manufacturer measures eachLED’s brightness, calculates the appropriatedot correction values, and programs theTLC5940 with the new dot correction values.This process can be repeated until the com-mand to turn all LEDs on at maximum bright-ness results in uniform brightness across thepanel. If an Osram LP E675 LED is used,43mA of forward current guarantees thedimmest LED produces at least 80 mcd.During a panel’s production test and calibra-tion, the brightness of all LEDs is measuredat the driver’s full-programmed current of43mA, which is set by an external resistor. Atypical pre-dot corrected LED luminousintensity histogram for sixteen LEDs mightresemble that in Figure 2. The data in theforeground is the LED current in mA, whilethe data in the background is the LED bright-ness in mcd. LED brightness variations areevident without dot correction, which may beunacceptable in higher-end displays. Themanufacturer can now use the pre-dot cor-rection information to calibrate the LED

brightness. When programmed to full bright-ness, the IC must dot correct the luminousintensity of LED1 from 83 mcd to 80 mcd.The TLC5940 has six-bit dot correction (63steps), which corresponds to a full-scale res-olution of 1.59 percent per step.

The following formula calculates the correctdot correction level for each LED:

DCProduction = LDesired / LInitial * 63

where DCproduction is the calculated dot cor-rection value at production, Lbaseline is thedesired brightness level, and Linitial is the ini-tial measured brightness at the referencecurrent. Since dot correction values arewhole numbers, the equation’s result mustbe rounded up or down.

After each LED’s DOT correction value is

calculated and stored, the TLC5940 auto-matically generates a uniform brightness inall LEDs. Since dot correction is controlledby the hardware system, the processor isnot concerned with variations in LED bright-ness. When the processor commands fullbrightness, the TLC5940 accepts the bright-ness command and automatically scales thecurrent in each LED to provide full bright-ness. Figure 3 shows the LED currents andresulting brightness after dot correction isapplied.

Implementing Dot Correction in the Field

Dot correction also simplifies calibration ofreplacement panels for existing video dis-plays. A replacement panel for an agedvideo display is already factory dot-correctedto ensure uniform brightness. However, thereplacement panel’s brightness varies fromthe existing panels’ brightness. The manu-facturer can calibrate the replacementpanel’s brightness much the same way theycalibrated individual pixels on the panel.



Figure 2 - LED brightness and toward current histogram before dot correction

After installing the replacement panel, the manufacturer measuresboth the new panel’s brightness and the older panel’s brightness.They then calculate a “replacement panel adjustment factor” usingthe equation below.

KRPAF = LumensOriginal_Panel / LumensReplacement_Panel

Where KRPAF is the replacement panel adjustment factor,LumensOriginal_Panel is the brightness of the original panels in the

video display, and LumensReplacement_Panel is the brightness of thereplacement panel.KRPAF scales each LEDs dot correction value to provide a uniformdimming for the replacement panel. The equation below calculatesthe new dot correction value for each individual LED.

DCReplacent_panel_LED = DCProduction* KRPAF

where DCReplacent_panel_LED is the new dot correction value for anindividual LED in the replacement panel, DCProduction is the originaldot correction value that was programmed in production at the facto-ry, and KRPAF is the replacement panel adjustment factor.This replacement panel calibration method results in an overall videodisplay with a uniform brightness. The calibration process is soft-ware-based, so it can be easily changed. This saves the additionaltime and expense required for hardware-only based calibration rou-tines.

Conclusion

Dot correction easily compensates for inherent variations in LEDbrightness, allowing display manufacturers to produce high-qualityvideo displays while still keeping costs down. The software-hardwarebased TLC5940 dot correction provides distinct advantages overother methods. The hardware portion allows cheaper, higher perform-ance systems by performing dot correction at the hardware level.

www.ti.com


Figure 3 – LED brightness and forward current histogram after dotcorrection

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 160.0

20.0

40.0

60.0

80.0

100.0

Luminosity After Dot Correction

componnts.systms.applications.automotiv.wirlss.mms.mbddd.softwar.

Sav tim whn you rgistr onlin

and mony whn you purchas your tickt onlin!

www.lctronica.d/tickt



With increasing interest in the development of electric traction andhybrid-electric vehicles, the demand for comprehensive systemsdesign and analysis to aid the development process is rising. Due toa current globalization environment the decentralized design processbrought a real challenge to the automotive engineering design com-munity. The particular stages of the product design as the componenttesting and the system/sub-system integration became more andmore difficult to be managed. Consequently, a uniform and easy plat-form of emulating those stages became a must.

With the high degree of complexity and the wide variety of domainsinvolved in this kind of design, there is a high demand for simulationand analysis tools. Experimental setups are too expensive and fre-quently only deliver a snapshot of the actual configuration. The explo-ration of design variations requires repetitive experiments that areeconomically questionable.

Ansoft’s unique solution for electrical machines and drives is charac-terized by the flexibility it offers its users. Many phases of the designprocess can be investigated with respect to one another to insureproper interoperability among domains. Designers, components engi-neers, analysts, and systems engineers no longer are limited in theirinteraction by stand alone software packages.

This work illustrates a simulation-based design environment capableof simulating the overall system behavior in connection with adetailed electrical model. The design environment combines eitherthe RMxprt solution, an electric machine-specific tool equiped withanalytical-solver that automates the model creation process for theuser, or Maxwell solution, an electromagnetic finite element solverthat includes static, harmonic, and transient analysis under a com-mon Simplorer environemnt, a sophisticated design tool offering elec-trical, electromechanical, and control simulation capabilities. InSimplorer the virtual prototyping is using VHDL-AMS language tomodel and to implement the control scheme model. The VHDL-AMSis a superset IEEE Std. 1076-1993, and is designed for the descrip-tion and simulation of multi-disciplinary mixed-signal systems.

Automotive engineers can now predict efficiency and fuel economy ofvehicle designs, power management strategies, and the electricaldesign in one integrated design environment. With Simplorer, engi-neers can simulate complete drivetrains by combining fuel cells, bat-teries, power electronics, and electric motors in one model.

Software Packages Description

RMxprt is a versatile software program that speeds the design andoptimization process of rotating electric machines. RMxprt is theentry point for the Ansoft automated motor and drive design flow.RMxprt automatically produces both system-level models and geo-metric data, allowing the preliminary design to be refined and inte-grated with power electronic and control circuitry.

Maxwell2D/3D is the leading electromagnetic design software for thesimulation and analysis of high-performance electromagnetic andelectromechanical components common to automotive, military/aero-space, and industrial applications. Maxwell2D/3D provides users avirtual laboratory on their desktop to study static, frequency-domain,and time-varying electromagnetic fields in complex structures.Maxwell2D/3D features a powerful Microsoft Windows®-based userinterface that provides a highly integrated architecture, unmatchedautomation, and productivity enhancements.

Simplorer is multi-domain, system simulation software for the designof high-performance electromechanical systems commonly found inthe automotive, aerospace/defense, and industrial automation indus-tries. With a wide range of modeling techniques, statistical analysiscapability and adherence to IEEE standards, Simplorer greatlyreduces engineering time and prototype iterations while improvingdesign performance of electrical, mechatronic, power-electronic, andelectromechanical systems.

Powertrain - General Description

An auto’s drivetrain or powertrain consists of all the components thatgenerate power and deliver it to the road surface. As seen in Figure1, this includes the engine, transmission, energy storage, electricalmachine, motor controller, driveshafts, differentials, and the final drive

Efficient Virtual Prototyping ofElectric Drive Systems

Improving electrical, mechatronic, power-electronic, and electromechanical systems.

Automotive engineers can now predict efficiency and fuel economy of vehicle designs,power management strategies, and the electrical design in one integrated design environ-

ment. With Simplorer, engineers can simulate complete drivetrains by combining fuelcells, batteries, power electronics, and electric motors in one model.

Marius Rosu, Ph.D., Group Leader Simplorer Modeling, AnsoftKoichi Shigematsu, Ph.D., Application Engineer, AnsoftThomas Liratsch, Country Manager, Germany, Ansoft

(drive wheels). Sometimes “powertrain” is used to refer to simply theengine and transmission, including the other components only if theyare integral to the transmission. A vehicle’s driveline consists of theparts of the drivetrain excluding the engine and transmission. It is theportion of a vehicle, after the transmission, that changes dependingon whether a vehicle is front wheel drive, four wheel drive, or rearwheel drive.

Mechanically actuated automotive systems, such as power steering,

fuel and water pumps and even the traction system are beingreplaced with systems that utilize electric motor technology.Consequently, the presence of semiconductors in automobiles isincreasing geometrically, in part due to automotive systems develop-ers’ desire for electronic motor control to address consumer require-ments for safer, more efficient cars.

One major contingent of the electrical machines group, brushlesstype motors, are ideal for automotive applications that continuouslyrotate, such as pumps, cooling fans, and stepper motors. Brushlessmotors equip positioning systems with start and stop functions wherehigh reliability is required. Furthermore, the electronic control thatbrushless motors offer is critical to statutory vehicle requirements,including energy-savings, reduced environmental and emissionsimpact, and the creation of safer vehicles. Brushless motors are alsouseful for variable-speed applications where space is tight, as in fuelpump control and electronic/electric power steering. In these types ofapplications, electronic control is essential because of the need forfault diagnostics and wide temperature and voltage operationalranges.

The embedded processor is a critical tool for automotive systemdesigners as they address the increasing needs and demands oftoday’s driver. The increased use of electronic controls enables auto-motive system designers to meet these needs, while meeting theirown requirements to develop low-cost, low-noise, high-accuracy sys-tems with faster time-to-market. A multitude of embedded-processorsolutions is available to automotive design engineers.


Figure 1. Hybrid Electric Vehicle use advanced electric motors andcontrollers within the transmission

components.systems.applications.automotive.wireless.mems.embedded.software.

Experience the world of electronics in all of itsdiversity: innovative components, software, systemsand applications for future progress. Take advan-tage of your visit to the world’s leading inter-national trade fair to enter into a direct dialog withkey industry suppliers from around the world. Additional information? www.electronica.de/2006

Get the whole picture.

www.global-electronics.net

electronica 2006components systems applications

22nd International Trade FairNew Munich Trade Fair CentreNov. 14–17, 2006

Save time when you register online

and money when you purchase your ticket online!

www.electronica.de/ticket

More Electric Drive System Topology

An example of electric drive system architec-ture is shown in Figure 2. This topology caneasily accommodate under the same simula-tion environment both the system levelanalysis and the very detailed study takinginto consideration different behaviouralaspects of each component.

As seen in Figure 3, either RMxprt design orMaxwell solution can be employed inSimplorer environment to define the rotation-al actuator behaviour in such system. Themain advantage of using this strategy is tocount accurately the electrical machine non-linearities and the end-effects on the systemconfiguration when analog/digital controlparameters design is involved.

For the current application the technology ofbrushless type motor is employed. Thebrushless motor does not use brushes forcommutation. Instead, the brushless motorcontains a rotor with permanent magnetsand a stator with windings, and performscommutation electronically. Commutation isthe act of changing the motor phase currentsat the appropriate times to produce rotationaltorque.

Stacked steel laminations on the stator equipthe brushless motor with windings placed inslots, which are cut axially along the innerperiphery. While the stator, in general, issimilar to that of an induction motor, themotor windings can be configured in a non-distributed format. Each winding is construct-ed of numerous small coils that are placed in

the slots and interconnect to form the largerwinding. Each winding is distributed over thestator periphery to form an even number ofpoles. Stator windings can be either trape-zoidal or sinusoidal, with each generatingdifferent types of back electromotive force(EMF). The phase current also has trape-zoidal or sinusoidal variations.

All rotors have permanent magnets of sometype, and can vary from two to several polepairs. The proper magnetic material withwhich to create the rotor is selected basedupon the required magnetic field density.Ferrite magnets have traditionally been usedto make permanent magnets. However, rareearth alloy magnets are becoming more pop-ular, as they generally have a higher mag-netic density per volume and enable the

rotor to compress further for the sametorque. Alloy magnets improve the size-to-weight ratio and deliver higher torque than amotor of the same size that is comprised offerrite magnets. A method to detect the posi-tion of the rotor magnets is required for suchmotors.

Brushless type motors are popular becausethey are fast, noiseless, efficient, and exhibita longer operating life. Brushless motors arealso popular due to their compact size, con-trollability, high efficiency, low EMI(Electromagnetic Inference) and high-reliabil-ity. Their compact size is a direct result oftechnological advances in magnets notedearlier that deliver efficiency improvements.Additionally, the ratio of torque delivered inbrushless type motors relative to motor sizeis higher than in brush type motors, makingbrushless motors an excellent match forspace and weight-sensitive applications.

Brushless type motors can be designed intosystems that are sensor-based or sensor-less. The implementation of sensorlessmotor systems eliminates the cost of HallEffect or optical sensors and their supportingelectronics. The sensorless operation is alsodesirable if the rotor is operating whileimmersed in fluid such as fuel, oil, or water.In sensorless control, back EMF zero cross-ing is used for commutation.

Since AC variable speed drives started togain popularity in the 1970s, their basicmode of operation has hardly changed. Theincoming AC supply is rectified by a diodebridge to produce DC, and the DC thenfeeds an output bridge that, in turn, pro-duces an AC output to drive the motor. Thefrequency and voltage of the drive’s AC out-put are continuously adjusted by a controlloop so that the motor delivers the requiredspeed and torque.

Almost invariant to the arrangement of thecomponents, the electric drive system has anumber of shortcomings. For example, recti-fier systems are notorious for feeding highlevels of harmonics back into the supply, andthey also have a poor power factor. In addi-tion, the so-called DC link between the recti-fier and the output bridge requires the use oflarge electrolytic capacitors. These not onlyoccupy a lot of space, they are also thecomponents most likely to limit the useful lifeof the drive, since they lose capacity overtime, especially when operated at high ambi-ent temperatures.

Finally, the link between the incoming supplyand the motor in this conventional design is



Figure 3. Ansoft’s RMxprt allows users to evaluate machine performance using advancedanalytical algorithms and creates a finite element model for detailed transient evaluation.

Figure 2. Electric Drive System Topology

Power Supply Inverter Actuator Mechanic Load

Analog Control Digital Control

www.e-emc.com

Face-to-face communication at its best –plan your participation today!

EMC

EMV 2007Stuttgart

International Exhibitionwith Workshopson ElectromagneticCompatibility (EMC)6-8 March 2007Messe Stuttgart

wwwe-emv.com

EMC EMC EMC EMCOrganizer:

Mesago Messe Frankfurt GmbH

Rotebuehlstraße 83-85

70178 Stuttgart / Germany

Phone +49 711 61946-0

Fax +49 711 61946-93

[email protected]

Including:y Measurement & Testing

y Filters and

Filtering Components

y Lightning and

Overvoltage Protection

y Shielding

y ESD-Protection

y Services and others

For complexchallengesin all aspectsof ElectromagneticCompatibility –the answer isEMV 2007 Stuttgart!

Europe’s leadingEMCevent!

irreversible - there is only the path availablefor energy from the motor to be fed back tothe supply with a special converter - which isvery expensive. In general during braking,not only is energy wasted, it also has to bedissipated, usually in a bulky, heat-producingbraking resistor.

The presented power electronic design isrelated to the implementation of a modularmatrix converter topology into the electro-mechanical system as illustrated in Figure 4.The three phases of the drive’s incomingsupply are connected directly to the motor,via a matrix of bi-directional semiconductorswitches. By correctly sequencing the opera-tion of these switches, the voltage and fre-quency of the output to the motor can beprecisely controlled.

The benefits of the technology include mini-mal generation of harmonics, regenerativebraking without the need for additional resis-tors, automatic return of energy to the supplyduring braking and enhanced operating life.Matrix-converter drives offer efficient opera-tion in all four quadrants. Ordinary drivesconvert the incoming AC supply to DC, andthen convert the DC back to AC at therequired frequency and voltage. In contrast,the matrix converter drives use an array ofsemiconductor switches - packaged insulat-ed gate bipolar transistors (IGBTs) - to con-nect the three phases of the supply directlyto the motor in a precisely timed sequence.This method of operation reduces harmonicgeneration to approximately 8% of the level

associated with conventional drives, whilealso offering almost unity power factor.

The voltage converters are based on pulse-width modulation (PWM) of their AC outputvoltages, which is performed through count-less switching operations running at high fre-quencies. Modern semiconductor powerdevices (GTO thyristors, MC thyristors, IGBTtransistors) are nowadays used for this pur-pose. The switching operations are con-trolled through sophisticated algorithmswhich at present are based almost exclu-sively on microprocessors (for instance Intel87C196MC, Siemens SAB 80C166), signalprocessors (Texas Instruments TMS320C40,Motorola DSP 96002) and application ofspecified integrated circuits (ASIC, FPGA).

The Simplorer environment provides with avery efficient way to implement this digitalcontrol strategy by using advanced hardwaredescription language as VHDL-AMS as seenin Figure 5. Immediate advantages of usingthis language for such system designs aredue to its features in supporting analog, digi-tal, and mixed signal modeling, offering solu-tions in different design domains at differentlevels of abstraction, co-simulating with spe-cialized simulation packages and proprietarycode, linking to Maxwell for high-fidelity mod-eling of electromechanical components.

Increasing demands laid upon accuracy, sta-bility and efficiency of electric drives opera-tion require careful selection and optimiza-tion of criteria for the PWM strategies. This



Figure 5. Analog and Digital system control can be defined using VHDL-AMS within the sameschematic.

Figure 4. Detailed schematic view of the hybrid-electric power system. Users can analyze the mechanics, electromechanical device and powerelectronics operation and interaction within a single view.

Mechanicalsystem

PM Motor

Power Supply

is a complicated multidimensional problemespecially in the case of feedback modula-tions. The simulation results of entire analog-digital system are presented in the Figure 6-7. Figure 6 presents the line permanentmotor currents (left) and the matrix converterinput currents (right) since the Figure 7shows the digital control signals representingthe FPGA response which are further con-verted into the IGBTs’ gates switching puls-es.

www.ansoft.com


Figure 7. MxC digital control signals

t0 32 005 00 10 00 15 00 20 00 25 00

MAKING MODERN LIVING POSSIBLE

Simplify your design!

Danfoss Silicon Power GmbH • Heinrich-Hertz-Straße 2 • D-24837 Schleswig, Germany • Tel.: +49 4621 9512-0 • Fax: +49 4621 9512-10E-mail: [email protected] • http://siliconpower.danfoss.com

E0 to E3 size Short and long pins Flexible pin-outIGBT’s and MOSFET’s from world class manufacturers Low and high voltage

For industry, transportation and automotiveWe design and manufacture to your needs.

9936

Figure 6. PM Motor Load Currents and the MxC Input currents

T R A N S I S T O R S


For the power switching element within such loads, these technicaldemands appear to manifest themselves as simply ‘increasedswitched power density!’ – in practice how can this be bestachieved?To some extent the demands listed above are contradictory. Forinstance, improving efficiency by increasing the power switching chipsize can reduce conduction losses and improve reliability throughreduced operating temperature, but this is at the expense ofincreased cost and a larger package size. In order to satisfy the demands for increased switched power densitywe need to consider all aspects of power device manufacture, name-ly the design and construction of both chip and package. When con-sidering breakdown voltages of up to 500V there are three potentialtechnologies to choose from: BJT, MOSFET, or IGBT.

IGBT

Taking these in reverse order, IGBTs combine bipolar and MOSFETphysics to create very efficient high power devices in larger pack-ages. Unfortunately the forward voltage drop of the additional emitterjunction adds to the on-voltage such that even in a package such asDPAK the typical on-voltages achievable are greater than 1.8V, andmay reach 2.8V. While the additional off-set voltage may be feasiblefor large current devices, in heat-sinkable packages with thermalresistance values down to below 1°C/W, it significantly increases theeffective on-resistance for low current devices in small surface mountpackaging with typical thermal resistance values of 60 – 200 °C/Wlimits.

MOSFET

MOSFETs are certainly ubiquitous and have established a dominantposition particularly in low voltage applications such as batteryportable load switches due to the development of modern trenchMOSFET processes. As such, they may appear at first sight to bethe natural choice in the 200V to 500V range. Alas the reality is thatthey exhibit poorer area-specific on-resistance than bipolar transis-tors as breakdown voltage capability increases. This is because thecontribution to resistance of the drift drain region increases rapidlywith voltage, as resistivity and thickness are increased to support thevoltage.

BJT

While the same must surely be true for bipolar junction transistors,the effect is mitigated by the bipolar phenomenon of conductivitymodulation, whereby operation in the saturation region causes the

injection of minority carriers into the collector region, resulting in acommensurate injection of majority carriers to preserve charge neu-trality. The effect of this increase in charge carrier density is to dra-matically reduce the resistance of the lightly doped collector regionduring the on-state, thereby reducing the area-specific on-resistance.The significant advantage of conductivity modulation does not how-ever automatically lead to superior power density in bipolar transis-tors.To deliver class leading performance a bipolar transistor must also becarefully designed to minimise series resistance and to ensure evencurrent density across its active area. This must be combined with acompact chip termination, the function of which is to terminate thecollector-base junction at the chip periphery whilst avoiding electricfield crowding that will result in premature or unreliable breakdown.For low-voltage devices the termination structure can be simple butfor higher voltage devices designed to fit in small footprint packagesit takes up a significant area. Considerable effort therefore needs tobe expended to achieve compact termination structures that max-imise the active area of a chip thereby allowing high power densities.

However it is only when an optimised chip design is put together witha package that’s engineered to maximise the chip area to packagefootprint ratio and achieve reduced thermal resistance, that the goalof high power density is realised. Careful internal design and selec-tion of materials is required to ensure that package robustness is notcompromised.

Latest bipolar developments

Consider for example the latest 400V NPN transistor from ZetexSemiconductors. The ZXTN08400BFF is capable of switching500mA continuously, 1 amp peak, within a SOT23 flat package lessthan 1mm high and with a footprint of just 7mm2. This equates to200W of switched power and a maximum 28.5W/mm2 switchedpower density. The effective RCE(SAT) is down to 350mOhm, a per-formance only previously available in the SOT223 package having 6times the footprint and one-third of the power density.Compared to the ZXTN08400BFF, a typical MOSFET of an equiva-lent voltage rating can achieve continuous current ratings of around3A in a DPAK package with a footprint of 64.3mm2, equating to amaximum switched power density of 19.3W/mm2 - only two-thirdsthat of the bipolar transistor. Despite the larger package and chip theMOSFET RDS(ON) is typically more than 4 times the minimumRCE(SAT) of the bipolar transistor.

Optimising the Power Switch inHigh Voltage Applications

High voltage bipolar transistors in SMT packageSmall high voltage loads are today found in abundance. Be they actuators, motors,

solenoids or transformers, power supply or power conversion circuits, all are subject to therelentless quest for better energy efficiency, improved reliability and reduced cost and footprint.

By Peter Blair, Discrete Product Development Manager, Zetex Semiconductors

Other selection criteria

Now clearly there are other selection criteria that come to bearbesides power density in the selection of a switching device for agiven application, and the illustration above does not tell the wholestory. A designer must consider carefully others factors such asdrive current capability, switching speed, nature of load, and ambienttemperature. A particular issue with the use of bipolar transistors inhigh voltage circuits is the avoidance of simultaneous high voltagesand high currents that may lead to secondary breakdown failure.Unclamped inductive switching is such an application that wouldrequire careful analysis. However, high voltage bipolar transistors invarious small SMT package styles are available to suit a number ofdemanding applications, some of which are briefly outlined here.Modern fax, modem and voice circuits require Data AccessArrangement (DAA) chip sets in order to interface with the PublicSwitched Telephone Network (PSTN). High voltage discrete transis-tors with breakdown voltages appropriate to the various nationalstandards are used to connect the DAA chips to the telephone lines.Figure 1 shows a DDA chipset interfaced to the telephone line usingbipolar transistors, Q1, Q2 and Q3. Typically 300V - 400V transis-tors are specified in SOT23, SOT89 and SOT223 outlines.

HID car headlamps require electronic ballasts that drive the lamp viaa transformer (Figure 2). During start-up the striking voltage mayexceed 20 kilovolts, which in turn requires the ballast and driving cir-cuitry to withstand 350V. Typically 400V NPN and PNP bipolar tran-sistors (Q5, Q6, Q7,Q8,Q9,Q10) are specified in the bridge high-sidedriver stage.

Piezo diesel injection valves are replacing traditional solenoid injec-tors driven by fuel emissions and efficiency considerations. Figure 3shows a typical circuit using 300V - 400V bipolar transistors to switchthe high-voltage supply to the piezo elements.

In conclusion, bipolar technology provides designers with the readymeans to achieve increased switched power density goals in highvoltage load driving circuits. The advantages the latest bipolar tran-sistors offer over MOSFET or IGBT alternatives, in terms of highbreakdown voltage, switched currents up to an amp and the smallestpossible footprint make them the power switch of choice in a widerange of demanding applications.

www.zetex.com

T R A N S I S T O R S


Figure 1: DDA chipset interfaced to the telephone line using bipolartransistors.

Figure 2: HID car headlamps require electronic ballasts.

Figure 3: Bipolar transistors to switch the high-voltage supply to thepiezo elements.


M O T I O N C O N T R O L

Digital current control

One typical control method is linear current control. xWith thismethod it is possible to obtain current with low ripple and thus smallnoise in the machine. The disadvantage of a linear controller is that itcan react very sensitive to wrongly set parameters. Due to the oscil-lations of the controller in case of instability the maximum current ofthe inverter can be exceeded. Since the electrical time constant ()varies widely with current and rotor position the parameters of thecontroller are difficult to adapt. Additionally a strong and dynamic dis-turbance (back EMF) causes poor performance of the controller. Thisis easily to be noticed during generator operation with small current(Figure1).

These problems are related to the nonlinear behaviour of themachine, which can be expressed by the voltage equation for onephase:

(1)

where is the back electromotive force (back EMF) term andis theincremental inductance. A mutual inductance term is neglected. Theequation (1) shows that the respective operating condition of the non-linear machine depends on position, current and speed.

( , )

( , ) ( , )

( , ) ( , , ).incr

d iu R idti di i du R ii dt dt

diu R i l i e idt

ψ θ

ψ θ ψ θ θθ

θ θ ω

= ⋅ +

∂ ∂= ⋅ + ⋅ + ⋅

∂ ∂

= ⋅ + ⋅ +

Current Control for theSwitched Reluctance Machinewith Enhanced Performance

Improved current control can be utilized effectively for torque controlSwitched reluctance machines (SRM) become more and more attractive for many industrialapplications due to their simple structure, high speed, high torque and low cost. One of the

major disadvantages of the SRMs is the torque ripple. The torque is often controlled via an innercurrent loop. To achieve good torque control, accurate command tracking by the current control

is required. Some methods for current control of the SRM are presented in [1-7].

By Joanna Bekiesch and Günter Schröder, University of Siegen

Figure 2: Experimental result showing the phase current and con-troller output, without EMF compensation and gain adaptation inmotor operation at speed 750 rpm.

Figure 3: Calculated value of the incremental inductance.

www.bodospower.com

With a small current, the incremental inductance is not saturatedand has both largest values and variation (Figure 3). Largeinductance means small closed loop gain and poor dynamics ofthe control loop. Figure 2 shows the phase current and con-troller output in motor operation (at 750 rpm). The increasing rip-ple indicates falling incremental inductance. The controller gainis optimized for the worst case (lowest incremental inductance).The total controller output is limited to ±250Vwhich is the ratedvoltage for the converter, after EMF compensation. The figuresshow the controller output before the final limit (see Figure 5).The back EMF is a rapidly changing voltage forcing the con-troller to react very fast which causes control deviations.

To solve these problems a PI gain adaptation for the incremen-tal inductance and an EMF compensation method are proposed.

Gain Adaptation for Incremental Inductance

The incremental inductance of the SRM is non-linear anddepends on the phase current and rotor position, as can beseen on the Figure 3. This means that at low current, where theincremental inductance reaches high values, the controller withconstant (gain of the P-part) can not fulfill the control task.Because of that the difference between reference current andthe measured one is particularly noticeable in generator opera-tion (Figure 1). Improper design can lead to either poor com-mand tracking or oscillations. To decrease the fluctuation, thegain of the PI controller has to be a function of the incrementalinductance. Figure 5 shows the complete structure of the controlloop. The transfer function of the current loop to be controlled is:

(2)

where is the sum of the small time constants: the time constantof the converterand half the ADC sampling time is the electricaltime constant, which equals:

(3)

In this paper it is suggested to make the, which is proportionalto the electrical time constant, therefore depending on the incre-mental inductance. The incremental inductance can be calculat-ed from the equation:

(4)( , )

( , )iL ii

ψ θθ

∂=

∂

( , )A

L iTRθ

=

( ) 1 1 1

( ) 1 1i

A

I sGU s R sT sTσ

∗= =+ +

Figure 4: Measured Value of the back EMF at reference speed.

All the Power you need

Tyco Electronics is now expanding its solderless

MiniSKiiP® product portfolio and offers both 600V

and 1200V products. The current standard range

includes MiniSKiiP® 2nd generation PIM (input

rectifier, brake chopper and SIX-PACK) and SIX-PACK

products. They are covering up to 100A/600V and

70A/1200V for PIM modules as well as for SIX-PACK

modules in size 1, 2 and 3 housings. MiniSKiiP® 1st

generation products are offered on special request.

For a complete cross reference please download

following file from our internet page:

http://em.tycoelectronics.com/pdf/MiniSKiiP_Cross_Ref.pdf

Tyco Electronics Power Systems Finsinger Feld 1 85521 Ottobrunn, Germany Tel.: +49 (0)89 6089 830 Fax: +49 (0)89 6089 833 [email protected]

Features:W Up to 100A / 600V

W Up to 70A / 1200V

PIM Module

SIX-PACK Module

Visit us at:Electronica

Munich/Germany, Booth B3.225



The flux was measured with the direct measuring method intro-duced in [8]. To omit unnecessary calculations for adjustment ofthe and accordinglyand to reduce the calculation time, the incre-mental inductance values can be stored in a loop-up table. Theelectrical time constant is multiplied by a constant to get the con-troller gain.

Back EMF Decoupling

The second problem for the current control is the influence of theback EMF. From the equation (1) it can be noticed that the EMFis a function of the position, phase current and speed. The backEMF is small at low speed and has not as significant negativeimpact on the controller. This influence increases at higher speed.The back EMF as a disturbance is already known from the DCmachine. Compared to the DC machine, the change of thisinduced voltage is much quicker. Therefore the back EMF at theDC machine, contrary to the SRM, has no significant impact onthe dynamic work of the controller. Figure 4 shows that the EMFis highly non-linear. Because of that, a controller without feed for-ward can not compensate for the quick and large changes of thisinduced voltage and a difference between reference current andthe measured one develops.

To prevent this, the EMF compensation method is proposed. Thetask of this method is to minimize variable disturbances in thecontrol loop which are related to the back EMF. This EMF com-pensation block contains a look-up table of the voltage levels forthe feed forward control. For the application of the table, theangular position as well as current and speed are needed asinput variables. At constant current and constant angular speedequation (1) is reduced to:

(5)

If the winding resistance is known, the back EMF can be calculat-ed and stored. For operation at arbitrary speed the compensationcan be calculated as:

(6)

With the data from the table the voltage canbe estimated on-line. The test results weredetermined with a regulation cycle time of40ms. Figure 4 presents the measured val-ues of the back EMF. The EMF compensa-tion is made for each operating conditiondefined by current, position and speed inorder to eliminate the variable disturbancesto a large extent. Current control with EMF compensation andconsideration of the variable time constantFigure 5 shows the block diagram of the pro-posed method. In addition to the PI con-troller, the EMF compensation is used andthe gain of the PI controller has to be propor-tional to the incremental inductance.

Figure 6 and Figure 7 show the experimentalresult of the application of the EMF compen-sation method in the control loop. In compar-

0

0

( , , )e e i ωθ ω

ω= ⋅

0( , , ).u R i e i θ ω= ⋅ +

Figure 5: Block diagram of current control with EMF compensation method and adaptation forincremental inductance.

Figure 6: Experimental results showing the phase current and currentand controller output, with EMF compensation and gain adaptation inmotor operation at speed 750 rpm.

Figure 7: Experimental results showing the phase current and currentand controller output, with EMF compensation and gain adaptation inmotor operation at speed 750 rpm.

Visit the Apec 2007 web sitefor the latest information!

www.apec-conf.comwww.apec-conf.com

2007February 25–March 1, 2007

Disneyland, Anaheim, CA

THE PREMIER

GLOBAL EVENT IN

POWER ELECTRONICSTM

THE PREMIER

GLOBAL EVENT IN

POWER ELECTRONICSTM

SPONSORED BY



ison to the current without EMF compensation depicted in Figure 1and 2 an improvement is to be noticed. The current now follows thedesired current value accurately. The controller output is almost zero.That means that the controller only reacts on the variable distur-bances which are not related to the back EMF. With this the con-troller’s capacity is not exceeded. The experimental results weredetermined for 10% of rated current and at speed 750 rpm.

Conclusion

The modification of the PI current control allows a minimization of thedisturbances in the control loop, which are related to the saturation orback EMF and variable time constant. This article focuses on theSRM with low current, because the incremental inductance hasextreme variations in that case and it has a significant negativeimpact on the dynamics of the controller. The experimental resultsshow that the current controller is capable of performing highlydynamic output variations and that the closed loop gain can be kepton a constant high value which allows good dynamic behaviour in thewhole operating range. This improved current control can be utilizedeffectively for torque control.

References

[1]. R.B. Inderka. Direkte Drehmomentregelung GeschalteterReluktanzantriebe, Dissertation, Köln, 2002.[2]. M. Khwaja, M. Rahman, S.E. Schulz. High-Performance FullyDigital Switched Reluctance Motor Controller for Vehicle Propulsion,IEEE Trans. on Industry Applications, vol. 38, no. 4, pp. 1062-1071,

2002.[3]. S.K. Sahoo, S. K. Panda, J.X. Xu: High Performance CurrentController for Switched Reluctance Motors based on IterativeLearning, Proc. Eur. Conf. on Power Elec. and Appl. EPE, CD-ROMPaper No. 1064, Toulouse, 2003.[4]. S. E. Schulz, K. M. Rahman. High Performance Digital PI CurrentRegulator for EV Switched Reluctance Motor Drives, Proc. IEEE Ind.Appl. Conf. Annual Meeting IAS, CD-ROM Paper No. 42P3,Pittsburgh, 2002.[5]. Z. Lin, D.S. Reay, B.W. Williams, X.He. High PerformanceCurrent Control for Switched Reluctance Motors with On-lineModeling, Proc. IEEE Power. Elec. Spec. Conf. Annual MeetingPESC, CD-ROM Paper No. 1246_12076, Aachen, 2004.[6]. A.Greif. Untersuchungen an Geschalteten Reluktanz-antrieben fürElektrofahrzeuge, Dissertation, Neubiberg, 2000.[7]. F. Blaabjerg, P. C. Kjaer, P. O. Rasmussen, C. Cossar. ImprovedDigital Current Control Methods in Switched Reluctance MotorDrives, IEEE Transansactions on Power Electronics, vol.14, no. 3,pp. 563-572, 1999.[8]. C. Cossar, T.J.E. Miller. Electromagnetic testing of switched reluc-tance motors, Proc. Int. Conf. on Electrical Machines ICEM, vol. 2, pp470-474, Manchester, 1992.

[email protected]

[email protected]

ICT Concept

Thyristor-type turn-off devices like the GCT are controlled by the gatecurrent [2]. During turn-off, the complete anode current has to becommutated from the cathode to the gate connection before theanode-cathode voltage starts to rise. To ensure this unity gain turn-off, the gate-cathode voltage has to be kept below approximately 0.5V. Using MOSFETs to shortcircuit this path, i.e. the MTO concept [3],a safe turn-off can only be guaranteed if the impedance in the com-mutation path is very low. This is not easy to realize in a hybrid fash-ion, but becomes feasible by directly connecting the MOSFETs to thewafer [4].

Alternatively, anadditional voltagesource can beimplemented in thegate-cathode path.Thereby, the influ-ence of the parasiticimpedances is mini-mized and theunity-gain imple-mentation is simpli-fied. Since the volt-age source isstressed with veryhigh current pulses,precharged capaci-tors are used. Notonly the parametersof the selecteddevices influencethe impedance, but

also the geometrical layout. Since a long current loop area yields ahigh inductive impedance, commercial drivers for GCTs minimize theimpedance by integrating the GCT power semiconductor into thegate drive unit.

Integrating critical parts of the GDU into the GCT housing canimprove the overall performance significantly. In the proposed ICT[5], the turn-off unit, consisting of MOSFETs and capacitors, is placednext to the GCT wafer inside the presspack housing (figure 2).Consequently, the remaining GDU is simplified and can be connectedvia cable to the ICT (figure 3). In addition, the coupling between GCTwafer and turn-off unit benefits from the shorter connection andimproves the commutation process during turn-off.

H I G H P O W E R S W I T C H


The Internally CommutatedThyristor (ICT)

A new GCT with integrated turn-off unit

Today, Gate Commutated Thyristors (GCTs) and Insulated Gate Bipolar Transistors(IGBTs) dominate the high-power semiconductor market. While IGBTs are mainly offeredin modules, GCTs are only available in press-pack housings. Since the operation of GCTsrequires a very low inductive and low resistive connection between the GCT device andthe gate drive unit (GDU), the GCT housing is directly connected to the GDU (figure 1),resulting in an Integrated Gate Commutated Thyristor (IGCT). This design requires closeattention to creepage distances and thermal stressing of the GDU. In addition, the gatedrive of IGCTs is usually much larger than an IGBT driver and consumes considerably

more power.

By Peter Köllensperger and Rik W. De Doncker; Institute for Power Electronics and Electrical Drives; RWTH Aachen University

Figure 2: Principle of the InternallyCommutated Thyristor


Integrated Turn-Off Unit

To enable the integration of the turn-off unit into the presspack hous-ing, two important aspects have to be considered: limited space andhigh temperatures (up to 125°C) inside the housing.

The size of the standard turn-off unit, that can be seen in the red boxin figure 4, is much too large for an integration into the press-packhousing. This requires a drastic volume reduction, which can only beachieved by a complete redesign of all components.

The standard electrolytic capacitors, which are very sensitive to hightemperatures, are replaced by Multi-Layer-Ceramic-Capacitors(MLCCs) using X7R material.

The limited ac-current capability of the electrolytic capacitors in thestandard GDU requires the parallel connection of many capacitors,resulting in a relatively high capacitance. In contrast, the necessarycapacitance of the new turn-off unit can be calculated by defining amaximum voltage drop during turn-off of the GCT. In addition, aseries of fast switching cycles is usually required without the possibil-ity of recharging in between. In an ICT, these two aspects are themain design criterion for the capacitors of the turn-off unit.Consequently, it is considerably smaller, e.g. 470 µF instead of 44.7mF for a 520 A device. In addition, the MLCCs have lower parasiticimpedances, which is also beneficial during turn-off.

Concerning the MOSFET switches of the conventional GDU, only thevolumetric ratio between silicon and packaging has to be improved inorder to minimize the required volume. This can be accomplished bythe use of DirectFETs from IR [6]. They basically consist of the MOS-

FET chip with a small metal shielding at the drain contact thatenables a direct soldering to the PCB and double sided cooling.Furthermore, this package is able to sustain pressure, which allowsfor pressure contacting.

The resulting turn-off unit for the first ICT prototype can be seen infigure 5 in comparison to an one Euro coin. Further details can befound in [5].

Measurement Results

Several tests were carried out to verify the function of the first proto-type ICT. Figure 6 shows a successful Safe Operating Area (SOA)test at 3.3 kV, 520 A and 125°C with the ICT, which equals the per-formance of the commercial version. The current commutation fromcathode to gate starts at t = 0 µs and is finished at t = 0.3 µs wellbefore the anode-cathode voltage begins to rise (t = 0.8 µs), whichapproves the hard-drive of the GCT.

The capacitors were precharged to -10 V, compared to the -20 V ofcommercial GDUs. This leads to a significantly reduced power con-sumption of the ICT GDU. Since the final GDU was not yet availablefor this test, the capacitors were not recharged quickly after the turn-off.

The gate current IG was measured with a small custom-madeRogowski-coil between turn-off unit and GCT. The gate current alsoshows a distinctive reverse-recovery of the gate-cathode diode partof the GCT, proving the hard turn-off and indicating potential forswitching off even higher currents [7].


Figure 4: Comparison between commercial and new turn-off unit

Figure 5: Internal turn-off unit

Figure 6: SOA test with ICT device

Figure 3: First ICT prototype



Remaining Gate Drive Unit

As stated in II, the GCT is a current controlled device. In order toturn-on both the IGCT and the ICT, a positive current peak is need-ed, followed by a continuous dc-current to be fed into the gate con-tact. However, the demands on the respective GDUs during turn-offswitching differ significantly. An IGCT driver has to carry the fullanode current of the device, whereas an ICT driver just has to issuethe turn-on signal for the internal MOSFETs of the ICT. In contrast tothe IGCT driver, the ICT GDU has no need for a low inductive, lowresistive current path to the GCT, enabling a simple cable connec-tion.

Since the turn-off unit, which scales with the ICT current rating, isintegrated, the remaining GDU can be used for different ICT ratings.Consequently, it is no longer necessary to build specific GDUs foreach wafer size and therefore possible to build application specificdrivers [8].

Reliability

If the reliability of the total ICT switch is investigated, three differentfactors can be distinguished: First, the reliability of the GCT waferand the presspack housing, secondly the reliability of the internalturn-off unit and finally the reliability of the GDU. Thyristor-type high-power devices in press-pack housings, e.g. GTOs and GCTs, usuallyhave excellent reliability records. Therefore, the GCT wafer shouldnot be a concern, since it is a proven design. In contrast, no workexperience with the internal turn-off unit, consisting of DirectFETsand MLC capacitors, has been gathered yet. The DirectFETs do notcontain any bond wires, produce only very low losses and are ther-mally coupled closer to the heatsink than to the wafer, since the ther-mal conductivity of the MLCCs is very high. The DirectFET housingcan sustain pressures of several hundred Newton without fatigueeffects [9].

MLC capacitors show no electrical wear out mechanism, e.g. they donot dry out like electrolytic capacitors and can operate over therequired temperature range without derating. The ceramic capacitorsare mounted vertically underneath the PCB and the complete turn-offunit is spring mounted into the press-pack housing. Thus, mechanicalstresses due to different thermal expansion coefficients between cop-per and X7R are absorbed and cannot cause cracks in the MLCCs.

The GDU can be located outside the press-pack stack with consider-able distance to heatsinks and ICT. In addition, no electrolytic capaci-tors are needed that could limit the allowable ambient temperature.Considering the total switch, consisting of ICT and GDU, an excellentreliability is expected.

Conclusion

The ICT concept with optimized gate drive possesses several advan-tages over the standard IGCT in three major areas. The turn-offcapability of the GCT wafer is improved by the lower stray inductanceand the lower ESR of the MLC capacitors compared to electrolyticcapacitors. Ease of use of the complete switch is better due to theflexible low power cable connection, the reduced number of differentGDUs, the lower power consumption and the larger ambient temper-ature range. Finally, the reliability is improved by absence of elec-trolytic capacitors and a lower temperature at the GDU (cable con-nection enables relocation).

Acknowledgment

The authors would like to thank Eric Carroll from ABBSemiconductors Ltd. for supporting this work.

References:

[1] “5SHX 06F6004 datasheet,” ABB Switzerland Ltd, 2002,www.abb.com[2] H. Grüning, B. Ødegard, J. Rees, A. Weber, E. Carroll and S.Eicher, “High-power hard-driven GTO module for 4.5 kV/3 kA snub-berless operation”, PCIM 1996, Nueremberg, Germany[3] D. Piccone, R. W. De Doncker, J. Barrow and W. Tobin, “The MTOthyristor-a new high power bipolar MOS thyristor”, IAS 1996, SanDiego, USA[4] Dirk Detjen, Stefan Schröder and Rik W. De Doncker, “ New High-Power BIMOS-Devices Based on Silicon-Silicon Bonding“, IAS 2002,Pittsburgh, USA[5] P. Köllensperger and R. W. De Doncker, “The InternallyCommutated Thyristor - a new GCT with integrated turn-off unit”,CIPS 2006, Naples, Italy[6] A. Sawle, C. Blake and D. Maric, “Novel power MOSFET packag-ing technology doubles power density in synchronous buck convert-ers for next generation microprocessors”, APEC 2002, Dallas, 2002,USA[7] H. Grüning and K. Koyanagi, “A new compact high dI/dt gate driveunit for 6-inch GCTs”, ISPSD 2004, Kitakyushu, Japan[8] Peter Köllensperger and Rik W. De Doncker, “Optimized GateDrivers for Internally Commutated Thyristors (ICTs)”, IAS 2006,Tampa, USA[9] “Application note an-1035”, International Rectifier, 2004,www.irf.com

[email protected]


N E W P R O D U C T S

LEM has introduced the AT series of AC cur-rent transducers, reducing the price point forproducts made to LEM’s demanding qualitystandards. An innovative core materialenables split-core current transformer tech-nology to reach new levels of performancefor this category of products.The AT transducers combine an accuratecurrent transformer and signal conditioningelectronics in a very light and compact caseto offer substantial savings in size. Theymeasure only 44.5 x 67 x 36.5 mm with a 16mm diameter sensing aperture for non-con-tact measurement and are, therefore, partic-ularly suitable for installation in tight environ-

ments.Being split-core and either self-powered orloop-powered they are easy to install andput into operation, which means they can beretro-fitted into existing installations withoutshutting down operation. This ease ofmounting, coupled with the transducers’ highreliability, will reduce installation and mainte-nance costs.The transducers have been designed tomeasure 50/60Hz AC signals with RMS(average) computation. They provide anabsolute accuracy better than 1.5 percent ofthe nominal current over a broad range ofinputs. A choice of primary current measure-

ment ranges from 5A to 50A are available,with a selection of industry standard outputtypes (4-20mA, 0-5VDC or 0-10VDC).SPS/IPC/Drives – Nürnberg Booth 7, 179

www.lem.com

Compact transducers for AC current

Mitsubishi Electric now presents the nextpioneering step of transfer mold technology,a new CIB (converter – inverter – brake)module rated from 20A up to 30A/600V aswell as from 10A up to25A/1200V driven and protected by a newlydeveloped 1200V HVIC.The DIP-CIB module for general purposeinverters and small power industrial use fea-tures latest CSTBT silicon chips and latestpackaging technology for low thermal resist-ance. Its compactconstruction in conjunction with the suggest-ed heat sink shape fulfils international safetystandards such as UL508 and IEC664-1. Allmodules are fully RoHS compliant. Suitable

for up to 5.5kW class drives the new moduleincludes three-phase input rectifier, brakechopper and three-phase inverteras well as a NTC for the temperature sens-ing of the base plate and provides open

emitter topology.A newly developed dedicated 1200V HVICwith its dual level voltage shift circuit and theseparate driver terminals for turn-on, turn-offand keeping off the IGBT offers a very reli-able and rugged driversolution when a functional isolation betweenthe controller and the CIB is required. Itscomplete protection functions for short circuitand under voltage have been tested in con-junction with the DIP-CIB on an evaluationboard.SPS/IPC/Drives – Nürnberg Booth 7.320 /

1.541 / 6.210

www.mitsubishichips.com

600 & 1200V Transfer Mold CIB Modules

The most powerful device on the market.Diotec presents the new SMD bridge seriesB40S2A to B380S2A. With output currents ofup to 2.3 Amp in SO-DIL package, theseparts are currently the most powerful SMDbridge rectifiers on the market. The highpower rating is achieved by an extremelylow forward voltage drop of below 0.95 Voltsat 2 Amp. Also unique is the peak forward

surge current of as much as 72 Amp for an8.3ms sine half wave.Typical applications for these rectifiers arepower supplies for industrial and consumergoods such as set top boxes, where up tonow not powerful enough SMD versionshave been available. Due to reduced forwardpower losses, the parts can be also used athigh ambient temperatures, e. g. in lamp bal-

lasts. Finally, the high surge current is animportant parameter for lightning protectionin telecommunication circuits. Start of seriesproduction of these devices will be in thebeginning of 2007.

Electronica Booth A5.127

www.diotec.com

2.3 A SMD Bridge Rectifier

Hanau / Frankfurt – Vacuumschmelze pres-ents its new high-precision current sensorfamily at this year’s electronica (14th to 17thNovember) in Hall B5, Booth 119. TypeT60404-N4646-X654, for example, meas-ures maximum effective continuous currentup to 50 A and peaks of up to +150 A. Thedimensions of the sensor housing are anultracompact 22.2 mm x 10 mm x 24 mm.

The electronics of these new-designVacuumschmelze sensors are almost com-pletely concentrated in an IC designed incollaboration with a leading semiconductormanufacturer.The sensor displays AC and DC current ofany wave form up to 100 kHZ and over. Itrequires only a unipolar 5-volt power supplyand generates output voltage proportional to

input current. Type X664 features an addi-tional reference pin to connect, say, with thereference voltage output or input of an A/Dconverter.

Electronica Booth B5.119

www.vacuumschmelze.com

150 A High-Precision Current Sensor for AC/DC

Mesago Messemanagement GmbH, Postfach 10 32 61, 70028 Stuttgart, Germany, e-mail: [email protected], phone +49 711-61946-44

Nuremberg

ElectricAutomationSystems and Components

Exhibition & Conference28 – 30 Nov. 2006

SPS/IPC/DRIVES/

Experience electric automationat its best!Come and see it all!Control Technology

IPCs

Drive Systems and Components

Human-Machine-Interface Devices

Industrial Communication

Industrial Software

Interface Technology

Electromechanical Components and Peripheral Equipment

Sensor Technology

www.mesago.com/sps



Tyco Electronics Power Systems, Inc.announced the debut of the small and pow-erful TLynx Series, the third generation ofthe highly successful Austin Lynx Seriesnon-isolated dc/dc converters. These newPoint-of-Load (POL) modules with industrystandard footprints are designed to providehigher power, higher density, and faster tran-sient response in board mounted powerapplications. Offering greater flexibility andefficiency for power designers, the 3A and6A PicoTLynx modules are the first introduc-tions of the TLynx Series. The TLynx series will be available in 3A, 6A,12A, 20A, and 30A versions that offer scala-ble cost-efficient solutions ideally suited for awide variety of power architectures includingHybrid and Intermediate Bus Architectures

(IBA). The input voltage range for theTLynx series has been widened to offera 2.4V to 5.5V (3.3V/5V) version and a4.5 to 14V (5V/12V) version.. Samplesof the 3A and 6A PicoTLynx are current-ly available with 3.3V/5V input voltage.The TLynx Series offers the newTunable-Loop (patent pending) featurewhich allows engineers to optimize thecontrol loop to significantly improvetransient response and reduce theamount of external filtering. When thecontrol loop is optimized using the Tunable-Loop feature on the 3A and 6A PicoTLynx,load transient response can be improved upto five times (5X) using identical externalcapacitance. The Tunable-Loop is easy toimplement requiring a minimal set of addi-

tional low cost components and no additionalpins.

Electronica Booth B3.225

www.power.tycoelectronics.com

Double Power Density in Lynx POL Converter

Semikron has developed a customer-specificMiniSKiiP semiconductor module for SMA’ssolar inverter Sunnyboy SB3300, which wasgiven top quality ratings in the latest inde-pendent market survey by German producttesting foundation “Stiftung Warentest”.

SMA’s inverter achieved the best rating in itsclass not only due to its state-of-the-art tech-nology but also for its high efficiency andsuperior environmental properties. Thanks tothe use of innovative, patent-protected pres-sure contact technology, purpose-designedcircuitry and the optimum silicon chip, thisinverter boasts an efficiency of up to 95.2%,not to mention very low losses during feedinto the power grid.

The innovative semiconductor technology isbased on spring contact technology, whichensures an even distribution of pressureand, consequently, thermal resistances with-in the power module. What’s more, thesprings can be positioned in numerous dif-ferent ways, allowing for a very high degree

of layout flexibility. The result is optimumchip layout and a highly efficient inverter withlow-loss switching.

Spring contact technology has been aroundfor 15 years now and has been used suc-cessfully in “trickier” applications with hightemperature cycle capability. In fact, morethan 300 million springs of this kind can cur-rently be found in working applications. Andit goes without saying that this technology isfully compliant with the requirements of theEU Restriction of Hazardous SubstancesDirective.

SPS/IPC/Drives – Nürnberg Booth 1.220

www.semikron.com

Top Test Ratings for SEMIKRON

November 14-17, 2006 an extremely power-ful, and at the same time environmentallyfriendly, nanophosphate lithium-ion recharge-able battery from the U.S. manufacturerA123Systems. The ANR26650M1 has a 10xlife cycle vs. conventional lithium ion batter-ies and allows for much higher power densi-ty, or rate of charge and discharge. The bat-tery has the ability to recharge to 90% of itscapacity in five minutes. 3000 Watts perKilogram guarantees the highest perform-ance in the smallest form factor.The new cell, which for the first time uses ananophosphate positive electrode (cathode),has a nominal capacity and voltage of 2.3Ah, 3.3V. The internal impedance (1kHz AC)is 8 m? typical. Maximum continuous dis-

charge is 70A and pulse discharge at 10 secis 120A for each cell. With an operating tem-perature range of -30°C to +60°C, theANR26650M1 allows for use in harsh envi-ronmental conditions.A123 nanophosphate lithium-ion high-per-formance cells do not use any toxic heavymetals and are therefore, in contrast to nick-el cadmium cells, particularly environmental-ly friendly. Due to the use of thermally sta-ble, non-combustible active materials, theANR26650M1 cell offers a high level of safe-ty, even without additional protection circuit-ry. This feature is a significant distinctioncompared with conventional cobalt-basedlithium ion cells. Because nanophosphatelithium-ion cells do not require any protection

components, this allows future high perform-ance rechargeable batteries to be manufac-tured much simpler than in the past andtherefore, also makes them suitable for lowvolume production quantities.Electronica Booth B2.479

www.bmz-gmbh.eu

Nanophosphate Lithium-Ion Cells



APEC 55

Apex 41

Coilcraft C3

CT Concept Technologie 5 + 63

Danfoss Silicon Power 49

Datatronics 17

Diotec Semiconductor 31

Electronica 43 + 45

Electronica China 39

EMC 47

Fairchild C2

Fuji Electric 15

Infineon/eupec 23

International Rectifier 21 + C4

Intersil 11 + 13

IXYS 37

LEM 3

Micrel 9

Microsemi Power Products Group 59

Mitsubishi Electric 25

National Semiconductor 33

PCIM China 57

Pemuk 51

Powersem 19

Semikron 29

SPS 61

ST Microelectronics 35

Texas Instruments 7

Tyco Electronics 53

Würth Elektronik 27

ADVERTISING INDEX

A high-voltage, high-frequency buck regula-tor control IC for AC-DC offline, non-isolatedapplications requiring multiple light-emittingdiode (LED) circuits or requiring DC-DCcolor-mixing capabilities has been introduced

by International Rectifier. Applicationsinclude indoor and outdoor signage as wellas architectural, entertainment, design anddecorative lighting.Rated at 200V or 600V, the IRS254x seriesincorporates a continuous mode, time-delayed hysteretic buck regulator to controlthe average load current within a toleranceof five percent, using an accurate on-chipband gap voltage reference. An externalhigh-side bootstrap circuit drives the buckswitching element at high frequencies up to500kHz. A low-side driver is also providedfor synchronous rectifier designs.The new ICs deliver a micro-power startup ofless than 500µA for extremely low losses at

turn-on and a deadtime of 140ns for continu-ous current regulation. Other featuresinclude auto restart, non-latched shutdownand PWM dimmable capability, incorporatedinto a compact 8-pin DIP or 8-pin SOICpackage, also available in tape and reel. These ICs utilize IR’s advanced high-voltageIC process which incorporates latest-genera-tion high-voltage level-shifting and termina-tion technology to deliver superior electricalover-stress protection and higher field relia-bility, in addition to other new features andenhancements.Electronica Booth A5.576

www.irf.com

Buck Control for Constant LED Current

Intersil Corporation introduced a new familyof core controllers for meeting Intel’s next-generation (Core Duo and Core 2 Duo)mobile platform. These devices offerimproved accuracy in power dissipationreporting and industry-leading transientresponse and efficiency enabled by Intersil’spatented R3 (Robust Ripple Regulator)Technology.Intersil’s new family of controllers reports theCPU power consumption through a highlyaccurate continuous analog output signal.Compared with conventional controllers,Intersil’s R3 controllers stabilize current flowand allow less time between wave peaks ina regulator circuit.To boost battery life, the ISL6260C,ISL6261A and ISL6262A supportDPRSLRVR (deeper sleep) functions andmaximize efficiency by automatically chang-ing operation modes. At heavy loads in the

active mode, thesecontrollers com-mand the continu-ous conductionmode (CCM) opera-tion. When theCPU enters a deep-er sleep mode, thecontrollers enablediode emulation tomaximize the effi-ciency at lightloads.The ISL6260C mul-tiphase controller –together withISL6208 external gate driver – provides acomplete solution to power Intel’s mobilemicroprocessors. This family of devices fea-tures power management innovations fromthe Notebook Power group. These parts are

designed to meet the exacting specificationsof laptop and mobile device manufacturers.

Electronica Booth A4.207

www.intersil.com/power

Improve Power for Intel Santa Rosa Platform

®

21 Napier Place, Wardpark North, Cumbernauld Scotland G68 0LL+44/1236/730595 Fax +44/1236/730627

RoHSCOMPLIANT

Our new LPS shielded inductors give you thebest combination of ultralow profile and highlevel performance.Highest saturation current ratings Comparedto competitive inductors of the same size, ourIsat ratings are typically 20 - 30% higher.Widest range of L values Only Coilcraft’s LPSfamily offers you so many inductance options:from 3300 µH all the way down to 0.33 µH.

And no one else has so many high inductancevalues in a 3x3 mm footprint.Rugged construction Their impact-resistantdesign withstands 1500 G’s deceleration in onemeter drop tests, making them the per-fect inductors for handheld devices.See why designing in our new LPSinductors is a really bright idea.Visit www.coilcraft.com/lps9

These new ultrathin inductorsThese new ultrathin inductorsreally shine in LED and ELreally shine in LED and EL

backlight applicationsbacklight applications

LPS30080.56 - 330 µH

Up to 1.8 A3 x 3 mm

0.8 mm high

LPS30100.47 - 330 µH

Up to 2.3 A3 x 3 mm

0.9mm high

LPS30151 - 330 µHUp to 2.0 A3 x 3 mm

1.4 mm high

LPS40120.33 -3300 µH

Up to 5.0 A3.9 x 3.9 mm1.1 mm high

LPS40180.56 - 3300 µH

Up to 4.8 A3.9 x 3.9 mm1.7 mm high

Visit usElectronicaHall B6 Booth 242

DeadtimeCircuitry

andLogic

HV LevelShifters

UVLO

UVLOLO

HO

IN

DT

COM

TOLOAD

Up to 600V

VS

VBVCC

VSS

Delay

SD

THE POWER MANAGEMENT LEADER

IR SETS THE STANDARDFOR 600V ICs

For more information call +44 (0)1737 227215 or +49 (0) 6102 884 311 or visit us at

www.irf.com/

Rugged, Reliable, Highly Integrated

IR’s 600V ICs for motor control, lighting, switch-mode power supplies, audio, and flat-paneldisplay applications deliver more features andgreater functionality to simplify your circuitdesign and reduce risk.

IR’s latest-generation high-voltage IC technologydelivers superior protection and higher fieldreliability in an intelligent, monolithic driver IC.

Our new ICs are offered with single or dual inputs,under-voltage lockout protection, and fixed orprogrammable deadtime for half-bridge drivers.

Features:• 3.3V logic compatible input

• Drive current up to 2.5A

• SO-8 package available*

• Separate COM and logic ground*

• UVLO protects VBS*

*select models

HALF-BRIDGE DRIVER ICs

Part Number Pin Count Sink/Source Current (mA) Comments

IRS2103(S)PBF 8 290/600 UVLO VCC

IRS2104(S)PBF 8 290/600 Input logic for shutdown; UVLO VCC

IRS2108(S)PBF 8 290/600 UVLO VCC & VBS

IRS21084(S)PBF 14 290/600Programmable deadtime;UVLO VCC & VBS

IRS2109(S)PBF 8 290/600Input logic for shutdown;UVLO VCC & VBS

IRS21094(S)PBF 14 290/600Input logic for shutdown; programmable deadtime;UVLO VCC & VBS

IRS2183(S)PBF 8 1900/2300 UVLO VCC & VBS



IRS21844(S)PBF 14 1900/2300Input logic for shutdown; programmable deadtime;UVLO VCC & VBS

INDEPENDENT HIGH- AND LOW-SIDE DRIVER ICs

Part Number Pin Count Sink/Source Current (mA) Comments

IRS2101(S)PBF 8 290/600 UVLO VCC

IRS2106/IRS21064(S)PBF 8 / 14 290/600 UVLO VCC & VBS

IRS2181/IRS21814(S)PBF 8 / 14 1900/2300 UVLO VCC & VBS

High-voltagewell

High-sidedrive stage

Low-sidedrive stage

600V half-bridge gate drive ICwith integrated UVLO protection

Deadtime/shoot-through protection

Programmabledeadtime

Input logicfor shutdown

See us on stand: A5.576.

issn: 1863-5598 zkz 64717 11-06 bodo´sbodo´spower ... and simulation ... rosu, ph.d., group leader...

Documents