CONTENTS

ViewpointThe show goes on—any show, any time, anywhere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Industry News

C & D announces acquisition of Datel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

International Rectifier investment in Wales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4

Agilent Technologies and Digi-Key Corporation signed a distribution agreement . . . . . . . .4

IEEE meeting addresses Power Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Texas Instruments and Ariston optimized washing mashines . . . . . . . . . . . . . . . . . . . . . . . 6

CoolSET provides lowest stand-by power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6

Cover Story

Active PFC with Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12

Power Player

Power processes will shape the portable market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10

Features

Technology Review—Design efficient Motion and Conversion Applications . . . . . . . . . . . 8

Power Semiconductors—Influence of MOSFET Transconductance . . . . . . . . . . . . . . .20

Power Supply/Simulation—Lossy Models Offer More Realistic Averaged Simulations .24

Capacitors—More Power for the Rails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26

Power ICs—Digital Power Control Highlights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28

Power Management—Power Manager is Easy to Design and Program . . . . . . . . . . . . .32

MOSFET Transistors—Power MOSFET Package Selection . . . . . . . . . . . . . . . . . . . . .36

Power ICs—Monolithic Inverter IC to Meet Strict Regulations . . . . . . . . . . . . . . . . . . . . .40

Power Protection—PolySwitch Devices to Protect Linear Transformers . . . . . . . . . . . . .42

New Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47-48

Member Representing

Eric Carroll ABB SwitzerlandBob Christy Allegro MicroSystemsPaul Löser Analog DevicesArnold Alderman AnagenesisDr. Uwe Knorr Ansoft Todd Hendrix Artesyn TechnologiesHeinz Rüedi CT-Concept TechnologyDr. Reinhold Bayerer eupecChris Rexer Fairchild SemiconductorAndreas Sperner IXYSDr. Leo Lorenz Infineon TechnologiesEric Lidow International RectifierDean Hendersen Intersil David Bell Linear TechnologyRüdiger Bürkel LEM ComponentsPaul Greenland National SemiconductorKirk Schwiebert OhmiteChristophe Basso On SemiconductorDr. Martin Schlecht SynQorDr. Thomas Stockmeier SemikronUwe Mengelkamp Texas InstrumentsPeter Sontheimer Tyco Electronics

Power Systems Design Europe Steering Committee Members

PowerSystems DesignE U R O P EPowerSystems DesignE U R O P E

2 Power Systems Design Europe June 2004

Approaching summer I have the impressionfrom attending many shows around the worldthat business is in recovery which will lead usinto a positive outlook for future growth.

The PCIM Europe show in Nuremberg hadattracted close to 5,000 engineers and it seemsthat companies have budgets again to let peo-ple travel to learn about new technology andtheir applications.

Power has been moved into a more impor-tant view since energy costs us more moneythan ever and we have to look for renewableresources. Just to burn fossils like wood cooland oil is not the way to hand over the planetto our kids.

The race for more efficiency can be found atany place in industry. The power supply in anylevel calls for higher efficiency. Therefore theactive switches from MOSFETs in low voltageup to IGBTs in 6.5 KV range get better in theirtotal losses, conduction loss as well the switch-ing losses at turn on and turn off.

The semiconductor industry is working hardto fine tune the last percentages at given tech-nologies. We have structures that constantlyshrink towards an ideal point. What we miss tosee is the breakthrough of promised technolo-gies like silicon carbide in active switches. Insemiconductor development, the developmentcyle of a new product, not modified devices onexisting processes, is about 3 to 5 times theduration of an elephant pregnancy (22 months).All our simulation tools have helped to get bet-ter products, as the products get more complexthe time element is critical.

It is a promising sign to see silicon carbidediodes with more ideal performance in moreapplications complement the active switch tominimize total switching losses between IGBTand diode.

So the European industry moves ahead toreduce power consumption in any design theycan. A big issue is the consumer having hisnever ending needs for comfort served by more sophisticated electronics that havesmart intelligent functions controlled by micro-controller putting devices to sleep when theyare not needed.

The automobiles go hybrid and reduce con-sumption by regenerative braking. So we con-tinue to look ahead to a number of Europeanevents in power. On the following pages forNews I included a short “Power Event” sectionthat gives you the focus on what is coming thatwe should pay attention to. I know that it is animpossible task to be at all of these confer-ences, but the proceedings are a good backupand can always be used later for self education.

PCIM had generated a number of well rec-ognized papers that inspire me to work withthe authors to have future subjects reflectedin upcoming articles in our magazine. This willgive the authors a broad recognition of theirresearch and design work as well as the ben-efit to our readers.

As we are Engineers we have learned howto keep a continued progress going. This is thefuel that motivates us for the next design chal-lenge. Looking forward to seeing you at ourindustry’s next show.

Best regards

Bodo ArltBodo.Arlt@powersystemsdesign.com

The show goes on—anyshow, any time, anywhere

VIEWPOINT

AGS Media Group

Katzbek 17a

D-24235 Laboe, Germany

Phone: +49 4343 421790

Fax: +49 4343 421789

info@powersystemsdesign.com

www.powersystemsdesign.com

Editor-in-Chief

Bodo Arlt, Dipl.-Ing.

Bodo.Arlt@powersystemsdesign.com

Publisher

Jim Graham

Jim.Graham@powersystemsdesign.com

Advertising & Marketing Director

Julia Stocks

Julia.Stocks@powersystemsdesign.com

Creative Direction & Production

Eyemotive

beata@eyemotive.com

www.eyemotive.com

Circulation Management

circulation@powersystemsdesign.com

Advertising Rates, Sizes and Digital File

Requirements available at:

www.powersystemsdesign.com

Subscription Request available at:

www.powersystemsdesign.com/psd/

subslogn.htm

Registration of copyright: January 2004

ISSN number: 1613-6365

Issue print run: 20,000

Printing by: Central - Druck Trost GmbH & Co.

Heusenstamm, Germany

AGS Media Group and Power Systems Design Europe

magazine assume and hereby disclaim any liability to

any person for any loss or damage by errors or omis-

sions in the material contained herein regardless of

whether such errors result from negligence, accident

or any other cause whatsoever.

Send address changes to:circulation@powersystemsdesign.com

Volume 1, Issue 5

www.powersystemsdesign.com4 5Power Systems Design Europe June 2004

INDUSTRY NEWS

C&D Technologies announced that it signeda definitive agreement to acquire Datel, Inc.in an all-cash transaction, terms of whichwere not disclosed.

Datel is a Mansfield, Massachusetts-basedmanufacturer with sales of approximately $60million for the twelve months ending March2004. Datel's business is focused primarilyon DC/DC converters, with additional productofferings in digital panel meters and dataacquisition components. The acquisition ofDatel will provide C&D with a broader productoffering, access to a diverse group of Original

Equipment Manufacturer (OEM) customersas well as an expanded international foot-print, notably, including operations in Japan.Datel has significant market presence in mid-power DC/DC converters, that will expandthe total C&D offering, thereby enabling theCompany to more fully satisfy customerrequirements. When completed, this transac-tion is expected to be immediately accretiveto earnings.

While subject to antitrust clearance andcustomary closing conditions, the transactionis expected to close in the second quarter.

Wade H. Roberts, Jr., president and chiefexecutive officer of C&D Technologies, stat-ed, “This transaction comports with our plansfor growing organically as well as via acquisi-tion. This acquisition expands our customerbase, product portfolio and international pres-ence. In this latter regard, Datel’s historicstrength in Asia and Europe considerablyenhances our international position. We lookforward to Datel joining C&D while recogniz-ing that this fine company has been extreme-ly successful on its own.”

C&D announces acquisition of Datel

International Rectifier investment in WalesInternational Rectifier CEO Alex Lidow

and Welsh First Minister Rhodri Morganannounced a major initiative to expand IR’sworkforce and double the production capacityat its facility in Newport, Wales.

The expansion will include the creation ofhighly skilled engineering and production jobsto fill the nearly 120 vacancies the Companyforesees as it brings on-line its 200 mm (8-inch) wafer fabrication facility. Many of thenewly created jobs will be devoted toInternational Rectifier’s latest High Voltage IC (HVIC) proprietary technology.

With today’s announcement, Dr. Lidowunveiled plans to expand IR-Newport’s char-ter beyond leading-edge low voltage devicesto include IR’s next generation of HVICs,considered by many to be the best in theindustry. This high value HVIC silicon,planned for production on the 200 mm line,will be used in integrated motor controllers for a variety of industrial, appliance and automotive applications.

“IR-Newport is one of our most advancedfacilities and, as such, an important part ofour business strategy. Since the acquisition,IR has invested over $66 million dollars toexpand our production capabilities in the 150

mm facility and has increased the staff bymore than 100. With the great skill sets avail-able here in Wales and greater Europe, andour strong relationships with the WelshAssembly and Welsh Development Agency,we are ahead of our original expansion plans.These circumstances coupled with strongmarket demand lead to our decision to bringthe 200 mm facility on-line,” said Alex Lidow,IR’s CEO.

First Minister Rhodri Morgan said,“International Rectifier is a hugely successfulcompany with an international lead in thesemiconductor market and this announce-ment will be very warmly welcomed inNewport and South Wales. I am very pleasedthat the Welsh Assembly Government hasplayed a key role in the company’s expansionin Newport through a Regional SelectiveAssistance (RSA) grant. Our backing for thisproject has been crucial in securing a futurefor Newport’s semi-conductor manufacturingskill base and demonstrates InternationalRectifier’s commitment to develop skilledR&D and Operations openings here in the region.”

International Rectifier acquired theNewport facility in 2002. Since that time,

the facility has met or exceeded all expecta-tions, including the consolidation of IR’s lowvoltage trench product line production, thedoubling of Newport’s initial productioncapacity, and providing exceptional supportto IR’s low voltage R&D program. Theseefforts have dramatically accelerated IR’sposition in this important market segment.

International Rectifier’s planned investmentof $40 million in IR-Newport will result inmore than 120 additional jobs. Furthermore,this investment will also provide the potentialto quadruple the facility’s production capabili-ty in the future.

Agilent Technologies and Digi-Key Corporation signed a distribution agreement

Agilent Technologies and Digi-KeyCorporation, one of the fastest growing elec-tronic component distributors in the world,today announced that they have signed a distribution agreement. The agreementallows Agilent’s Semiconductor ProductsGroup (SPG) to make a broad array of itsproducts accessible to designers via Digi-

Key's vast distribution channel and real-timeonline service and support site.

Digi-Key’s inventory of Agilent productsincludes a full range of LEDs and displays,optocouplers, motion control encoders, RFand microwave devices and infrared and optical transceivers. These products are usedin applications that span industrial, office

automation, consumer electronics, homeappliances, signs and signals, wireless, networking and automotive markets.

Agilent’s semiconductor products are nowavailable through the Digi-Key Web site.www.digikey.com

www.irf.com

www.cdtechno.com

www.digikey.com

6 Power Systems Design Europe June 2004

IEEE meeting addresses Power Electronics

More than a hundred international partici-pants of the IEEE meetings in May 2004 atGerman capital Berlin received an insight intothe usage of power electronics in electricpower systems. In particular having experi-enced 2003's large area grid outages inEurope and United States, the benefits ofrapidly controllable high power converters as

well as the challenges related to their con-nection with the electric network have attract-ed increasing attention:

Siemens AG—Power Transmission andDistribution—High Voltage Division—present-ed state of the art and R&D activities regard-ing the use of power electronics in high volt-age transmission systems, e. g. HVDC andFACTS. The challenge of high voltage highpower switching is faced using actual dedi-cated thyristors as semiconductor valves,taking into consideration functionality likedevice light triggering or indispensable con-verter short circuit current limitation.

Alstom Power Conversion GmbH—General Drives—gave an insight into drivesystems in the Megawatt range, comprisingpower electronics—e. g. a multi level mediumvoltage converter—and machines—e. g. a5MW permanent magnet generator. The sys-tems have been designed for applications asdifferent as offshore wind parks, pump stor-age plants or railway traction.

Besides leading industry, Berlin Universityof Technology had invited to learn about main activities of Electrical Machines, Drivesand Renewable Energies Group and PowerElectronics Group within Department ofElectrical Engineering and Computer Science,where machines or converters for renewableenergy and medium voltage drive systemsplay an important role, proving the relevanceof the subjects of current scientific researchalso for industrial development projects.

Prof. Dr.-Ing. W. Leonhard—TUBraunschweig—gave a thoughtful lectureabout sustainable supply of electric energy

It is remarkable that this interdisciplinaryevent, linking power electronics and powerengineering, has been jointly organised byIEEE Joint IAS/PELS/IES and PES GermanChapters, represented by several Germanand international officials, in conjunction withVDE—a fruitful cooperation to be continuedin future.

eupec displayed its advanced IGBTprodTI’s TMS320C24x digital signal con-trollers leverage Field Orientated Control inthree-phase asynchronous motor for low-cost, algorithm-based speed control.

Ariston, one of the leading brands ofMerloni Elettrodomestici (MERI.MI), and

Texas Instruments Incorporated (TI)(NYSE:TXN) today announced that TI’s -based TMS320C24x controllers reduce noise levels and improve washing efficiency.Ariston's Super Silent line of washingmachines. Using TI's low-power controllers,leading white goods manufacturer Merloni

Elettrodomestici has implemented a moreefficient three-phase alternating current (AC)motor with Field Oriented Control (FOC) in its designs.

For more information on TI’sTMS320C24x, see www.ti.com/merlonipr

www.fairchildsemi.com

Texas Instruments and Ariston optimized washing machines

www.ti.com

INDUSTRY NEWS

Infineon announced its third generationintegrated multi-chip power IC family, rein-forcing the company’s position as a leadingsupplier of power semiconductors forswitched-mode power supply (SMPS) appli-cations. The CoolSET F3 family allows SMPSmanufacturers to quickly design lighter, morecost-effective power supplies with high relia-bility and optimized efficiency.

“The U.S. government estimates that thetotal amount of electricity flowing throughexternal and internal power supplies in thatcountry alone is more than 207 billionkwh/year, or about 6 percent of the nationalelectric bill, and that more efficient designscould save an estimated 15 to 20 percent of that energy,” said Arunjai Mittal, VicePresident and General Manager, PowerManagement & Supply Business Unit,Infineon Technologies AG. “With the indus-try’s lowest stand-by power consumption,

the CoolSET F3 family can contribute signifi-cantly to achieving those savings.”

The stand-by power consumption ofInfineon’s CoolSET F3 products is the lowestcurrently available, exceeding the specifica-tions of such standards as Energy Star andthe German Blue Angel Eco Norm. For exam-ple, in a typical 30 watt (W) DVD recorder,the stand-by power consumption of aCoolSET F3 device is less than 100 mW. Themaximum allowed for 15 W-to-50 W suppliesunder the Energy Star and European energycommission target specifications is 500 mW.The lowest consumption achieved by a com-petitive device in the same type of applicationis above 150 mW measured on a 10 W board.

Further information on Infineon’s CoolSETproducts is available at:www.infineon.com/coolset

www.infineon.com

CoolSET provides lowest stand-by power

Features▼ Info & Online Store▼

Power Events

• York EMC 2004, July 1-2, York Racecourse

UK, www.yorkemc.co.uk/emcyork

• EPE-PEMC 2004, September 2-4, Riga, Litvia;

www.rtu.lv/epe-pemc2004

• H2Expo 2004, September 15-17, Hamburg,

www.h2exp0.de

• Automotive EMC 2004, October12, York

Racecourse UK, www.AutoEMC.net

• Electronica 2004, November 9-12, Munich,

www.global-electronics.net

• Surface Mount 2004, November 16-18,

Brighton UK, www.smartgroup.org

• SPS/IPC/DRIVES 2004, November 23-25,

Nuremberg, www.mesago.de

www.powersystemsdesign.com8 9Power Systems Design Europe June 2004

TECHNOLOGY REVIEW

The world is demanding more andmore power to serve the peoplesneed of communication, traveland leisure in life. It starts with powermanagement in handhelds like cellphones and a wide variety of computeroriented tools and toys. It is our life styleenhancements and work that can be donewith much more ease and efficiency. In this case we are at the low level of power consumption. The amount ofdevices being built reflect the growingconsumption and demand for more.

All around the world engineers areworking to optimize device technologyand create new topologies to reduceenergy consumption waste. The IC andmicro-controller manufactures are theones that look from the bottom to the top of power applications. They are optimizing chip processes and generatethe right elements to control the powerswitches. What ever these switches are. Starting with the popular MOSFETs,having the IGBts and continuing up tomodified turn off devices in the kilo volt range.

Power supply manufactures havingthe MOSFETs as the active low voltageswitch in the DCDC converters. Nowthey use the MOSFETs more and moreto boost efficiency as they replace therectification diodes by MOSFETs achiev-

ing the active rectification. As a resultwe have only the Rds on from theMOSFET and gain a few percentageperformance that does not heat up our product. DCDC conversion getsmore and more important as point ofload supply is the key to moderndesigns. It does not matter if you lookinto the design on a board or morebroad in a system. The driving factor tohave point of load supply is the increaseof transistors in ICs at higher operatingfrequency continues to improve ICprocesses with lower operating voltage.The result are higher peak currents atthese loads. Therefore to optimize thesupply, multiphase designs havebecome a standard in the industry. Alsothe amount of equipment being aroundthe computer and having interface con-nections by wire and get powered andcharged through the data link of thecomputer. This excess capability mustbe given in the system design to sensestatus and function to the demand andavailability of power. All IC design manu-factures have a focus to serve this market.

To make the power supply more easyto adapt digital controlled power supplyis on its way. A great benefit in systemsis fine tuning if requested to match vari-ations in end-products that have slightlydifferent parameters to be served. Thatmeans set the voltage exact to the

best performing level to supply themicro-controller.

All of that helps in saving energy. If welook ahead we can see that energy stor-age is the way to recycle the energy andreduce overall consumption. Very popu-lar and efficient is the storage of break-ing energy from transportation systems.The train, the streetcar and the hybridvehicle do it, generate power duringbraking and feed it back. It is a questionhow much energy gets back and whatsource is best to store it. Super-capaci-tors, flywheel and batteries are the typi-cal storage solutions. To manage allthese functionable, semiconductorswitches are requested. Here we seethe IGBT doing it’s best.

The IGBT is a little younger than thePCIM Europe show. Both are matureand we have seen the blocking and cur-rent handling capability of IGBTs grow-ing over the years from line voltage atthe mains, which are about a few hun-dred volts, up to substitute today GTOapplications in the kilo volt range. IGBTshave reached the 6,5 kilovolt breakdowncapability together with an improvedsave operation area they are movinginto the traction applications. PCIM 2004Conference had a number of goodpapers to listen to. Everybody that didnot make it to the show can contact the

organizer at www.pcim.de for proceed-ings of the last Conference in May.Thenext event to look to is the PESC andchips in Aachen at the end of June.

For us in Europe Automotive is a widefield for power electronics. The classiccar equipped with a combustion enginegets more and more electronics toimprove combustion motor managementby micro-controller and having again theIGBT as the switch optimizing the ignition.Ignition IGBTs are tailored to serve thespark plug with the proper electrical pulse.Also including ABS and all the audio and video systems available in our cars.Next generations will go to higher busvoltage. DCDC converters will help tobridge old to new standards. For alter-nator/generator designs we need evenhigher bus voltages than 42 volt. Toimprove efficiency without decreasingperformance we need to go to thesemore complex power systems designs.Automotive applications require for thereliability at higher temperature, a stronginterest to have a monolithic device toreduce the failure rate in systems.

Smart Power IC process technologycontrols functions and switches likeMOSFETs on the same die. ICs needmore than double amount of wafer pro-cessing steps than a MOSFET. Producinga MOSFET in the IC process technologywill require the IC cost for the die space.So the technology must benefit doingthings like that. Higher operation fre-quency can be one point. Heat dissipa-tion has been one major topic besidecost to make that product widely usable.Breakdown voltage for these processesin general are running at and below 100 volt. More or less the amount ofmasks are needed makes the MOSFETvery costly. So what we see is a limita-tion of max current in most applications.Often we have the option to add anexternal MOSFET as the switch forextended power range.

The range of Smart Power solutionshas a wide range of application. DI elec-trical isolation is a way to generate reli-able monolithic power ICs that have agood immunity against reverse inductivevoltage that can trigger the thyristorestructures in ICs based on junction

isolation. Harris had pioneered in thepast the DI process technology to gen-erate reliable products for space. It hadbeen adapted for line interface ICs intelecom. Now we see the approachmoving to higher voltage and generatingthe platform for full inverter designsbeing achieved on a single chip includ-ing the right optimized switches.

Having the right ability to envision theprogress and team up with major semi-conductor makers and being present atmost important events like PCIM inNuremberg and others on my list arePESC, EPE and APEC. I just named afew that get my attention to serve youwith crucial information to do the bestpossible job in design.

Simulation had made a significantcontribution to power systems designs.Nowadays we have the product virtuallyready to go for production with veryclose to reality results. The more andmore advanced software and the extrac-tion of data from devices in semiconduc-tors including thermal details gives theplatform that allows to work straightahead. Simulation starts on the semi-conductor design level and continues tothe device level and goes step for stepup and includes the surrounding ele-ments to simulate total power systems.

The computers of today haveincreased calculation speed togetherwith “unlimited memory space”. Workingwith these tools today we have muchmore detailed information to describe all elements with more accuracy in themodel. That includes today thermalbehavior, something very crucial tosmart power.

Thermal management is finally thekey to any long-term stable design.Simulation is the platform to do it quiteefficiently. The junction temperature isthe key parameter that gives the leadingedge for any calculation. The maximumwhich is allowed and the one you chosewill give you the lifetime by taking thecycling into account. Simple to say. Noteasy in some cases to achieve, but thatis what physics delivers to us.

When we look into power manage-ment, it depends upon the application.Individuals working on automotive con-trollers may define 42 volts as being ahigh voltage, high power. Conversely,those in Europe operating at 240 voltsor 420 volts, consider any value of volt-age below these values as low voltage.However, this voltage region is the mostcommonly used in industrial and com-mercial electronics.

The IC manufactures are looking toapplications in communications, com-puter and the portable and wireless por-tion of that business. Here we have thestrong demand to use the given energyat the best performing efficiency.

My listing for contacts below is just anintroduction to key players to learn from.

www.artesyn.comwww.abb.com/semiconductorswww.advancedpower.comwww.allegromicro.comwww.ansoft.comwww.bergquistcompany.comwww.IGBT-Driver.comwww.curamik.comwww.danfoss.com/siliconpowerwww.epcos.comwww.eupec.comwww.fairchildsemi.comwww.ferrazshawmut.comwww.fujielectric.comwww.egraf.comwww.hitachi-eu.com/pddwww.infineon.com/powerwww.irf.comwww.intersil.comwww.intusoft.comwww.ixys.comwww.lem.comwww.linear.comwww.microsemi.comwww.mitsubishichips.comwww.onsemi.com www.ohmite.comwww.powersem.dewww.powerint.comwww.pcim.dewww.semikron.comwww.ti.comwww.thermacore-europe.comhttp://em.tycoelectronics.comwww.zetex.com

Power Systems solutions are key to success Power starts at the device level based on controlling functions. The next step is to handle the powerswitches. The goal is to build efficient systems. To have valid cross communication for improvement

By Bodo Arlt, Power Systems Design EuropeEditor-in-Chief

TECHNOLOGY REVIEW

10 Power Systems Design Europe June 2004

POWER PLAYER

Advancements in portable powermanagement technologies con-tinue to grow rapidly, keepingpace with continued consumer demandfor portable devices. Portable powermanagement circuits thrive in a diversemarket, with solutions ranging fromsmall, multi-channel power convertersfor wireless communication products tohigher integration battery and powermanagement circuits for digital videorecorders and players. With such diver-sity, there are many factors affecting theentire portable power market. Two tech-nology factors, power semiconductorprocesses and packaging, play impor-tant roles in driving these advancements.

Looking at semiconductor processes,the continual move to smaller geome-tries in high-density digital chips directlyenables greater capabilities within ana-log and power products. New techniquesand methods to improve the way systempower is handled as well as drivinggreater efficiency improvements andpower management scheme flexibilityare becoming a focus for these newprocesses. For example, by combininghigh-density logic with power, greatersystem control and flexibility is achievedwithout sacrificing size and cost. In thepast, many of the control functions wereimplemented utilizing fixed discretecombinations of analog and digital con-trol schemes. Adding dense digital logic(0.25 micron or smaller) to a traditionalpower process allows digital logic to addmore complex and adaptive controlfunctions as well as accomplish themore traditional control functions.

Why is this important? The answerlies partly in the ability of the powermanagement products to react effectively

to the changing input and outputrequirements. As systems continue toincrease in power complexity due tomultiple rails and startup/shutdownrequirements, digital control providesthe functionality and flexibility to meetthe requirements. Input transients duringstart-up can require over-design of thepower system. Load changes as theCPU switches operating frequency oradding or removing peripheral systemsvary the load regulation requirements.Managing the power regulation under all of the varying operating conditions isa complex challenge. Although analogcomponents can provide high speedand small size when managing linearcontrol applications, extending thesemethods to more complex non-linearcontrol may require a more complex,larger solution. Since the circuit sizereduction for analog components on theintegrated circuit is not as great as it isin digital logic, adding digital controlunder some conditions may yield asmaller solution size without sacrificingcapability. With a dense digital logicprocess, greater control over the powerregulation can be maintained than is

currently available in a pure analogprocess. This includes the ability to re-program a standard power managementsolution for a variety of applicationrequirements as well as add nonlinearcontrol loops to improve load regulationover wider operating conditions.

Adding dense digital logic to a con-ventional power process also enhancesmany power management features. Forexample, a high-speed digital communi-cations interface on the power manage-ment unit (PMU) allows the system toadjust output voltages, modify the powersequencing and obtain the status of theconverter. This control and status canlead to better efficiency and thermalmanagement. Adding on-chip non-volatile memory further enhances thePMU by allowing flexible system config-uration and improved analog perform-ance through precision voltage refer-ences and timing control trimming.Power management IC’s with multiplepower converters capable of supplying 1to 2A can exist cost-effectively on thesame die with dense digital logic,improving system control, conversionefficiency and load regulation over widerinput and output conditions.

In summary, power processes withdenser digital capability together withchip-scale packaging will help shape theportable power market for years tocome. Combining the increased “digitalpower” capability along with the ability todeliver solutions in a variety of footprintswill enable system developers to offermore portable systems features andcapabilities in their portable systemswhile driving solutions sizes smaller.

By Dave Heacock, Vice President of Portable Power, Texas Instruments

www.ti.com

12 Power Systems Design Europe June 2004

COVER STORY

Ensure the functionality of the power grid systems

Design concept for an active Power Factor Correction with power modules with special focus into power loss minimizing and inrush current control.

By Ernö Temesi, Michael Frisch and Peter Sontheimer, Tyco Electronics

As the application of electronicpower modules is constantlyincreasing, new problems arearising beyond the actual discussionabout the power factor (cos w) inconnection with the Power FactorCorrection (PFC).

Why Power Factor Correction?Generally, mains AC voltage is trans-

formed into DC voltage via RectifierBridge and capacitor. This configurationis used for most power electronics applications.

In this configuration, the DC-link cir-cuit will only be loaded, when the sinusmaximum of the mains voltage (VIN) ishigher than the voltage at the Dc-linkcapacitor (VCap) circuit voltage. This

leads to short and very high currentpulses (ICap), which may interfere other users in the public power supply network.

Actions have to be taken to ensurethe functionality of the nationwide power grid systems with a fast growingnumber of power electronics connected.Applicable international standardsdemand a sinusoidal current drain.Thus, the developers of power applica-tions to be connected to the publicpower supply network are now chal-lenged to care for its realization.

Challenge and Chance For the development of applications

with sinusoidal current consumptionmore design work will be required thanever before. New national and interna-tional standards and laws demandincreased activities. And compliance of the applicable standards is a musttoday. But an active PFC also generatesadditional advantages, which does notgenerally lead to additional costs.Precondition is a system design thatuses the advantages of an active PFCas smaller DC-link capacitor, loss reduc-tion in the application connected to the

Figure 1. DC voltage via RectifierBridge and capacitor

Figure 2. Current pulses.

14 Power Systems Design Europe June 2004

COVER STORY

output achieved by the increased andconstant output voltage. Applicationexamples of active PFC solutions areinverter welding machines, as by appli-cation of a PFC, the performance canbe increased without affecting the mainsfuse. Also converter-controlled fans forclean rooms should be mentioned. Inclean room production facilities typicallyhundreds or thousands of such ventila-tors are used and the current drain fromthe mains have to be controlled in orderto keep the system in function.

General RequirementsSome general requirements are

mandatory for all these applications:compact design, low interference level,and power loss optimization. To realizean optimally compact PFC design, theswitching frequency must be maximizedso that extremely small PFC choke coilscan be applied. To control by thiscaused increased power loss, the coilshould have a high-performance corewith thermal contact to the cooling ele-ment. New semiconductors must besuited for higher performance in order to gain smaller space requirements. Atthe same time it has to be preventedthat the advantage is lost by thedemand for a larger cooling element.Here new technologies of the semicon-ductor industry are opening doors. Highswitching frequencies are leading toconsiderable cost savings. A compactdesign leads to completely newmechanic concepts for an application,which may also lead to considerablesavings in system costs. Also a positiveinterlinking effect can be utilized. Higherswitching frequencies allow the applica-tion of smaller components for chokeand EMC filter. By its compact design,EMC compatibility will be easier. Andthis will lead to further savings in systemcosts and development time.

Fundamentals of Active Power Factor Correction

An active PFC switch is basically anAC/DC converter, as its core is a stan-dard SMPS (Switch Mode PowerSupply) structure, which controls thecurrent supplied to the consumer via a“Pulse Width Modulation” (PWM). ThePWM triggers the power switch, which

separates the intermediate DC voltagein constant pulse sequences. This pulsesequence will then be smoothened bythe intermediate DC capacitor, whichgenerates DC output voltage.

PFC—Boost Switch (Boost Topology)The boost topology acts as boost con-

verter by converting the input voltageinto a higher output voltage. This switch-ing system is mainly used for PFC.

Two different modulation procedurescan be applied, the continuous modeand the discontinuous mode.

Discontinuous ModeIn the discontinuous mode the transis-

tor is switched on only then, when theenergy contained in the choke coil iscompletely transferred via the diode tothe electric DC output circuit. Whenswitching on the transistor the chokecontains neither energy nor current. Thisoperation principle has therefore theadvantage that switch-on losses will notbe generated. Another benefit is that thechoke can be very small. But this princi-ple generates strongly increased wavi-ness and switch off losses.

Continuous ModeThe topology is the

same as in the discontin-uous mode, but herevaries the current in thechoke closed to the sinu-soidal average. In thecontinuous mode theselarge upper waves andthe very high switch-offlosses (caused by thedouble as high switch-offcurrent) can be avoided.Therefore, this processwill be preferably beused for capacities fromabout 250W.

Figure 3. PFC—Boost Switch.

Figure 4. Discontinuous mode.

Pho

to: I

nter

natio

nal R

ectif

ier

www.powersystemsdesign.com16 17

COVER STORY

Design of a Continuous Mode PFC

Switch-On LossesIn standard PFC switches normally

relatively low flow losses are generatedcompared to the switching losses.Consequently mainly the transistorswitching losses are limiting the maxi-mal switching frequency. On the otherhand, the reverse current of the diodehas a great influence on these switchinglosses generated in the power switch.

Reverse currents of the diode

increase the switch-on losses of thetransistor and the application of a fastboost diode with a low reverse recoverycharge (Qrr) can thus be of great impor-tance. The boost diode has a significantinfluence into the switch on losses of thepower transistor in the boost circuit.

Switch-Off LossesAny parasitic inductance in the output

causes a voltage-overshot witch endan-gers the semiconductors and increasesthe switch off losses. Switching-off thetransistor results in a current change.This causes a voltage spike by the cur-rent change in the parasitic inductancesaccording to:

VCE(peak) = VCE + L x di/dt

To minimize switch-off losses, preven-tion of parasite induction loops at thePFC output is of great importance. Aparticularly elegant solution is to shortthe inductive loop with a capacitorattached as closed as possible. A fastcapacitor integrated into the power com-ponent would be the optimum.

MOSFET vs. IGBTIGBT or MOSFET? To answer this

question, the efficiency of a frequency-dependent technology isn’t a suitablestandard. The reason is the switching of variable currents. In the range of thesinus maximum, only a short pulse willbe required to increase the voltage frome.g. 325 V (VPeak at 230 VAC) to 400 VDC. In the zero flow range the pulse fre-quency is higher, but the current to beswitched is then lower. For 230 VAC/ 400 VDC applications and switching frequencies of 60 kHz and more theMOSFET seems to be the better andcheaper solution.

Applications with focus on a wideinput range (e.g. 90 VAC, 240 VAC) gen-erate a completely different result. Whenapplying 90 VAC at the input and 400VDC at the output, also in the sinus maxi-mum, a relatively long pulse will berequired for the transformation from 127V (VPeak at 90 VAC) to 400 VDC. For suchapplications the static losses are moredecisive for the performance balance ofthe PFC level. For switching frequenciesup to nearly 100 kHz, application of veryfast IGBT’s seems to be more attractive.To optimize system costs, switching andflow-through losses shall always benearly equal. As a rule, PStat/PSwitch = 1shall apply. Balancing of switching loss-es with flow-through losses shall alwaysbe the target of product designers inlow-cost developments.

Power Systems Design Europe June 2004

Input Start Switch as Short-CircuitProtection and Protection againstHigh Switch-On Currents

Output current rates of > 500 Wrequire a specific input control, whichlimits the higher start-up current gener-ated by the loading procedure of theDC-link capacitor after activation of thePFC. The switching system, for which apatent has already been filed, will notonly solve this problem - moreover italso provides a short-circuit protection atthe output (Figure 9).

Description of functions:1. After application of AC voltage at

the input, the SCR´s of the semi-con-trolled input rectifier are not activated.The output capacitor is loaded via a current limiter and auxiliary rectifier. Assoon as the controller begins to contactthe PFC transistor, the PFC coil will beloaded with current. When switching-offthe transistor, the output voltage of thePCF coil will be limited to the initial out-put voltage level at the PFC capacitorvia the PFC diode.

2. When the voltage is high enough,the voltage reduced by the winding ratioN1/N2 via the auxiliary winding at thePFC coils activates the semi-controlledrectifier, in order to limit the losses at thecurrent limiter.

3. In case of a short-circuit at the PFCoutput, the PFC diode clamps the volt-age at the PFC coil down to zero anddisables the semi-controlled rectifier.The current limiter reduces the short circuit current.

PFC Solution with Tyco’s PFC-IPMThis intelligent power module (IPM) is

a complete, universally applicable PFCsolution for currents up to 1 kW. ThePFC-IPM is designed t reduce the inputcurrent upper waves and has therequired features to realize a perform-ance factor > 0.99. The switch is basedon a boost topology, provided with asemi-controlled input rectifier (Figure 10).

This solution has an input start switchto limit the input current and a short-cir-cuit protection for consumers connected

to the output. Another characteristic feature is the zero load capability. Thismeans that the PFC-IPM can generate a stable DC output voltage without thenecessity that a consumer must be connected to the output—a demandedmust, when the application requires ahighly stable output voltage (Figure 11).

The following characteristic featureshave been realized in this flexible, uni-versally applicable PFC solution:

• Nominal input voltage 230 VAC• Output voltage 400 VDC• Output current (zero load) 0.1 kW• Efficiency ca. 95%

(including the Choke)• Power factor > 0,99• Starting current limiter < 12A• Short-circuit protection for static

and dynamic failures at the output

PFC Solution with Tyco’s flowPFC 0The following characteristic features

have been realized in this flexible,power integrated concept:

Figure 6. Switching On Losses.

COVER STORY

Figure 5 Continuous mode.

Figure 7. Switch Off losses.

Figure 8. Short the inductive loop. Figure 9. Switching System.

Figure 10. IPM PFC solution.

Figure 11. IPM Device Picture.

18

Features:• All semiconductors integrated in one

module, no additional efforts for thermal contacting required

• Integration of a temperature sensorfor detection of the substrate tem-perature

• The proven flow concept allows acompact PCB design. Plugs withsimilar voltage are concentrated to voltage islands

• Symmetric design of the PFC transistor for parallel or alternatingoperation

• Low-inductive current measurementwith a shunt for precise control of the PFC switch

• Capacitor for low-inductive bypassingof the high frequency in the module

PFC-Solution with Tyco’s flowPIM+PDrive solutions with PFC require the

additional integration of one fast switch-ing transistor and boost diode.

Similar to the standard TYCO flowPIMmodules, all power semiconductors areintegrated in one module, so that noadditional action for the handling of the power dissipation to the heat sink is needed.

Features:• 1 Phase Input Rectifier • PFC Transistor + extreme fast Diode• 3 Phase Inverter IGBT + FRED• HF-Capacitor in DC Link • Current sense shunt in the DC–• Current sense shunt for PFC

controlling in the DC–

• NTC temperature sensor• Approved flow concept for easy

routing off the system PCB• Clip In for mechanical fixing into

the PCB

Advantages of PFC solutions withtyco Modules

The Tyco PFC modules offer compactmodules and the flow concept offeropportunity to get a compact and easyPCB design. All Tyco PFC solutionsshort the high frequency with a ceramiccapacitor inside the module. The excel-lent EMC behavior is only possible withmodule solution and can never achievedwith discrete components. The Clip Inhousing of the modules flowPFC0 andflowPIM0 is adjustable to different PCBand makes the assembly easy and reli-able. The integrated temperature sensorprotects the module and the application.The UL listing of the module shortensthe time to get the application certified.These advantages are a basis for an innovative and cost effectivePFC solution.

Figure 12. Flow PFC solution.

Figure 13. Flow Device Picture. Figure 14. TYCO flowPIM modules.

www.tycoelectronics.com

Power Systems Design Europe June 2004

COVER STORY

www.powersystemsdesign.com 2120 Power Systems Design Europe June 2004

POWER SEMICONDUCTORS

Differences in MOSFET cell struc-tures and technologies providethe power electronics designcommunity many options as to theselection of an appropriate device for agiven application. Medium transconduc-tance (MT) MOSFETs may be appropri-ate for some applications requiringextended periods of linear operation,whereas high transconductance (HT)MOSFETS may be the best device forother applications. However, failure toconsider differences in gate charge and thermal impedance may result indifferent operation.

Differences between Medium and High Transconductance Power MOSFETs

HT MOSFET technology has a higher channel density than MTMOSFETs. Structural geometric differ-ences (Figures 1 and 2) depict a signifi-cantly reduced die specific on-resist-ance (Rsp) for the HT structure, render-ing these devices superior to the MTstructure of Figure 1 for switch modeoperation. Circuit design challengesoriginate from failing to consider theeffect from HT technology’s higher gatecharge per unit area, higher transcon-ductance, faster switching slew rate,increased thermal impedance from the use of a smaller die, reduction of

device safe operating area (SOA), andlower unclamped inductive switching(UIS) capability. These differences arediscussed in this article.

Issues guiding the selection of MOS-FET technology can be roughly brokendown into three application categories:

• high frequency switched modepower supplies

• switch mode motor control • linear mode regulation

High Frequency Pulse WidthModulation (PWM) Power Supply

HT devices are best suited to achievesystem efficiency improvement. Toachieve best operating performance, the gate drive circuitry may be optimizedwhen substituting an MT device. A figure-of-merit (FOM) frequently referenced for device comparison is:rDS(ON)(TJ)HT x QG_HT ≤ rDS(ON)(TJ)MT xQG_MT as gate charge becomes animportant loss contributor at high switching frequencies.

PWM Motor ControlBoth MT and HT MOSFET devices

are appropriate technologies. Equivalentor lower loss performance is possiblewith smaller die size HT MOSFETs. Itmay be incorrect to simply use the samerDS(ON) HT device when substituting aMT device. An rDS(ON) reduction may be necessary to maintain equivalentperformance. An approximation fordetermining HT rDS(ON) requirement is to use a power–density-equivalency

Linear and switch mode applications are the focus

HT MOSFET devices offer faster switching slew rate performance and reduced conduction lossesrelative to MT, desirable for SMPS power supply design and SMPS motor control,

but not beneficial for linear mode regulators.

Alain Laprade and Alex Craig, Fairchild Semiconductors

Figure 1. Medium transconductanceMOSFET using planar technology.

Figure 2. High TransconductanceMOSFET using trench technology.

22 Power Systems Design Europe June 2004

where: rDS(ON)(TJ)HT x RuJC_HT ≤rDS(ON)(TJ)MT x RuJC_MT.

Differences in gate charge character-istics are typically not a significant con-sideration as motor drive MOSFEToperating frequencies rarely exceed 20KHz. The gate drive circuit may requireoptimization when replacing MT deviceswith HT for similar EMI system perform-ance with controlled switching slew rate.Snubbers are useful for this purpose.

Use of parallel HT devices in applica-tions meant to have slow slew rateswitching transitions may be more diffi-cult. The consequence of mismatchedthreshold voltage (VGS(th)) betweendevices must be carefully considered toavoid inter-device gate oscillation.Under some circumstances, oscillationcan occur in spite of the use of individ-ual gate decoupling resistors. Layoutmodification should also be consideredparticularly if one is replacing a numberon MT devices with a lesser number ofHT devices.

Linear Mode RegulationIt is incorrect to attempt die size

reduction for linear mode regulation byusing HT devices as an alternative toMT devices. Die volume (ZuJC) is theonly factor of consequence when per-forming junction temperature calcula-tions. There are no operational benefitsfrom the use of HT technology.

It is not possible to generalize theeffect of MT and HT MOSFET technolo-gies on device performance. Die size,packaging technology, layout parasitics,

temperature dependent transconduc-tance and gate threshold voltage char-acteristics, and SOA operating pointinfluence performance. Power applica-tions with die sizes requiring larger pack-ages (e.g. TO220, TO263) are subject touneven die surface temperature distribu-tion, and some current focusing mayresult. The negative temperature coeffi-cient of the voltage transfer characteristicaccentuates the extremes in die surfacetemperature, resulting in increased cur-rent density focused at the die thermalcenter. HT devices are more susceptibleto this than are MT devices. MT technol-ogy may be the more appropriate tech-nology in applications having high linearmode power dissipation.

HT MOSFETs are optimized for highfrequency operation. A consequence of optimizing MOSFET die designs forhigh frequency operation is that devicetransconductance is greater than that ofMT devices, rendering them more sensi-tive to electrical noise and susceptible to high frequency oscillation with circuitparasitics. For same die size devices,HT device SOA is less than that of MT. It is possible to design a device usingtrench technology optimized for linearmode operation with increased Rsp as atrade-off. At the time of this writing, suchdevices are commercially unavailable.

Other IssuesUnclamped inductive switching (UIS)

performance of HT technology is lessthan that of MT for same die sizedevices. As HT devices offer reducedrDS(ON) at a reduced die size, use of suchdevices are often made and occasional-

ly encounter problems when testing out-side normal operation. While UIS energycapability depends on die size in a givenapplication, the temperature at whichthe device fails is a function of the epidoping. The energy required to reachthat temperature is a function of timespent in avalanche (tAV) and avalanchecurrent (IAS); Avalanche energy EAS = _ x 1.3 x BVDSS x IAS x tAV.

MOSFET SOA capability is die sizeand the channel density dependent.Failure temperature is also a function of epi doping. Depending on the SOAevent, the current, voltage, and timerequired to reach failure point in HTdevices may be less then that requiredfor a same size MT device.

Circuit designers are often unaware ofthe device die sizes, so careful consid-eration must be given to avoid devicefailures. MOSFET manufacturers mustassist the circuit designer to avoid such failures. To this end, FairchildSemiconductor provides SOA (Figure 3)and UIS rating graphs (Figure 4). Thecriteria for safe usage are to find the circuit’s peak load current and tAV, andto plot the information on the UIS graphin the data sheet example shown inFigure 4. If the operating point is belowand to the left of the appropriateTJ(START) line, the part is being usedwithin its capability. Also provided is a simple transient thermal impedancemodel which may be used to obtain a reasonable junction temperatureapproximation during transient events.

HT MOSFET devices offer fasterswitching slew rate performance andreduced conduction losses relative toMT, desirable for SMPS power supplydesign and SMPS motor control, but not beneficial for linear mode regulators.HT device selection as a replacementfor an MT device in PWM designsshould be performed using a power-density-equivalency or figure-of-merit.Not to be forgotten is the potential effectof UIS in a circuit using HT technology.

Figure 3. Forward bias SOA curve. Figure 4. Unclamped inductiveswitching capability.

www.fairchildsemi.com

Get your own copy of Get your own copy of

Sign up today online at www.powersystemsdesign.comSign up today online at www.powersystemsdesign.com

your trade journal for the power engineer!your trade journal for the power engineer!

Pho

to c

olla

ge: M

icro

sem

i Cor

pora

tion

POWER SEMICONDUCTORS

www.powersystemsdesign.com24 25Power Systems Design Europe June 2004

, eq. 6

By solving eq. 6 and plugging I intoeq. 5, we obtain the complete transferfunction of the BUCK affected by staticlosses:

, eq. 7

Equation 7 shows that losses areweighted accordingly to the timesequence in which they play: Rlf is pres-ent during both Ton and Toff whereasRon and Rd are respectively active during Ton (D multiplication) and Toffonly ([1-D] multiplication).

In previous averaged models, thestate-space averaging technique orswitch waveforms analysis were usuallyapplied over perfect elements, non-inclusive of the above ohmic losses.However, if these elements play anactive role in the DC transfer function,they affect the small-signal AC analysisquite significantly by introducing variousdamping effects. In a paper presented in

PCIM Nuremberg 2001, Sam Ben-Yaakov from the Ben-Gurion University(Beer-Sheva, Israel) presented his mod-ified Generalized Switched InductorModel (GSIM) where all conductionlosses were modeled [1]. Without enter-ing into the details of the model deriva-tion, Ben-Yaakov did not depart from hisoriginal model but added in series withthe inductor the losses specific to agiven time interval (e.g. Vf and Rd dur-ing D' in a BUCK, etc.). One very inter-esting feature consisted in including thetrue diode SPICE model and the realMOSFET RDS(ON)@Vgs if necessary.Figure 2a shows how to implement thismodel in a BOOST voltage-mode appli-cation (see Figure 2a and 2b).

The internal Ron was passed as astandard resistor parameter to keep thesimplest implementation. The diodemodel is however kept external. A DCsweep was performed on the schematicwhere VDon was swept up to 900mV(90% duty-cycle). Figure 2b reveals theresults showing the latch-up characteris-tic of the BOOST when the ohmic lossesbecome more significant compared tothe load. The above model subcircuitnetlist can be downloaded in bothIsSpice4 and PSpice from the authorwebsite [2]. Other models include lossyBUCK in voltage and current-mode

control. Please note that more complexmodels including dynamic switchinglosses were derived by the ColoradoPower Electronics Center (CoPEC). Anextensive documentation related to thesubject, including tutorials, is availablefrom the university web site, http://ece-www.colorado.edu/~pwrelect.

Christophe BASSOON Semiconductor 14, rue Paul Mesplé - BP1112 - 31035 TOULOUSE Cedex 1 - France33 (0)5 34 61 11 54 e-mail: christophe.basso@onsemi.com

1. “The Generalized Switched InductorModel (GSIM) Accouting ForConduction Losses”, Sam Ben-Yaakov,Isaac Zafrany, PCIM Nuremberg 2001

2. http://perso.wanadoo.fr/cbasso/Spice.htm

POWER SUPPLY/SIMULATION

Figure 2a. The BOOST model features a true diode SPICE model.

Figure 2b. Vout versus D exhibit the classical BOOSTlatch-up effect.

When one derives the equationsrelated to a given topologyoperation (e.g. BUCK, BOOSTetc.), perfect or ideal elements are usu-ally installed on the schematic and vari-ous drops are omitted in the calcula-tions. If this option obviously simplifiesthe analysis, it perceptibly alters the DCoperating point as well as the ACresponse. Figure 1a and 1b depicts aBUCK converter that includes lossesattributed to a) the inductor wire resist-ance, Rlf b) the power switch RDS(ON),Ron c) the diode forward drop, Vf and its dynamic resistance

Rd = = :

When the switch closes, the inductorvoltage VL and capacitor current Ic can bedefined through the following equations:

, eq. 1

, eq. 2

At the switch opening, the current Ikeeps circulating in the same direction,but now flows through the free-wheeldiode to keep the amp-turns constant inthe inductor. Equations become:

, eq. 3

, eq. 4

Theory dictates that the average volt-age across the inductor must benull when the converter has reached theequilibrium: = D . VLON + D' .VLOFF = 0, which by combining eq. 1and eq. 3 gives:

, eq. 5

The above statement regarding theinductor average voltage also translatesto a capacitor where it's average current shall be null when the converteroperates in steady-state: = D .IcON + D' . IcOFF = 0, which by combin-ing eq. 1 and eq. 3 gives:

Standard averaged models do not include the various losses that degrade the converter efficiency but also it’s small-signal AC response.

Christophe Basso, ON Semiconductor

Including the true diode SPICE model

Figure 1a. Component arrangement during the ON time. Figure 1b. Component arrangement to the switch turn-off.

www.onsemi.com

POWER SUPPLY/SIMULATION

www.powersystemsdesign.com 2726

CAPACITORS

Power Systems Design Europe June 2004

If you frequently travel from place toplace via tram or the underground,you are familiar with the problem ofpower outages. Travel suddenly comesto a stop, delays occur, and within a few minutes there may be several trainsstuck one behind the other somewherebetween stations—typically at peakhours, when all of the coaches are full.For individual passengers, this initiallyonly causes a few minutes’ unpleasantdelay, but in a sort of chain reaction,such an incident can affect large por-tions of the rail network, particularly ifseveral trains must resume their jour-neys after being backed up in a tunnel.If they all draw energy from the samenetwork at practically the same time,they can pull the network voltage belowa critical level.

The local transport authorities of several European cities such asCologne and Madrid, as well as severalmetropoles in the US such as Portland,Oregon, are now tackling this previouslyunsolved problem with an innovativeenergy storage system that is alsodesigned to recover braking energy.This system is called SITRAS SES, and it was developed by the engineersof Siemens Transportation Systems. It allows system operators to achieveenergy savings of up to thirty percent.SITRAS SES also makes a decisivecontribution to stabilising the network,which not only enhances the reliability ofmass transit systems, but also improvesthe tempers of passengers.

Underground trains that feed brakingenergy back into the electricity supplysystem first entered regular servicearound twenty years ago. When such atrain brakes, its electric motor acts as agenerator and feeds the regeneratedenergy back into the supply lines.However, this excess energy can onlybe used if there is an increased energydemand somewhere else in the network,which can for instance arise from a trainjust starting off. Otherwise, only approxi-mately sixty percent of the regeneratedenergy can be used in normal operation.The remainder is dissipated as heat inthe braking resistors of the vehicles,without being put to good use. Sinceenergy costs form a significant portion ofoperating budgets, amounting to 25,000to 150,000 euros per year depending on

the vehicle type, system operators havea major interest in reducing their costsby achieving energy savings.

Around four years ago, the engineersfirst started thinking about an energystorage system that could absorbbraking energy and release it later on to trains that are starting to move.Simulations and practical tests in vari-ous cities showed that using a suitableenergy storage system operationally foraround 22 hours per day could reducethe annual primary energy demand by as much as 500,000 kilowatt-hours or 30 percent. That corresponds to areduction of CO2 emission of 300 tons.

Before choosing ultracapacitors asenergy storage source the engineersfirst decided to use flywheel storagesystems. But already after the firstextended service tests, it was clear thatthey were not suitable for long-term use,due to their complex maintenance.That’s why the engineers then concen-trated on developing capacitor energystorage systems.

The Maxwell ultracapacitors used inthe SITRAS SES system have a capaci-tance of 2600 farads with a size of asmall soda can. Each device is operatedat a voltage of 2.3 volts. The completeenergy storage unit contains approxi-mately 1300 of these 2.3-volt units. Thesystem provides a peak power capacityof one megawatt.

Ultracapacitor energy storage systems help mass transit systems avoid sudden power outages,and as a pleasant side effect, they yield energy savings of up to 30 percent.

By Dr. A. Schneuwly, J. Auer, G. Sartorelli, Maxwell Technologies

Concentration on capacitor energy storage systems

Figure1.A SITRAS SES unit canabsorb the braking energy released by all stopping trains within a radius of up to three kilometres.

Figure 2a and b.A glimpse into theheart of the SITRAS SES system: The2600 farad BOOSTCAPs can be easilyrecognised.

The system, which includes a connec-tion unit, a voltage converter and controlelectronics in addition to the capacitors,is housed in two rows of cabinets, eachof which is 3 metres long and 2.7 metreshigh. If one or more trains start at thesame time, the SITRAS SES systemreleases the energy absorbed duringbraking, and the network voltage neverdrops below the critical level. The sys-tem thus not only saves energy, with itsfast response time it also prevents asudden loss of power for the trains andthus avoids leaving hundreds of frustrat-ed passengers sitting stranded.

Ultracapacitor technologyIn Maxwell ultracapacitors, carbon

powder electrodes are combined with anorganic electrolyte. By using extremelylarge surface area carbon electrodes,such capacitors can contain an activesurface area corresponding to two foot-ball pitches in a volume of half a litre.Compact in size, ultracapacitors canstore an incomparably higher amount ofenergy than conventional capacitors.Indeed, ultracapacitors from Maxwelloffered under the trademark BOOST-CAP are currently available on the mar-ket with capacitance ranges from pris-matic 5 and 10-farad cells up to cylindri-cal 2600 Farads. They can repeatedlyprovide very high power pulses,recharge as fast as they are discharged,while little affect on the life of the prod-uct. Because they are capable of cyclingmillions of times, they are virtually main-tenance-free over the life of any productin which they are used. As a result, theyneed not be disposed—making them avery “green” form of energy storage.

Maxwell is able to supply BOOSTCAPsin volumes and at price points that areopening numerous market opportunitiesfor energy storage and peak powerdelivery. To facilitate adoption of ultraca-pacitors for applications which requireintegrated packs consisting of multipleultracapacitor cells, Maxwell providesfully integrated power packs that satisfy

the energy storage and power deliverydemands of large systems.

Our large cell ultracapacitors havebeen designed into industrial applica-tions such as uninterruptible back uppower systems, pitch systems of windturbines and transportation applicationssuch as hybrid buses and trucks, electri-cal rail systems and capacitive startingsystems for diesel engines. Our smallcell ultracapacitors have been designedinto consumer electronics such asremote transmitting devices, digital cam-eras, bar code scanners, computermemory boards and transportationapplications such as electrical rail alarmsystems and electric actuators, or latch-es, for aircraft and automobile doors.Many of the end products into whichBOOSTCAPs have been designed noware ramping into commercial production.

Product newsMaxwell recently developed a new

high power ultracapacitor that features a revolutionary case design. TheBCAP0350 D cell BOOSTCAP is thefirst in a series of new ultracapacitors tobe standardized on battery-sizing todrive down the costs and ease the inte-gration of the technology. By standardiz-ing its ultracapacitors, Maxwell is reduc-ing its manufacturing costs by more than50 percent and passing that savings

directly to OEMs, as well as reducingtime-to-market by providing engineers a known form factor for seamless, rapidproduct integration.

Figure 3a and b.New high powerBCAP0350 D cell BOOSTCAP and itsRagone plot showing the specific powerdensity against the specific energy density.

The 350-farad ultracapacitor isdesigned for maximum power through-put. Weighing only 60 g, it features apower density of up to 25 kW/kg andenergy density of 5.1 Wh/kg. The newBOOSTCAP is designed for a rated volt-age of 2.5 V and can withstand a peakof 2.8 V. The operating temperaturerange extends from -40 to 70 °C. Thanksto the outstanding performance as wellas the low cost design this cell is ideallysuited for application such as automotivemultiple zone electrical distribution sys-tems, wind turbine pitch systems, aircraftdoor opening systems, actuator applica-tions, medical systems, battery supportapplications in general and many highvolume consumer applications.

Authors:Dr. A. Schneuwly, J. Auer, G. Sartorelli, Maxwell Technologies SA, 1728 Rossens, Switzerland

Figure 2a.

Figure 3a. Figure 3b.

Figure 2b.

www.maxwell.com

CAPACITORS

www.powersystemsdesign.com28 Power Systems Design Europe June 2004

Power requirements for high-per-formance microprocessors havebecome increasingly demanding(see Figure 1). Prior to the Pentiumclass of processors, 5 Volts directly fromthe AC-DC silver box was used as theprocessor core voltage. As core voltagesdropped and current demands increased,the industry moved to using 12V fromthe silver box as the input voltage to asynchronous Buck converter to generateVcore. In an attempt to decrease thehigh ripple associated with regulatingVcore from 12V, the use of a multiphasebuck converter became the standardstarting with Pentium 4 class proces-sors. In order to satisfy processor powerspecifications, such as soft start, powersequencing, VID and load line specifica-tions, dedicated multiphase controllerICs like the HIP6301 from Intersil wereintroduced. The ICs were conceivedbased on analog PWM technology.

In addition to voltage and current con-trol, the new multiphase controllers wererequired to satisfy many other specifica-tion requirements: VID programming,load line regulation, power sequencing,phase current balance, and monitoringand protection. The PWM function in the multiphase controller is becoming a

small part of the controller functionality.This demand for more functionality from the controller and the continuingadvance of digital technology haspushed many companies to look at digital multiphase control. The proposedsolutions vary from the use of digitalprocessors (DSP), microprocessors,and microcontrollers, to the latest soft-ware-programmable mixed-signal IC,the ISL6590 introduced by Intersil.

Benefits of Digital ControlDigital controllers offer many advan-

tages over their analog counterparts:improved system reliability, flexibility,and ease of integration and optimiza-tion. Overall, they offer an elegant solu-tion to many requirements in the Vcorepower regulation specifications.

A system based on a digital controllerrequires fewer components, whichdecreases the mean time before failure(MTBF) of the system. For example, all

Digital controllers offer many advantages over their analog counterparts: improved system reliability, flexibility, and ease of integration and optimization. The use of software to change

the controller functionality makes a system based on a digital controller very flexible.

Zaki Moussaoui and Greg Miller, Intersil

Digital controller requires fewer components

POWER ICS

the components for the feedback loopare eliminated; the select on test andselect according to design specificationcomponents are also replaced by soft-ware programming. A change to thedesign to meet a new requirement maynot require new board layout and moreengineering time; the changes could beimplemented in software.

The added capability of monitoringprotection and prevention will alsoincrease the system reliability. Forinstance, an engineer can choose tomonitor the system temperature todecrease the current limit level, or turnon a fan. This scenario will decrease thestress on the power components andfans that in turn will improve the systemreliability and would eliminate over spec-ification of components.

The use of software to change thecontroller functionality makes a systembased on a digital controller very flexi-ble. The digital controller offers the abili-ty to add, eliminate or change any sys-tem parameter in order to meet newrequirements, or to optimize and cali-brate the system. For example, thesame voltage regulator model (VRM)can be programmed to meet differentprocessor specifications such as LoadLine (LL), voltage identification (VID),and current or voltage requirements,without any hardware changes. Due tothe ease of integrating communicationcapability, the digital controller also facil-itates the ability to integrate and cas-cade multiple systems together. Forexample, in multi-VRM boards, currentsharing could be implemented through astandard communication bus without theneed for any hardware additions.

Digital Control IC ImplementationIn order to choose the ideal digital

controller IC for the application, thepower supply engineer will have takeinto consideration the performance andthe capabilities of many digital IC blocksthat normally do not exist in an analogcontroller IC. Three blocks in particularform the heart of a digital controller: ananti-aliasing filter, an analog to digitalconverter (ADC), and a digital pulse

width modulator (DPWM). See Figure 2.

a) Anti-Aliasing FilterThe Sampling Theorem indi-

cates that a continuous signal canbe properly sampled, only if it does not contain frequencycomponents above one-half of thesampling rate. So, if the samplingrate is 2MHz, then an anti-aliasingfilter with –20dB gain at 1MHzwould be practically expected.The importance of the anti-alias-ing filter arises from its effect onthe total system bandwidth. Forexample, assume that we onlycan have an RC type of low passfilter, with transfer function of the form:

To have –20dB gain at 1MHz,half the sampling frequency of2MHz, the pole of H_Alias has tobe at 100kHz, which will introducea phase lag of 45° at 100kHz. If apower stage is switching at 1MHz,an analog control loop with acrossover frequency of 200kHzcan easily be obtained. With thisdigital implementation, thecrossover frequency will be limit-ed to below 100kHz if we do notallow any phase lag to be intro-duced in the feedback loop by theanti-aliasing filter. To solve thisproblem we could use an activefilter with more poles, or increasethe sampling frequency.

b) A/D Converter The A/D Converter block con-

sists of two main sub-circuits: thesample and hold and the ADCitself. The sample and hold blockadds a delay in the loop accord-ing to that can beapproximated as:

Figure 1. Voltage/ Current Road Map.

30 Power Systems Design Europe June 2004

where TADC is the A/D Convertersampling period.

In order to decrease the delay, i.e.decrease the phase lag, the samplingperiod TADC of the A/D Converter needsto be increased.

A wide choice of ADC architecturesexist that differs in resolution, band-width, accuracy, and power require-ments. The major ADC architectures areflash (all decisions made simultaneous-ly), successive approximation (where asuccessive approximation shift registeris the key defining element), andpipelined with multiple flash stages.Each has it own unique set of pros and cons.

The Flash Architecture: Sets of 2n-1comparators are used to directly meas-ure an analog signal to a resolution of n bits. The flash architecture has theadvantage of being very fast becausethe conversion occurs in a single cycle.The disadvantage is that it requires alarge number of comparators; the num-ber of comparators needed for an n-bitADC is equal to 2n-1. For instance, a 10-bit A/D converter needs 1023 com-parators, which makes it hardware inten-sive even for an integrated controller.

The Successive ApproximationsArchitecture: This approach uses a single comparator over many cycles tomake its conversion. The successiveapproximation (SAR) converter onlyneeds a single comparator to realize ahigh resolution ADC, but it requires ncomparison cycles to achieve n-bit reso-

lution. SAR has a major disadvantagesince it takes too many cycles to convertthe analog signal, which will introduce alarge phase lag in the feedback loop.For example, a 10-bit conversion willintroduce 10TADC delay.

The Pipelined architecture withMultiple Flash: This is the middleground between the fast flash converterand the high resolution SAR. A pipelinedconverter divides the conversion taskinto p consecutive stages. Each of thesestages consists of a sample and holdcircuit, an n-bit flash converter, and ann-bit D/A converter (DAC). A pipelinedconverter with p-pipelined stages, eachwith an n-bit flash converter, can pro-duce a high-speed ADC with a resolu-tion of k = p x n bits using p (2n-1)comparators. For example, a 10-bit, two stage pipelined converter wouldrequire 62 comparators as compared to1023 for the flash, and will take only twocycles for the conversion compared to 10TADC on a SAR.

From the discussion regarding theanti-aliasing filter and the delay intro-duced by the Pipelined or the SAR A/Dconverter, we see clearly the need forhigh sampling rate.

The number of bits needed can bebased on the resolution of the measure-ment required. In order to satisfy a spec-ified output voltage regulation ∆Vout , theADC resolution has to have an errorless than the allowed variation of theoutput voltage.

If the maximum output regulation is ∆Vout , the maximum voltage of the

ADC is ∆Vout max and the output voltageVout is scaled by a gain G to meet theADC voltage levels, then the least signif-icant bit of the ADC has to be less thanthe product of the maximum ripple andthe Gain.

where k is ADC number of bitsSolving for k we get:

For example, if the output voltage rip-ple ∆Vout is 5mV, the output is 2V, andthe maximum ADC input voltage is 1V, ascaling of G = 0.5 will be required beforethe input to the ADC. At a minimum, anADC with kADC = 10 bits is required inthis case.

c) Digital Pulse Width ModulatorAn analog controller has no inherent

limit on the possible pulse width gener-ated. On the other hand, a DPWM pro-duces a discrete and finite set of PWMwidths. From the output point of view insteady state, only a set of discrete output voltages is possible. BecauseDPWM is part of the feedback loop, it is necessary that the resolution of theDPWM be high enough so that the out-put will not display what is known as alimit cycle. In a limit cycle the output

goes into an oscillation of fixed amplitude and frequency, irrespective of theinitial state.

The minimum number of bits neededso as not to display any limit cycledepends on the topology, the outputvoltage, and the ADC resolution. Forexample, in the buck converter:

Vout = DVin where D is the Duty Cycle.

If we differentiate the above equation,we get:

∆Vout = Vin ∆D

Replacing Vin by Vout

, we get:

Therefore, if we have kpwm bits in ourPWM modulator, the least significant bit is:

From the preceding paragraph, theminimum output ripple is:

Using the result from above we get:

Where kADC is the ADC number of bits.

As was shown in the previous exam-ple, if ∆Vout = 5mV, the output is 2V, themaximum ADC input voltage is 1V, anda scaling of G = 0.5, then a minimum of10 bit ADC is required. With kADC = 10bits and assuming a minimum duty cycleof 0.1 (D = .1), a minimum of 14 bits isrequired for the Digital PWM.

Over the years, power supplydesigners have gained a strongknowledge of how to choose ananalog power supply controller inorder to meet the required specifi-cations. With the introduction ofdigital controllers, the choices arenot as clear. The digital controllerrequires the designer to payattention to more complex ICblocks. In order to achieve thedigital controller advantages ofimproved system reliability, flexi-bility, and ease of integration andoptimization, the power supplydesigner must select the correctICs for the application. The anti-aliasing filter, analog to digitalconverter (ADC), and the digitalpulse width modulator (DPWM)are the heart of the digital con-troller IC. Special design consid-erations of these blocks are keyto achieving superior overall sys-tem performance.

POWER ICS

Figure 2. Major Digital Controller Blocks.

www.intersil.com

D

www.powersystemsdesign.com 3332 Power Systems Design Europe June 2004

POWER MANAGEMENT

The Power Manager series fromLattice Semiconductor is capableof performing all power supplymanagement functions for contemporarylow voltage and multi voltage systems ina single chip. The Power Manager is afamily of in-system-programmable com-ponents which facilitates setting up to 12 monitoring thresholds, setting precisedelays, generating absolutely flexiblemonitoring and sequencing functionsand outputting both logic signals as wellas driving n-channel-MOSFETs. Theprogrammable logic core also controlsthe output voltages of these MOSFETs.This article covers a typical supply volt-age sequencing and monitoring applica-tion from a component design and pro-gramming view.

Being the first mixed signal PLD in the world the Power1208 provides thefollowing programmable modules toimplement the sequencing and monitor-ing functions:

• On-board ispMACH sequence con-troller CPLD

• Delay timers (for sequence delaysand watchdog timers)

• Internal oscillator for precision timingand clocking the CPLD

• Multiple analog inputs with program-mable precision analog threshold

• Buffered comparator outputs• Programmable N-channel MOSFET

drivers (Power1208 only)• General-purpose open-drain digital

logic outputs• General-purpose logic inputs.

A typical applicationFigure 1 shows a typical power

control system, such as that which mightbe used on a PCB board with CPLDsand FPGAs requiring multiple supplyvoltages. In this circuit, the ispPAC-POWR1208 monitors the input supply(+5VIN), and controls the local supplies,as well as the interfaces to on-boardlogic through SYSRESET and SHUT-DOWN signals. To power this systemup, the 3.3V supply must be brought upfirst, the 2.5V supply next, and finally the5V supply. The system should pauseapproximately 10msec between ramping

each supply in order to allow for stabi-lization of the circuitry on the board.Additionally, SYSRESET must be heldLOW until all supplies have been acti-vated. When all the supplies are activat-ed, SYSRESET is brought high. Whenthe SHUTDOWN line is brought HIGH,the power supplies must be turned off in the reverse order; 5V, 2.5V, andfinally 3.3V.

Development Tool is PAC-DesignerThe programmable functions, which

make the ispPAC-POWR so flexible, areeasily configured by use of a specialsoftware tool called PAC-Designer. ThePAC-Designer, currently available asrev. 2.1, offers a hierarchical designentry system, which allows the user todefine all analog and I/O characteristicsthrough the use of Windows basedgraphical user interfaces in conjunctionwith dialog boxes.

For the design of the programmablelogic core inside ispPAC-POWR a frontend called LogiBuilder is used. It is adesign environment that exists withinPAC-Designer and provides a simple,easy-to-learn method for configuring thesmall embedded PLD incorporated with-in the ispPAC-POWR devices. TheLogiBuilder shields the user from the

Programmability reduces prototype-debugging time

Modern electronic systems have complex power requirements; the control and sequencing of the various voltage supplies that are incorporated into these complex systems is a major

problem faced by today’s design engineer.

By Johannes Fottner, Lattice Semiconductor

POWER MANAGEMENT

complexity and details of the underlyingPLD logic by presenting a higher-levelarchitecture, which the design engineeruses to construct the required function.

For building sequence controllers,LogiBuilder provides an instruction setcalled LogiBuilder Sequence Controller(LBSC). Instead of worrying about prod-uct terms and macrocell configurations,the designer develops configurations asa program containing short sequencesof LBSC instructions, which are sequen-tially executed. In addition, LBSC sup-ports both looping and conditional exe-cution through GOTO and IF-THEN-ELSE instructions. WAIT instructions areprovided to support both waiting forspecified events and user-defined

delays. In the ispPAC-POWR1208 abank of four independent timers(TIMER1-TIMER4) is provided for imple-menting user-defined delays.

Looking at the application problem infigure 1 again, from a “programming”perspective the sequence can be para-phrased as follows:

1. Wait for +5V to rise2. Activate 3.3V supply3. Wait ~10msec4. Activate 2.5V supply5. Wait ~10msec6. Activate Local +5V supply7. Bring SYSRESET HIGH8. Wait for SHUTDOWN9. Turn off +5V supply

10. Wait ~10 msec

11. Turn off +2.5V supply12. Wait ~10 msec13. Turn off 3.3V supply14. HALT

As can be seen from figure 2 theLogiBuilder vocabulary contains justthese basic functions:

OUTPUTWAIT FOR WAIT FOR USING TIMER

IF THEN GOTO ELSE GOTO GOTO NOPHALT.

Because LogiBuilder is dialog-driven,as opposed to a free-form text language,it is easy to use, and does not requirelearning a large number of syntacticdetails, as the various dialogs guide theuser in supplying the necessary informa-tion. Not only does it support sequencesbut also a capability called an exception.Exceptions provide the ability for aLogiBuilder program to quickly respondto conditions that are not specified in themain sequence. If we add over-voltagedetection to the example cited above,the circuit must be modified to permitsensing of the output voltages.

As soon as an over-voltage conditionis seen, the specified action will be toturn everything off fast and HALT. This can be accomplished without anychanges to the sequence. Exceptionsare listed in LogiBuilder in the exceptiontable (Figure 3), which is usually displayedimmediately beneath the sequencetable. Each exception occupies one row in the exception table and allowsbranches to a step in the sequence window or to change outputs directly or both.

In addition to supporting the creationof user-defined control sequences,LogiBuilder also supports the implemen-tation of simple logic circuits, both com-binatorial and synchronous designsusing ‘D’ flip-flops can be easily realized.To achieve this, the function of therespective output pin must be defined

Figure 1. Power1208 controls power sequencing and generates supervisory signals.

Figure 2. The first LogiBuilder example

34 Power Systems Design Europe June 2004

as combinatorial or D-type flip-flop, and the equations entered in a windowsimilar to the exception window.

During the design phase, LogiBuilderprovides immediate feedback as soonas inconsistent or erroneous input isidentified. Instead of waiting until thedesign has been entered to report errorsLogiBuilder checks equations for incon-sistencies and reports errors as soon as

the individual equation is entered.During the compilation run LogiBuilderprovides numerous post-design checks.Not only does this result in near-instan-taneous feedback to the user of bothreal and potential problems with adesign, but also by flagging these prob-lems at an early point in the processLogiBuilder can provide concise feed-back as to the exact cause of the prob-lem, and where it is located in thedevice configuration.

If required, for very complex designsthe ABLE code, which is generated dur-ing the compilation process, may also

be edited, PAC-Designer provides ABELediting for these cases, as well as alogic simulation tool.

As the Power Manager componentsare in-system programmable devicesprogramming is performed through a 4-wire, IEEE 1149.1 compliant serial JTAGinterface. Once a device is programmed,all configuration information is storedon-chip, in non-volatile E2CMOS memo-ry cells. In the evaluation and prototyp-ing phase the Lattice ispDOWNLOADcable may be used to connect the paral-lel port of a PC to the part’s JTAG inter-face. The download process itself iscontrolled by the PAC-Designer soft-ware or another suitable JTAG program-ming tool like the ispVM System fromLattice Semiconductor.

Integration and ProgrammabilityBenefits with Power Manager

The power manager family fromLattice Semiconductor greatly simplifiesimplementation of complex power-sup-ply sequencing and monitoring systems

and reduces design time. Programmabilityreduces prototype-debugging time, andby programming the threshold voltages,power supply margining tests can beperformed even on the production line.

The PAC-Designer software packagewhich can be downloaded from theLattice Semiconductor website (www.lat-ticesemi.com) free of charge fully sup-ports the parts´ re-programmability andoffers an easy to learn approach todevice configuration and programming.

Contacts:Lattice SemiconductorJohannes FottnerSenior Analog Field Application EngineerLilienthalstrasse 27D-85399 HallbergmoosGermany

Johannes.Fottner@Latticesemi.com

Lattice Semiconductor UK LimitedBernie PerrinEuropean Marketing Manager1st Floor Melita House 124 Bridge Road ChertseySurreyKT 16 8LHUnited Kingdom

bernie.perrin@latticesemi.com

Figure 3. Exceptions allow reaction at any time.

POWER MANAGEMENT

www.latticesemi.com

www.powersystemsdesign.com36 37Power Systems Design Europe June 2004

MOSFET TRANSISTORS

Today’s DCDC converters can generate large amounts of powerdespite their small size. Powerarchitecture changes, including theintroduction of an intermediate bus voltage rail, have resulted in a shift topoint-of-load (POL) implementations forlower-voltage and non-isolated convert-ers. Consequently, the area