viewpoint - power systems design · 2013-10-16 · viewpoint it is toy train time ... germany,...

www.powersystemsdesign.com 4

CONTENTS

ViewpointIt Is Toy Train Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Industry NewsDanfoss Strengthens its Research Activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4

Fairchild Reiterates Guidance for Q4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4

Intersil Online Design Tool for Op Amps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4

Tyco Launches New Website . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6

ON Semi Expands Development Center . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6

Microsemi to Acquire Advanced Power Technology, Inc. . . . . . . . . . . . . . . . . . . . . . . . .6

World’s First Mixed-Signal FPGA Designed for the Power Sensitive Applications . . . . .8

Saving Energy Needs to Be Price/performance Neutral . . . . . . . . . . . . . . . . . . . . . . .10

MarketwatchPowering Next Year’s Data Center . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12

Cover StoryHigh-Brightness LED Thermal Management Made Simple . . . . . . . . . . . . . . . . . . . . .14

FeaturesPower Conversion—Three-Level, Four-Quadrant Power Converter . . . . . . . . .20

Power Management—Design Challenge for Synchronous Buck Controller . . . .24

Transient Protection—Robust Discrete AC Protection Power Semiconductors . . .28

Super Capacitors—Ultracapacitors in Industrial Design . . . . . . . . . . . . . . . . . . . .32

Power Semiconductors—Safe Operating Limits in Linear Mode . . . . . . . . . . . . . .36

Optoelectronics—Photocoupler for High Speed and Isolated Switching . . . . . . . . .41

IGBT Modules—IPOSIM, a Flexible Selection and Calculation Tool . . . . . . . . . . . . .44

New Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49-52

Member Representing

Eric Carroll ABB SwitzerlandArnold Alderman AnagenesisHeinz Rüedi CT-Concept TechnologyClaus Petersen DanfossDr. Reinhold Bayerer eupecChris Rexer Fairchild SemiconductorDr. Leo Lorenz Infineon TechnologiesDavin Lee Intersil Eric Lidow International RectifierDavid Bell Linear TechnologyRüdiger Bürkel LEM ComponentsRalf J. Muenster MicrelPaul Greenland National SemiconductorKirk Schwiebert OhmiteChristophe Basso On SemiconductorBalu Balakrishnan Power IntegrationsDr. 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

http://www.fairchildsemi.com

http://www.curamik.com

2 Power Systems Design Europe December 2005

Winter has shown up and the holidays areupon us, this marks my annual ritual to pull outthe toy trains and run them throughout thehouse. My grand nephew enjoys helping me,this is a good start to get them interested inengineering and encourage them to one dayperhaps learn a technical profession.

The last events have shown a positive indi-cation for the upcoming year. SPS/IPC/DRI-VES in Nuremberg has reported a very positiveincrease in exhibitors and visitors. Therefore,SPS/IPC/DRIVES show has clearly reachedan international high acceptance. Not only wasthe Drives and Control manufacturing industryrepresented but the device manufactures havealso started to participate in this show.

Mr. Rath from Mesago can be proud of thewell-established SPS/IPC/DRIVES show inNuremberg. I remember the early days ofSPS/IPC/DRIVES in Sindelfingen. I had givena paper for Harris on power switches and theirareas of application. At that time a device calledMCT was very popular. There was talk that itwould become the ideal switch. The MOSControlled Thyristor had got a lot of credit andmost of the analysts had not been able to seethe technical limits in the device characteristicsand behaviour.

It is like today, I hear that SiC switches will beready next year. We will have to wait and see.Time had shown that MOSFETs and the IGBTsserve the majority of tasks in electrical motordrives and as well in power supply applications.

So SPS/IPC/DRIVES complement in a cer-tain way PCIM perfectly. The thought thatdevice and system manufactures are differentanimals has changed. An example is Semikronwith its move to inverters as we reported in theOctober cover story. Historically Semikron hadbeen a module manufacture.

Japanese companies, such as Toshiba andMitsubishi, are examples of the synergy betweenthe semiconductors and the inverters for drivesin one systems approach. Also in Europe areABB drives and ABB semiconductors as well asDanfoss Drives and Danfoss Silicon Power.

Productronica in Munich had a lot of activi-ty, which indicates a sign of positive develop-ment for our economy.

This month’s cover story is on Bergquist’sThermal Clad Insulated Metal Substrate (IMS) on the cooling benefits for LED lighting applications.

Some of the other technical features in thisissue includes the University of Bremen whichhas taken today’s state of the art MOSFETComponents and evolved their switching speedand low conduction losses.

Texas Instruments has a solution forsequencing and margining capability to simpli-fy power management.

STMicroelectronics takes a closer look atthe safe operating area in MOSFETs for anom-alous failure mode which limits the ability tooperate in linear mode.

Infineon introduces a software to have prop-er modules selected for the individual conditionsregarding DC input and AC output voltage,switching and output frequency, inverter power rating, overcurrent and overload capa-bility as well as the influence of different cooling conditions.

We wish for a peaceful holiday season for all.

Merry Christmas and a Happy New Year!

Bodo ArltEditorial DirectorThe Power Systems Design Franchise

AGS Media Group

Katzbek 17a

D-24235 Laboe, Germany

Phone: +49 4343 421790

Fax: +49 4343 421789

[email protected]

www.powersystemsdesign.com

Publishing DirectorJim [email protected]

Associate PublisherJulia [email protected]

Editorial DirectorBodo Arlt, [email protected]

Editor-in-Chief Power Systems Design China Liu [email protected]

Creative Direction & [email protected]

Circulation Management

[email protected]

Advertising Rates, Sizes and Digital File

Requirements available at:

www.powersystemsdesign.com

Your Free 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:[email protected]

Volume 2, Issue 10

VIEWPOINT

It Is Toy Train Time

http://www.lem.com


INDUSTRY NEWS

Danfoss Strengthens its Research ActivitiesAs of January 1,

2006 Dr. FrankOsterwald will takeover the manage-ment of the researchand developmentactivities of DanfossSilicon Power,Germany.

From 1988-1995 Dr. FrankOsterwald studied

electrical engineering specializing in controlengineering, electronics, microelectronicsand semiconductor technologies at the tech-nical University Berlin. As part of a project

regarding the reliability of power semiconduc-tors at Fraunhofer IZM, he earned a doctorsdegree in 1999 and thereafter took up lead-ing positions within industry responsible forthe design of microelectronic solutions.

Dr. Ronald Eisele, presently Director ofR&D for Danfoss Silicon Power, Germany,has been appointed Professor of Mechatronicsat the University of Applied Science in Kiel,Germany. The collaboration with theDepartment of Mechatronics in Kiel enablesDanfoss to continue and further develop itsactivities within fluid cooling of power mod-ules, assembly and interconnection technolo-gies as well as integration of sensors.

Danfoss Silicon Power situated in Schleswig,

Germany, develops and produces electronicpower modules for various applications, suchas frequency converters, power supplies andelectric vehicles. The competences of thecompany include lead-free soldering of largesurfaces and thermal design methods. Thecompany is part of the Danfoss Group andhas 100 employees who develop and pro-duce customer specific electronic powermodules. Plans have been made for newinvestments of 30 percent in extensions ofclean-room capacity for the production ofcomponents for the automotive industry; itwill result in 50 new permanent workplaces.

www.danfoss.com/siliconpower

Europe Sales Offices: France 33-1-41079555 Italy 39-02-38093656 Germany 49-89-9624550 Sweden 46-8-623-1600 UK 44-1628-477066 Finland 358-9-88733699 Distributors:Belgium ACAL 32-0-2-7205983 Finland Tech Data 358-9-88733382 France Arrow Electronique 33-1-49-784978, TekelecAirtronic 33-1-56302425 Germany Insight 49-89-611080,

Setron 49-531-80980 Ireland MEMEC 353-61-411842 Israel AvnetComponents 972-9-778-0351 Italy Silverstar 39-02-66125-1Netherlands ACAL 31-0-402502602 Spain Arrow 34-91-304-3040Turkey Arrow Elektronik 90-216-4645090 UK Arrow Electronics 44-1234-791719, Insight Memec 44-1296-330061

Fairchild Semiconductor reiterated guid-ance for fourth quarter 2005 revenue toincrease about 5% and for gross margins toincrease 200 – 300 basis points sequentially.

“I’m very pleased with the broad-basedstrength in order rates so far in the fourthquarter,” said Mark Thompson, Fairchild’spresident and CEO. “We’ve booked andscheduled enough demand within the quarterto approximately meet our guided revenuesand are now working to take advantage ofquick turn opportunities for potential upsidesin December. Demand continues to be sea-sonally strong for our products serving thecomputing and consumer end markets,

especially for notebooks, televisions, DVD’sand other consumer electronics. Bookingshave also increased for products supportingpower supply and battery charger applica-tions within our industrial end market seg-ment. Our blended lead times have graduallyincreased during the quarter, and now rangefrom 8 – 10 weeks, so most of the demandwe’re booking today is for deliveries in thefirst half of 2006.

“From a product perspective, order rateshave been up across virtually all of our prod-uct lines,” continued Thompson. “Demandhas increased for our system regulators andpower conversion analog products support-

ing the computing, consumer and industrialend markets. We have steadily increasedbacklog for our low power and high powerswitches supporting a wide range of applica-tions in these same end markets. We’rebuilding on our successful efforts to increasedistribution re-sales and improve the invento-ry mix at our distributors, while further reduc-ing channel and internal inventories,” statedThompson. “The seasonally strong demandthrough the first two months of the fourthquarter has provided a nice tailwind to thisinventory reduction process while improvingour backlog position for the first half of 2006.”

Fairchild Reiterates Guidance for Q4

www.fairchildsemi.com

Already Successful with its iSim Tool forPower Management Devices, Intersil AddsOp Amps to its Suite of Online Tools. IntersilCorporation announced a free, interactive,Web-based tool for selecting and simulatingdevices from Intersil’s broad operationalamplifier portfolio. Based on input and output specifications provided by the user,the iSim for operational amplifiers tool willfind all suitable Intersil devices for a givenapplication. In many cases, simulation is also available for immediate feedback on circuit performance.

Intersil’s iSim simulator tool has an appli-cations-based Intersil analog signal process-ing solution selector with dynamic input fieldsto match a designer’s input and outputrequirements. Options include inverting andnon-inverting gain, transimpedance, differen-

tial, and low and high pass filter configura-tions. All applicable Intersil devices are listed,with many of them available for simulation.When available for simulation, a referenceschematic is generated for simulation, basedon the designer’s specifications.

Making a Designer’s Life Easier in TwoSimple Steps iSim for operational amplifiersis a two-part tool. First, it contains a dynamic“solution selector” to sort Intersil’s portfoliobased on the specifications of the applicationrather than the specifications of the devicesthemselves. This saves the user a step indetermining which Intersil product would bestsuit the end application. This search functionis independent of the simulation side of thetool. Once application specifications are loaded,the sorted devices are displayed with iconsindicating which are available for simulation.

The second part of iSim for operationalamplifiers is the simulator itself. The simula-tions run on the well-established SIMetrix/SIMPLIS platform, which is based on SPICE.The simulator builds a circuit using the appli-cation specifications loaded into the selector,as well as the basic SPICE model for the following analyses: AC analysis (bode plot),transient pulse and transient sine. In addition,the circuit can be downloaded to iSim: PE(personal edition) for offline editing and AC analysis.

Intersil’s iSim for operational amplifiers toolis free to registered users. To register, go to www.intersil.com/isim.

Intersil Online Design Tool for Op Amps

www.intersil.com/isim

http://www.linear.com/automotive

http://www.linear.com/automotive

6

INDUSTRY NEWS

Power Events

• EMV 2006, March 7-9, Düsseldorf www.mesago.de

• ECPE 2006 SiC User Forum, March 14-16,Nuremberg, www.ecpe.org/news/news_e.php

• APEC 2006, March, 19-23, Dallas TX,www.apec-conf.org

• PCIM China 2006, Mar. 21 - 23, Shanghai www.pcimchina.com

• PCIM Europe 2006, May 30 - June1, Nuremberg,www.pcim.de

• SMT/HYBRID 2006, May 30-June1, Nuremberg,www.mesago.de

• SENSOR/TEST 2006, May 30-June1, Nuremberg,www.sensor-test.de

• MICROSYSTEM, October 5 –6 , Munich,www.mesago.de

• ELECTRONICA 2006, Nov. 14 – 17, Munich,www.electronica.de

• SPS/IPC/DRIVES 2006, Nov. 28 - 30, Nuremberg,www.mesago.de

Irvine, CA Microsemi Corporation and andBend, OR Advanced Power Technologyannounced the signing of a definitive agree-ment for Microsemi to acquire APT. With thisacquisition, Microsemi expands its portfolio ofanalog mixed signal offering in the RF market-place and also its high reliability offering in thedefense/aerospace and medical marketplace.

The transaction combines two strong highperformance analog companies offering bothdifferentiated RF product into niche end mar-kets as well as high reliability product address-ing a strong defense/aerospace and medicalmarket. APT is a leading designer, manufac-turer and marketer of high-performance RFand switching power semiconductors. WithAPT’s strong focus on the high-power, high-speed segment of the power semiconductormarket, APT’s RF and power switches willgreatly expand Microsemi’s product offeringwithin its existing channels. Additionally, APT

has an advanced development effort in siliconcarbide that offers significant advantages infuture systems ranging from military to note-book applications.

In addition to the strategic opportunity thatthis acquisition provides, the combined com-pany should have a profitable operating modelthat should generate significant cash. Thecombined entity has potential consolidationopportunities inline with previous restructuringefforts at both companies to improve efficien-cies. Together, the companies should in timeperform at greater than our 50% gross marginand 27% operating margins goals. Microsemiexpects the transaction to be accretive to itsthird quarter fiscal year 2006 results.

Microsemi to Acquire AdvancedPower Technology, Inc.

www.microsemi.com

As part of its continuing commitment todevelop efficient power management compo-nents for use in consumer, computing andoffice electronics, ON Semiconductorannounced plans to double both the staff andthe square footage of its Power ConversionIntegrated Circuit Development Center inChandler, Arizona.

By the end of the year, the Chandler facilitywill be expanded to 16,000 square feet andup to 20 new power integrated circuit engi-neering positions will be added. Work at theChandler center - one of 8 design centersON Semiconductor operates worldwide - isfocused on the research, design and testingof power-efficient integrated circuits andpower supply solutions used in the conver-

sion of both AC and DC power. The devicesdesigned at the Arizona facility utilize uniqueand often proprietary technology to convertpower from a wall outlet or battery to digitalconsumer or computer electronics with mini-mal energy loss.

ON Semiconductor’s proprietary powermanagement devices have been implement-ed into a wide range of products, includingcomputers, LCD monitors, printers, flatscreen televisions, set top boxes and batterychargers for cell phones. Additionally, ONSemiconductor products support ENERGYSTAR® qualified consumer products.Recently, the company introduced a uniquedual-edge PWM controller that will facilitatethe industry-wide development of DC-DC

power supplies with highly efficient “activemode” performance for computing and con-sumer electronics applications. For threeconsecutive years, the China EnergyCommission (CEC) has honored ONSemiconductor with its “Top 10 DC-DC”award for development of energy-saving DC-DC conversion components. Earlier thissummer, ON Semiconductor unveiled itsGreenPoint portfolio with the introduction ofthe industry’s first open ATX Power SupplyReference Design certified to meet 80 PLUSperformance requirements. AdditionalGreenPoint introductions are planned.

ON Semi Expands Development Center

www.onsemi.com

Tyco Electronics Power Systems announcedthe launch of its new customer-friendly website.Supported by a powerful content managementsystem, the website provides users with themost up-to-date information on Tyco Electronics’world-class power systems products.

The homepage of the new website isdesigned to make searching for power productseasier and faster by offering easy navigationmenus with recognizable industry standard cat-egory names. Users can click on these names

and be linked to product family pages, wherethumbnail images of product families, detailedproduct descriptions, and product documenta-tion links are easily located. Product documen-tation sub-categories for application notes andproduct manuals are especially helpful to usersseeking specific product information.

Also included are cutting-edge features suchas our interactive Product Viewer that makes iteasy to find our most popular DC-DC andEnergy Systems products. The Featured

Products and Innovations sections will allowusers to view the most current products andservices that appeal to and benefit a wide vari-ety of customers. Additional website elementsdesigned to enhance user experience includeonline video, which will include instructionalsegments for product demonstrations, designtools such as product configurators and design-ers, as well as product assessment tools.

Tyco Launches New Website

www.power.tycoelectronics.com

www.advancedpower.com

Power Systems Design Europe December 2005

http://www.powerint.com/piexp6


Satisfying a strongdemand from systemarchitects for a device

that simplifies design andunleashes their creativity, theleading vendor of single-chipprogrammable solutions ActelCorporation recently has held apress conference in Beijing,China to announce the immedi-ate availability of the ActelFusion™ ProgrammableSystem Chip (PSC), the world’sfirst mixed-signal FPGA family.

Replacing the high-cost dis-crete devices

By integrating mixed-signalanalog, flash memory andFPGA fabric in a monolithicPSC, the Actel Fusion devicesenable designers to quicklymove from concept to complet-ed design and deliver feature-rich systems to market. TheActel Fusion PSCs bring thebenefits of programmable logic to appli-cation areas, including; power manage-ment, smart battery charging, clock gen-eration and management and motorcontrol, that until now have only beenserved by either costly and space-con-suming discrete analog components ormixed-signal ASIC solutions.

The Actel Fusion PSCs present newcapabilities for system development byallowing designers to integrate a widerange of functionality into a singledevice while at the same time offeringthe flexibility of upgrades in the field ordeep in the production cycle. The ActelFusion devices provide an excellent

alternative to costly and time-consuming mixed-signal ASICdesign. In addition, when used inconjunction with Actel’s ARM7and 8051-based soft MCU cores,the Actel Fusion technology rep-resents the definitive PSC platform.

Designed for power sensitive applications

The Actel Fusion devices inte-grate a configurable 12-bit suc-cessive approximation register(SAR) analog to digital converter(ADC) with frequencies up to 600ksps. The flexible analog blocksupports MOSFET gate driveroutput and multiple analog inputsfrom -12 volts to up to +12 voltswith optional prescaler, thus enablingdirect connection and control of awide variety of analog systemssuch as a voltage, differentialcurrent or temperature monitor.

The Actel Fusion PSC family is the only programmable logic solutionto include embedded flash memory—upto 1 Mbyte per device. The flash memo-ry offers 60-nanosecond random accessand a very fast 100MHz access in read-ahead mode. The high performanceflash memory offers a user configurabledata bus supporting x8, x16 and x32 bit

Unprecedented integration of analog peripherals, Flash memory and FPGA fabric

in a monolithic programmable system chipredefines how products are designed

widths. The memory also offers errorcorrection circuitry (ECC) with single-biterror fix and two-bit error-detect capabil-ities. Pseudo EEPROM can be achievedwith an available endurance extender IPavailable from Actel. Further, the ActelFusion PSCs give designers unprece-dented levels of flexibility by allowingthem to easily reconfigure analog blocksettings to perform widely different func-tions by simply downloading data fromembedded flash memory.

Actel’s Fusion PSCs extend the corebenefits of the company’s flash FPGAtechnology—live at power-up (LAPU),single-chip, firm-error immunity and lowtotal system cost—to the world’s onlymixed-signal FPGAs. The new ActelFusion PSCs deliver the core analogblocks required for applications in theindustrial, medical, military/aerospace,communications, consumer and auto-motive markets.

For power sensitive applications, theActel Fusion PSCs offer ultra low-powersleep and stand-by modes and arespecifically designed to handle Level 0LAPU system supervisory activities,such as system board power-upsequencing and configuration.

Fusion innovationThe Actel Fusion technology and

devices enable designers to design atboth very high and very low levels ofabstraction. Fusion peripherals includehard analog IP and hard and/or soft digi-tal IP. Peripherals communicate acrossthe FPGA fabric via a layer of soft gatesæ the Smart Backbone. Much morethan a bus interface, the SmartBackbone integrates a micro-sequencerwithin the FPGA fabric and will configurethe individual peripherals and supportlow-level processing of peripheral data.

“With the Actel Fusion PSC we areremoving the handcuffs from systemarchitects and allowing them to focus on

adding unique features and enhancingend-product value,” said John East,president and CEO of Actel. “Designerswill be able to treat the Fusion PSCs like a mixed signal ASIC without all theASIC penalties of long design cyclesand high costs.”

To support these new devices andhelp maximize designer productivity,Actel has developed a comprehensivedesign environment that includes soft-ware tools, intellectual property (IP)and reference designs.

www.actel.com

http://www.ixys.com


New developments and a system-based approach to design cansave energy in motion control

applications and vehicles without chang-ing the price/performance ratio.

Our standard of living is increasinglydefined by how efficiently we use theenergy we extract from oil, coal, wood,nuclear power, hydroelectric powerschemes, and other energy sources.The more efficiently use of this ‘energybudget’, the better we live.

According to the Energy InformationAdministration, our global energy budgetin 2002 was 412 quadrillion BTUs. Thisfigure is set to get much bigger, withpredictions that by 2025, it will grow toan even more staggering 645 quadrillionBTUs. The big problem here is not somuch finding energy, but consuming it insuch a way that we minimise harmfuleffects on the environment while main-taining or raising our standard of living.And this means more efficient energy use.

Unfortunately, while the need toimprove efficiency is clearly recognised,very few governments, organisations orindividuals want to pay extra for efficien-cy improvements. This means that suchimprovements must be ‘price/perform-ance neutral’ when compared to existingsolutions. Take motion control—insideproducts that are part of daily life suchas home appliances, air conditioningsystems, fans, pumps, elevators, andconveyor belts in factories. In total,these motion control applicationsaccount for over half of global electricityconsumption. Analysis shows that over85% of these motors are cheap, ineffi-cient ‘click on, click off’ single-phase ACinduction motors. It is estimated that,simply by combining variable speedmotors with intelligent controls in all ofthese applications, we could reduceglobal electricity consumption by 50%.

But how can this be done withoutincreasing the price/performance ratio?

International Rectifier’s experiencewith washing machines provides a goodillustration. Ten years ago, we created a$250 variable speed motor drive systemfor a washing machine. Now that’s a bigslice of the cost of a washing machine,and clearly few people were willing topay a premium just for greater energyefficiency. However, the variable speedmotor improved functionality—it wassilent and washed clothes better so theylasted longer—functionality for whichretailers could demand a premium. Theprice may have increased but theprice/performance equation was essen-tially neutral and the consumer was get-ting energy savings as part of the bar-gain. Furthermore, as mixed-signal andanalogue power semiconductor tech-nologies and packaging have evolved,so too has evermore sophisticated con-trol products and methodologies. In theintervening ten years, we have beenable to create true system-level solu-tions that can use much lower cost motorswhile significantly reducing overall com-ponent count. The result? High efficiency,high performance washing machines

that are now nearly equal to the cost ofa purely mechanically-actuated alterna-tive. With transportation representing 20– 25% of global energy needs, and withcar ownership growing rapidly, improvingvehicle efficiency will have a significantimpact on global energy consumption.

The starting point is in cutting fuelconsumption without compromising costor performance. To help achieve this,International Rectifier has concentratedon three key automotive systems whichinclude the Integrated Starter Alternator(ISA), the Electric Power Steering (EPS)system, and the climate control and cool-ing systems with associated blowers,pumps and fans. By combining advancedpower semiconductor devices with pro-prietary analogue, digital and mixed-signalsolutions, it will soon be possible to imple-ment all three of these systems in a mid-range vehicle for only a slight cost premi-um. Again, there is almost no excusenot to pursue the more energy efficientoption. It is anticipated that by 2013, costswill be within a few percent for a car withfuel economy that is up to 60% better.

The imperative to save energy andthe impact of energy conservation hason our standard of living cannot be under-estimated. Only the most conscientiousconsumers or organisations, however,will buy something simply because it usesless energy. They will buy it if they arecomfortable with the price/performanceratio. This means that implementingmore efficient systems should either bea ‘cost neutral’ exercise, or should leadto added functionality and performancethat can legitimately demand a premi-um. And this requires companies suchas International Rectifier to take an inte-grated system-orientated approach todesign – to understand the real require-ments of the system in terms of cost,performance, features and function.

www.irf.com

By Dr. Alex Lidow, International Rectifier’s CEO

http://www.ansoft.com


By Chris Ambarian, Senior Analyst, iSuppli Corporation

Alot of us in the power industryhave for years felt that whilepower wasn’t as James Bond

sexy as some of the more visible chipsout there, power management is surelymuch more important than people realize.Maybe this is because we’ve really mar-keted ourselves badly—or maybe it’sbecause we’ve been doing our jobs sowell to date that we haven’t really beena headache to the industry.

As most of you know by now, howev-er, this is all changing. We are now infact being seen alternately as a reallyimportant area of design challenge, oras a major headache, depending onwhom you talk to. Power managementis now a primary and up-front designconsideration in product designs innearly every corner of the equipmentindustry. Power management is oftenthe pacing, edge-of-the-envelope con-sideration that determines product capa-bilities (or limitations). And nowhere isthe importance of power managementbeing made clearer than in today’s net-work facilities and data centers.

The Wall Street Journal recentlyreported (thanks to Astec’s Bharat Shahfor this story) on the saga of the StateUniversity of New York at Buffalo.[Please note: Power management, frontpage, Wall Street Journal. How far wehave come!] It seems that the Universitybought a $2.3M Dell supercomputer, inan effort to increase its ranking in theworld supercomputer standings. Butwhen they went to plug it in, they foundthat they didn’t have enough electricalpower to the building to run it (at leastnot all of it). They needed to do a sub-stantial upgrade to their electrical sys-tems to be able to use their new boxes.

This also presumes that once theyupgrade their electrical supply, they canstill afford to pay their electrical bills!Operators of data centers and switchingfacilities are reporting that their electricalrequirements are ramping upward at

an exponential rate that simply cannotcontinue. Even moderately-sized userssuch as hospitals are talking aboutmegawatts.

And the Buffalo story is just the tip ofthe iceberg. It pales in comparison tothe scale of the challenges faced byfolks designing new (or worse, trying toretrofit existing) data centers. The powerdensities of these facilities are now run-ning around 1kW/m2—with some exoticones running at a staggering 5kW/m2.Individual pieces of equipment arepushing the envelope at 40kW/m2. Onefalse move in any of the power supply orpower removal, and you have a dead,multi-million-dollar piece of equipmenton your hands.

And yet, I have heard it said amongpower management people that we’redoing everything that they can toincrease efficiency, but to some extentit’s out of our hands. The logic goes likethis: “We can improve another 2 or 3%in efficiency of that converter brick—butwhat about all the power being dissipat-ed by the micros and the memory andthe fans and… So really, whatever we’reworking on really isn’t the pacing item inreducing the losses in the room.” This

seems like a pretty good line of thinking,but nonetheless I’m here to argue withit. (Sorry, Mom, a guy can change onlyso much.)

I think that in order to understand ourpart in the challenge as semiconductorand power supply folks, it is helpful totake a look at the facility level, to seewhere the power is going.

According to numbers from Intel andfrom iSuppli’s own calculations and con-versations with system designers, here’san overall view of where the power sup-plied to a modern server room is going:

As you can see from Figure 1, half ofthe power needed goes right to theinfrastructure of the room—cooling andcontinuity of power. As mature as themotor drives and HVAC industries are,there is still much room for improvementin the efficiency of HVAC systems. Unlessa sophisticated motor control with com-munications capability is being used andactively adjusted to the room needs ona real-time, proactive basis, chancesare that significant amounts of energyare being wasted in the chiller/air han-dler loop. These technologies are avail-able—the question is: is anyone mind-

MARKETWATCH


ing the switch? The operating point ofthe system needs to be adjusted to itsoptimum—which presumes a system todetermine what that point is, and toadjust the system to that point (preferablyproactively, based on system events).More on this later.

Still in infrastructure, there’s the UPS.A typical UPS can dissipate 265 BTU/hrper kVA—which translates to about 89%efficiency. Appropriately, a lot of work isbeing done to improve on UPS efficien-cy; of all the dissipaters around the sys-tem, this one is the most stand-alone.Efficiency gains here will be direct, andmore or less independent of the rest ofthe system.

Which brings us at last to the reasonwe’re here: the equipment itself. Let’stake a look at a typical server system.We’ll consider an average configuration,somewhere between a typical 1U(around 900W) and 2U (around 1100W).For such a 988W system, the dissipa-tions at rated capacity look like in Table 1.

The potentials for reduction I’m toss-ing in there are a best estimate as aninterested, more or less objective observ-er of what’s going on in these areas. Allof the processor players (both large andsmall) are introducing new ideas forlower leakage and lower in-use dissipa-tion, and some of the technologies holdtremendous promise relative to today’ssolutions. Some of that 80% numberhas already been significantly chippedaway at by recent introductions andannouncements such as those by AMDAnd I threw 20% in there for memorybecause, frankly, I don’t know a damnedthing about memory but there must besomething that they can do to improve

on the situation, and I didn’t burdenthem too badly with that. Power conver-sion gets a big chunk of the work: I figurethat with anticipated improvements inmagnetics and in digitalization of thePWM controls so that operation can beoptimized at multiple points, we ought tobe able to gain another couple of percent.Throw in some MOSFET and diodeimprovements, and AC/DC can be broughtup another couple of percent. We canargue the exact amounts of these num-bers, but broad side of a barn, we cansee a path to about 40% reduction infull-throttle dissipation. Which not onlyenables us to pack the racks in evenmore densely—it saves a lot of energyand money. Hey, 40% or so isn’t smallchange. But that’s only part of the story.

The requirements on the data centerare anything but constant. There arevery well documented time-load profiles,and then of course there’s the unpre-dictable load that comes up (like thathumorous old story about the big flushthat overwhelmed the sewage plantafter the commercial break of the SuperBowl broadcast). But leaving everythingrunning full-throttle all the time inde-pendent of demand is obviously not asmart solution. The smart solution willbe the ability to eke the maximumamount of efficiency out of the system atany given time—and as we’ve seenfrom this facility-level view, that meansscaling back on that expensive air con-ditioner at the same time that you’rescaling back 62% of your server racks.Real time. If we’re able to do that, thennot only are we saving that 40%, butwe’re saving a significant portion morethan that by idling unneeded capacity allaround the system.

This idea begins to provide a glimpseinto what we see as a critical need for apower operating system to manage allof the parts of a complex power systemfor maximum efficiency, while still main-taining the ability to bring the whole sys-tem up to full capacity more or lessinstantly. In data centers in particular,this need must be met within the next12-18 months. In order to be able tomanage all of the pieces of a systemlike this, we need to be able to notmerely sense (e.g., via thermostats), butto digitally predict what’s happening orgoing to happen throughout the system,and then direct the system to operate inthe most efficient mode for that condi-tion. This requirement is facilitated bythe digitalization of power conversion. In fact, this is what we believe to be thepunch line of digitalized power: as acomponent-level phenomenon, it’s animportant development, but at a systemlevel, it’s a REALLY important develop-ment. Digitalized power converters willbe able to communicate useful informa-tion to a power operating system, whichcan then use the information to deter-mine the best state for every level of thesystem to be operating at—whether thatlevel is the adjustment of the speed of avariable speed drive on the air handlingunit, or the level of setting the operatingpoint of the control loop in a thousandDC/DC converters.

Of course, this will have some verysignificant implications on the softwareside. That will perhaps be the subject ofa future editorial.

So to bring us full circle, for us semi-conductor and power supply people, it’snot out of our hands. In fact, the onlyway that we’re going to get to the nextlevel of meeting our users’ needs will beto stop thinking at the component level,and to figure out how we’ll fit in at thesystem level. When we begin to do that,the first thing that we realize is that we’lleither be driving that choice, or we’llhave to take what we get.

The system awaits…

MARKETWATCH

Figure 1. Power dissipation in a modern server facility (source: Intel, iSuppli). www.iSuppli.comTable1. Power dissipations and potential improvements (sources: Intel, iSuppli).


COVER STORY

Use similar techniques as withpower transistors

Bergquist’s Thermal Clad Insulated Metal Substrate (IMS) offers many of the same cooling

benefits for LED lighting applications as it does for power electronics.

By Rick Samuelson, Market Development Manager and Justin Kolbe, Sr. R&D Engineer,

The Bergquist Company

Like surface mounted power tran-sistors, High-Brightness LightEmitting Diodes (HB-LEDs)

require cost effective thermal manage-ment for long-term reliability. Consistentcolor and power magnitude are directlyaffected by the thermal managementsolution as well. Generally, as the lightoutput of HB-LEDs increase, so doespower dissipation or watts per HB-LED.

Currently, three- and five-watt LEDsare commonplace and many industryexperts are predicting 10-watt LEDsavailability in the next few years.Packaging and thermal managementsolutions for HB-LEDs are parallelingsurface mounted power devices.

HB-LED of greater than one watt arealmost always surface mounteddevices. This is because the axial leadsto the die in a leaded package do notconduct enough heat away from theLED. Chip-on-board (COB), flip chipsand thermally efficient packages areemerging as the standard thermal man-agement packaging solution for HB-LED.

The Effect of Temperature HB-LED die have several temperature

dependent properties. The color, orwavelength will change with tempera-ture. As die temperature increases, thewavelength of the color gets longer.

= Change in dominant wave-length (nm)

= Change in die junction temper-ature (∞ C)

The temperature effect on wavelengthis depicted in Figure 1.

An area in which this is of particularimportance is with white light. Thehuman eye can differentiate small colorchanges in white light. When HB-LEDsare populated in an array, consistentthermal resistance from one die to thenext assures consistent color. Becauseof the comparatively low thermal resist-ance Thermal Clad offers as comparedto FR-4, die temperature is less affectedby slight variances in junction-to-casethermal resistance that occurs with tineutectic or epoxy-die mounting techniques.

Power magnitude, or watt density, isdirectly related to light output. The morepower that is applied to the HB-LEDwhile maintaining the desired die tem-perature, the higher the light output willbe. It may also be possible to pack thedie more closely in an assembly that utilizes good thermal management tech-niques, thereby reducing the effects of temperature.

The following graph shows the dra-matic change in relative light intensitybased on the forward current applied to the LED.

Figure 1. The temperature effect on wavelength.

http://www.ohmite.com

16

COVER STORY

Finally, the junction temperaturedetermines HB-LED lifetime. The heatrise in LEDs is due to the energy that isnot turned into light energy.

P = I2R

Generally, a 50 percent drop in lightoutput for a constant-forward currentindicates end-of-life for HB-LEDs. Withproper thermal management, HB-LEDlifetimes can exceed 100,000 hours. Inaddition to long life, other factors thatmake LEDs viable as compared toincandescent or fluorescent lighting are:Versatility, safety factors (low voltage),total cost of ownership.

Although the initial cost of HB-LEDs ishigher, many applications such as trafficsignals, architectural and display lightinghave demonstrated HB-LED lighting isthe most cost effective solution.

Cost of Heat SummaryBetter thermal management allows for

more forward current applied to the LED,which means more light and possiblyreducing the number of LED required forthe desired light output. Maintaining acooler assembly at an equivalent powerequates to more light per die. Ninetypercent of the electrical energy in a redAllnGaP LED will be converted into visi-ble light. Only part of this light energycan be ejected from the chip. Theremaining energy, which cannot leavethe chip, will be converted into heat.

Heat Transport MethodsDetermining the exit path for thermal

energy is important. Conduction, con-vection and radiation are the primarypaths for the removal of heat from theHB-LED die. Conduction, or heat trans-fer through a solid body, is the most effi-cient thermal path, followed by convec-

tion, which is usuallythe final thermal pathto ambient. Radiationis generally negligiblein managing the tem-perature of HB-LEDs.

The following calcu-lation is used for thestack’s conductive heattransfer for the stack,from the heat source tothe air.

Power that can bedissipated under

a given temperature constraint is theninfluenced by:

area of the layer [A],the thermal conductivity [K], the thickness of the layer [dx] and the power [q].

In the case of a typical HB-LED application, several thermal resistanceinterfaces are present in Figure 3.

Moving through the stack, thermalimpedance depends on thermal conduc-tivity and thickness. Figure 4 depictsthermal impedance versus thermal con-ductivity at different thicknesses for thedielectric layer.

Note that as the dielectric thicknessgets thinner, thermal conductivity hasless affect on thermal resistance. Asthis relates to heat removal from theassembly, the total thermal resistance of each component in the stack is con-sidered, and then assembly-to-ambient.Thermal Clad has a 75-micron dielectricwith relatively high thermal conductivity.This insulation layer is critical in thethermal management of HB-LEDsbecause the isolation layer is potentiallyone of the highest thermal resistanceinterfaces in the stack.

For the assembly-to-ambient inter-face, different options include an infiniteheat sink (such as the chassis of anautomobile or metal enclosure), forcedair, or a dynamically cooled assembly(such as a fan or thermal electric cool-er), often modeled as an infinite heat sink.

Figure 2. Relative Intensity vs. Forward Current.


Figure 3. Several thermal resistance interfaces.

Figure 4. Thermal impedance versus thermal conduc-tivity at different thicknesses.

http://www.IGBT-Driver.com

www.powersystemsdesign.com 1918 Power Systems Design Europe December 2005

conductor areathat the LED ismounted on getssmaller, thermallyconductive materi-als become more valuable. Buildingthe mounting con-ductor larger, theLED improvesthermal resistanceas shown in thefollowing graph.

Packaging ConclusionSeveral options are available for thermal management of

HB-LEDs. Most common are packaged die from the manufac-turer and direct-die-attach to the circuit board or substrate bythe HB-LED integrator. The most critical thermal path in thestack is the one with the highest thermal resistance. Goodpractice suggests that you reduce the thermal resistance ofthat layer with Thermal Clad instead of FR-4. Typically, HB-LED packages from the manufacturer have a thermal resist-ance of > 15∞C p/Watt. When HB-LED packaging has thermalresistance this high, the circuit board choice has less of animpact on thermal management. When the opportunity existsto use tin eutectic gold die attach methods, as was the case in

the model and bench test described earlier, the circuit boardmaterial greatly affects HB-LED thermal management.Bergquist’s Thermal Clad material is a cost effective option forthe thermal management of HB-LEDs thereby improving heattransfer as compared to emitters. However, thermal manage-ment of HB-LED emitters is also improved on Thermal Clad ascompared to FR-4, though the impact on thermal manage-ment is less than COB packaging.

COVER STORY

Figure 6. Modeling Results – Infinite Heat Sink.

www.bergquistcompany.comFigure 7. Light Output of Die on Different DielectricMaterials @ 50°C.

Removing the heat from the die isonly effective if it can also be removedfrom the assembly. An infinite heat sinkis the most effective solution thermally,but often impractical or too expensive.

Therefore, when using an infinite heatsink approximation, the problem becomesone of conduction only. This assumptionis generally valid if the removal of heatfrom assembly is greater then the gen-eration of heat, assembly is mounted toa large heat sink and assembly is beingcooled dynamically.

Still air (or natural convection) is prob-ably the most common method of cool-ing but still requires air flow (e.g. holesin a cover or enclosure). Otherwise, airis just a good thermal insulator.

Use the following equation as a ruleof thumb when figuring heat dissipationin still air.

q = dissipated power (W)A = area (m2)ΔT = (temperature plate – tempera-ture ambient) (K)L = plate length (m)h = convective heat transfer coeffi-cient (W/m2-K)a = coefficient 1.32 for top of plate,0.59 for bottom, dependent on shapeand orientation

Use the following equation as a ruleof thumb when you have moving air.

q = dissipated power (W)A = area (m2)ΔT = (temperature plate – tempera-ture ambient) (K)L = plate length (m)h = convective heat transfer coeffi-cient (W/m2-K)a = coefficient 0.0366 for turbulentflow, 0.59 for laminar flowB = 0.8 for turbulent flow, 0.5 for lami-nar flowV = velocity of fluid (m/s)ρ = density of air (kg/m3)μ = viscosity of air (Poise)NPr = Prandtl Number = Cp m / kCp = heat capacity of air (J/kg K)k = thermal conductivity of air (W/m-K)

Overall thermal resistance of the build is:

Ri is bulk or interfacial resistance -Conduction Rb = L/k AInterfacial resistance must be deter-

mined empirically (generally small, butnot always)

-Convection R = 1/h A

Overall power dissipation in a giventemperature rise is:

Overall temperature rise for givenpower dissipation is:

T2 = T1 + qRTotal

or the overall temperature rise for thegiven power dissipation is:

T2 = T1 + qRTotal

These simple equations are good forfirst approximations of temperature risein HB-LED designs. However, bench toptesting and data confirmation are advis-able to verify the design.

Circuit Board Comparison Models Insulated Metal Substrate (IMS) and

standard FR-4 are commonly used cir-cuit board materials in conjunction withHB-LEDs. The Bergquist Company’sThermal Clad IMS is a thin, thermallyconductive layer bonded to aluminum or copper substrate for heat dissipation(see Figure 5).

Thermal Clad circuit board materialsare available from The BergquistCompany in two different thermal con-ductivities. Multi-Purpose (MP) and LowThermal Impedance (LTI). Figure 6 illus-trates the modeling results of a CreeXB900 die mounted on each of thesecircuit board materials, along side anFR-4 board. This thermal model showsthe temperature of the die when a con-stant current is applied to the die on dif-ferent circuit board materials. The Creedie is modeled on two-ounce circuit foil,.020" square. Tin gold eutectic dieattach was utilized with gold wire bondsto complete the circuit. Power dissipa-tion was set at one watt. The resultingdie temperature for the FR-4 is 63.7∞C,MP is 36.9∞C and LTI is 34.2∞C.

Bench Test ResultsA bench test setup is used to verify

the thermal model. The test is adaptedfrom Voltage from Base to Emitter(VBE) testing used in power electronics.The die is calibrated by measuring volt-age at different temperatures using asense current small enough so as tocause negligible heat rise. In this case,one milliamp was used. Following areall of the test parameters common inpower electronics to determine junctiontemperature.

• Sense current - 1 mA• Heating current and heating voltage

from power supply• Electrical resistance dependent on

temperature• Know V and I, can get T as a func-

tion of R• Thermal resistance of the system is

Tj-Ts/(V*I)

Once the LED die was calibratedusing the described set up, the forwardcurrent to the die was adjusted in orderto regulate the die temperature. Figure7 illustrates the light output of the sameLED die on different circuit board mate-rials the die temperature was maintainedat 50°C. This bench test shows the ther-mal advantages of Thermal Clad versusFR-4, resulting in higher light output.

The thermal resistance is both a function of the thermal conductivity ofthe dielectric and the pad size. As the

Figure 5. The Bergquist Company’sThermal Clad IMS.

COVER STORY

Figure 8. Thermal Resistance TestExperiment Results.

http://www.advancedpower.com


600V CoolMOS devices in 400VAC motion control application

Today’s state of the art MOSFET Components offer compelling advantages in switching speed

and low conduction losses. These devices have a maximum voltage rating of 600V but they can

be employed in standard 400VAC power systems in multi-level topologies.

By Timo Christ, Uwe Werner and Bernd Orlik, IALB University of Bremen, Germany

Frequency converters generate an output voltage by switchingbetween at least two DC voltage

levels. The average in time of thesepulses is the desired output. A converterwith more than two connections to theDC link is regarded as a multi-level con-verter. The industry standard two-levelconverter switches the output nodeeither to the positive or the negativepotential of the DC link. In a three-levelconverter, an additional voltage level isintroduced. The diode clamped configu-ration takes this voltage from a tap inthe DC link energy storage capacitor.Many converters employ series connec-tion of DC link capacitors in any case toachieve the required voltage rating.

Multi-level converters feature a num-ber of advantages and some disadvan-tages over two-level converters. Theswitches in a multi-level converter onlyneed to withstand the voltage of onelevel, in other words the blocking volt-age requirement is cut in half for athree-level converter with the same totalDC link voltage as a two-level converter.This allows lower voltage and potentiallyfaster or lower conduction loss devices

POWER CONVERSION

Figure 1. Half-Bridge Structures.

http://www.ti.com/T2-e


to be used. Switching only half the totalDC link potential also cuts the voltagegradient dv/dt in half for the sameswitching speed. This can ease EMCproblems and reduce cost for screeningand filtering. Alternatively, switchingspeed may be increased to reduceswitching loss. Approximating the outputvoltage with three instead of two levelsalso reduces harmonic distortion signifi-cantly, allowing reduction of switchingfrequency or less filtering for the samequality of output. Switching loss is thesame in two- and three-level converters.This is because the average switchingfrequency of each switch is halved, butthe number of switches is doubled. Theswitching losses are distributed in agreater number of switches. This raisesthe opportunity to double switching fre-quency for the same losses per device.

Multi-level converters consist of signif-icantly more parts than two-level con-verters. The three-level converter needstwo additional switches and two diodesper half bridge, and system cost is driv-en up because additional gate driversand a more powerful controller arerequired. More parts also bring aboutinferior reliability. System cost increasefor a three-level converter of 20kVAoutput power was estimated as 60%over an equivalent two-level converter.Presently, no half-bridge drivers withintegrated protection functions are avail-able for three-level converters. Thesedrivers make it easy to build a robusttwo-level converter, three-level driversmust be custom designed. Half bridgepower modules are not available either.Today’s digital signal processorsdesigned for motor control are aimed attwo-level PWM and contain correspon-ding peripherals. Making these compo-nents work in a three-level converter isan additional engineering effort.

To balance the voltages of the DC-linkcapacitors a simple regulator is used. Itconsists of an IGBT half bridge whoseoutput node is connected to the neutralpoint of the DC link via an inductor. Thehalf bridge is controlled with a hystere-sis controller which consists of voltagedividers, high speed comparators and asmall microcontroller. When the neutral

point voltage sags to below half of thetotal DC link voltage minus the hystere-sis, the upper IGBT is turned on.Accordingly, the lower IGBT is turned onwhen the neutral point voltages risesabove half of the total DC link voltageplus the hysteresis. In this way, the con-troller keeps the neutral point alwayscentered between the positive and neg-ative DC link potentials. This ensuresequal voltage sharing on the CoolMOSdevices. With this regulator, the modula-tion algorithm of the inverter does notneed to take charge balancing of thecapacitors into account.

Fiber optical signal transfer has thedistinct advantage over copper wiring ofimmunity against electro magnetic inter-ference (EMI), which can be a major

asset in a power electronic environment.The attained galvanic isolation betweensignal and power processing is also of a quality which cannot be matched by other technologies presently.Disadvantageous is the high cost of allfiber optic components. The VersatileLink fiber optic system from AgilentTechnologies was employed, using thetransmitter component HFBR-1524 andreceiver component HFBR-2524 and an interconnection of 1mm plastic fiberof 1m length. Twelve channels arerequired to control the inverter. Thetransmitters with their drivers are mount-ed on a PCB which can be connected tothe DSP system directly. Receivers areintegrated on the gate driver PCBs.

Figure 2. a) Two-Level and b) Three-Level Phase-to-Phase Output Voltages.

POWER CONVERSION

The digital signal processor (DSP)employed is TMS320F2812 by TexasInstruments. This is a device with twointegrated peripheral blocks designed toproduce two-level PWM output for threephase systems. These blocks are calledEvent Managers and contain counters,comparators and auxiliary logic circuitsto make control of two-level converterseasy. Unfortunately, these blocks don’thelp much at all in controlling a multi-level converter. This problem wassolved by having each Event Managercare for only one half wave. EventManager A only generates switchingpulses during the positive half cycle andis switched off during the negative halfcycle, for Event Manager B the reverseis true. An external logic circuit con-structs the twelve pulse control signalsfor the three-level inverter from the sig-nals generated by the DSP and pro-vides additional protective functionssuch as shoot through protection anddead time.

An experimental converter which fea-tures active rectifier and inverter stageswas built in our laboratory and initialoperative testing was conducted with

this device. The main PCB, which con-tains four 70μm copper layers, holds all power components. That is, 24CoolMOS devices and 12 clampingdiodes in TO-247 packages are mount-ed underneath the board in contact witha substantial heatsink, the DC-Linkcapacitors and gate interface compo-nents are mounted on top of the board.Low parasitic inductance is of extraordi-nary importance with these very fastswitching power devices. The gate driv-ers are constructed on separate PCBswhich are mounted close to the mainboard. They are implemented basicallywith the fiber optic receiver and a MOSFET gate driver IC, namely theMC33151D by ON Semiconductor. Thismodular approach allows experimenta-tion with different gate drivers and easi-er exchange of power components,however at the disadvantage of longergate wiring. The converter was success-fully tested with DC link voltages of upto 680V.

This experimental multi level convert-er demonstrates the serviceability ofCoolMOS Devices in 400VAC motioncontrol applications. The converter is

able to drive up to 15kVA loads withhighly dynamic performance and lowvoltage gradients compared to industrystandard two level converters. The useof fast power devices and the multi leveltopology allow high switching frequen-cies—the border is set by the speed ofthe controller. Active rectifier stageallows regenerative braking, power fac-tor control and DC-Link voltage control.An independent charge balancing regu-lator keeps the voltage levels synchro-nised without interfering with the input or output stage modulation. The opticallink between signal and power process-ing stages provides a fast, safe, reliableand EMI-immune interface. The systempresented in this article is merely aresearch platform, but commercialimplementations of the same technologyare completely feasible.

Figure 3. Output of Experimental Converter.

www.ialb.uni-bremen.de

POWER CONVERSION


Sequencing- and margining-capability simplifies power management

Many applications require multiple voltage rails (e.g. a DSP with different core and I/O voltages),

and depending on the part and manufacturer used, there are different guidelines specifying

how to control two or more voltage rails to minimize the possibility of latch-up or

excessive current draw during power-up and power-down.

By Dirk Gehrke, Texas Instruments

The TPS40100 controller fromTexas Instruments features asequencing capability that simpli-

fies the external circuitry required toimplement these sequencing scenarios,as well as offering many other additionalbenefits over existing controllers. Thedevice is a synchronous buck controllertargeted at applications requiring supplysequencing and output voltage margin-ing features, and is targeted at applica-tions with bus voltages in the range of4.5V to 18V.

Three main schemes are commonlyused to power-up multi-rail power supplies: Sequential Sequencing,Ratiometric Sequencing andSimultaneous Sequencing, and themost appropriate scheme for a givenapplication depends on the requirementsof the device(s) used. Manufacturer'sdata sheets often do not specify whichpower sequencing scheme to imple-ment, but instead outline the voltageand timing conditions to be observed.(Note: Some devices allow out-of-bounds conditions for a short period oftime.) Using the pin conditions and

waveforms presented below, a sequenc-ing methodology can be chosen to meetthe various processor requirements.

Sequential Sequencing requires thatone power supply rail ramps up and set-tles at its final regulation voltage beforethe second power supply rail ramps up(after a predefined delay). This methodcan be used to initialize certain parts ofthe circuit to a known state before acti-vating the rest. An example, in which thecore supply is applied before the I/Osupply, is shown in Figure 1. Such a

scheme can be easily implemented withthe TPS40100 by using the power goodfunction of the core converter to enablethe I/O converter: the I/O supply will beonly enabled once the core supply iswithin regulation. An RC network couldalso be included to add some delaybetween these transitions.

With a Ratiometric Sequencing implementation, both power supply outputs ramp at the same time—and inproportion—until regulation is reached,

as shown in Figure 2. This sequencingscenario can be implemented with theTPS40100 controller by using a commonsoft-start capacitor for multiple controllers.In the drawing shown, three separatecontrollers share a common soft-startcapacitor.

Simultaneous Sequencing is similar toRatiometric Sequencing, in that bothpower supply outputs ramp at the sametime. However, with SimultaneousSequencing, the objective is to minimizethe voltage difference between the twosupply rails during power-up, as shownin Figure 3. This sequencing method isuseful for devices that have “sneakpaths” between supply pins, or drawexcessive current during start-up if internal logic is not in a stable state.Simultaneous Sequencing can be imple-mented by using an uncommitted ampli-fier included in the TPS40100: this canbe used to implement a separate controlloop that regulates the output accordingto a reference voltage applied to the

TRKIN pin. Typically, a master voltageramp such as the I/O supply is used toact as the reference for the ramp upvoltage, however, an RC network (whichgenerates a voltage ramp as its capaci-tor charges up) could also be used.Figure 3 shows what such a ramp upsituation looks like.

Since all of these sequencing scenar-ios can be implemented using theTPS40100, in general it is recommend-ed that placeholders for configurationcomponents are included during thedesign phase, so that changes to thesequencing scheme can be implement-ed by adding or replacing components,rather than redesigning the complete

Figure 1. Sequential Sequencing.

Figure 2. Ratiometric Sequencing.

POWER MANAGEMENT POWER MANAGEMENT

http://www.danfoss.com/siliconpower


system. Such an approach can savetime during development, and oftensaves money during mass production(e.g. if the power supply sequencingrequirements change because theprocessor or FPGA used is changed).

Margining ControlIn addition to the various sequencing

implementations, the TPS40100 incor-porates a so-called Margining Controlfunction. This feature allows the usertemporarily to adjust the output voltageabout its nominal value by a smallamount. This feature is useful during

automated testing of the final PCB dur-ing production because it allows themanufacturer to verify the board’s cor-rect operation at the limits of the powersupply tolerance. A voltage tolerancelevel of ±5% is common for a supplyvoltage like 3.3V, however, in systemsusing lower core voltages such as 1.5V,a tolerance of ±3% is more typical. Thisfeature can also be used to troubleshootcustomer-returned boards exhibitingmarginal performance, thereby reducingthe risk of sending them back with thedescription ‘failure not found’.

The hardware required to implementthis margining feature consists of aninexpensive N-Channel MOSFETconnected between either the MGD(Margining Down) or the MGU(Margining Up) pins and GND, likeshown in Figure 5. In this case, the sup-ply output voltage will change by about±5%, relative to its nominal value. If aresistor in the range 25_ to 37_ is includ-ed in the GND path the part changes thesupply output voltage by ±3%.

Pre-Bias Start-Up The TPS40100 employs synchronous

rectification to improve efficiency, asillustrated by the simplified circuit shownin Figure 4. During power up, someapplications require that the converterdoes not sink current during startup if apre-existing voltage exists at the output.In this drawn example, the core voltagecomes up first and applies I/O voltagethrough the external series diode (orESD diodes in the connected load)before the converter stage generatingthe I/O voltage is enabled. This isdefined as a pre-bias condition, in whichvoltage is applied to the I/O converter’soutput before the converter is enabled.When the converter’s PWM controller isenabled, it soft-starts the high-side FET,and its duty cycle (D) ramps graduallyfrom zero to that required for regulation.However, if, during soft-start, the syn-chronous rectifier (Q2) FET is on whenthe high-side FET is off (synchronousrectifier FET duty cycle = 1-D), the syn-chronous rectifier FET (Q2) sinks currentfrom the output (through the inductor),tending to cause both the I/O and corevoltages to drop.

Since synchronous buck convertersinherently sink current some method ofovercoming this characteristic must beemployed. The method used in this con-troller, is to not allow the synchronousrectifier FET (Q2) to turn on until therethe output voltage commanded by thesoft start up ramp is higher than the pre-existing output voltage. This isdetected by monitoring the internalpulse width modulator (PWM) for its firstoutput pulse. Since this controller usesa closed loop startup, the first outputpulse from the PWM will not occur until

Figure 3. Simultaneous Sequencing.

the output voltage is commanded to behigher than the pre-existing voltage.This effectively limits the controller tosourcing current only during the startupsequence. During this phase inductorcurrent flows through the parasitic bodydiode of the synchronous rectifier FET,rather than through its channel.

If the pre-existing voltage is higherthan the intended regulation point forthe output of the converter, when thesoft start time has completed, the converter will start and sink current.

The need for this pre-bias start-upfeature also arises when using sequen-tial sequencing with certain ASICs, inwhich a leakage path within the devicecauses some I/O voltage to appear onthe core before the core converter is enabled.

Figure 5 shows a typical implementa-tion of the TPS40100, using all its fea-tures. The shaded section shows theoptional implementation of the remotesensing feature, which allows the ana-log ground and the power ground pinsto be separated. All analog circuitry isconnected to the signal ground pin,

with a separate trace connected to theground terminal of the load in order tocompensate for losses in the powerground path. The positive Vout terminal is sensed directly at the load and fedback to the error amplifier section of the TPS40100, marked ‘Vout Sense’. If the remote sensing feature is notused, the PGND and GND pins shouldbe connected together using a singlepoint scheme (i.e. not routed via aground plane).

The controller uses a current feed-back mechanism to make loop compen-sation easier for loads that have widecapacitance variations. Current sensingis used for current feedback and overcurrent detection. The measurement istruly differential and can be implement-ed using the inductor’s DC resistance.This approach requires only a resistorand filter capacitor in parallel with theoutput inductor. The overcurrent level isprogrammable via the network connect-ed to the ILIM pin, and is independentfrom the amount of current feedback.This function provides protection againstshort-circuits and power supply overloads.The implementation in the TPS40100uses a hiccup overcurrent recovery

implementation, that periodically attemptsto recover after an overcurrent conditionhas occurred.

The UVLO (Under Voltage LockoutVoltage) and hysteresis thresholds areprogrammed independently via a resis-tor divider connected to the UVLO pin.This function allows the controller tostart only after the input voltage hasrisen above a certain threshold; the hys-teresis ensures that there is a voltagedifference between turn on and turn off,preventing oscillation if the input voltageis around the UVLO threshold voltage.

The optimum power supply implemen-tation depends not only on voltage, current, size and efficiency, but also onmeeting the complete power managementobjectives for reliability, availability andmaintenance. This requires intelligentpower management, rather than justsimple power conversion. The TPS40100includes these much-needed system-level features, simplifying design andbringing real benefits to the system it is used in.

www.power.ti.com

POWER MANAGEMENT POWER MANAGEMENT

Figure 4. Two converter Output Stages shown during a Pre-Bias Start-Up.

Figure 5. Schematic of typical implementation with all features.


As part of our continuing strategy to deliver the most complete line of circuit protection solutions,

Littelfuse acquired Teccor Electronics, a reputable producer of thyristor products that

includes over voltage protection device, SIDACtor.

By Richard Norris, Littelfuse Inc.

Packages with choice of connectionsand electrical isolation

Athyristor is any semiconductorswitch with a bi-stable actiondepending on p-n-p-n regenera-

tive feedback. Thyristors are normallytwo or three terminal components foreither unidirectional or bidirectional cir-cuit configurations. Thyristors can havemany forms, but they have certain com-monalities. All thyristors are solid stateswitches that are normally open circuits(very high impedance), capable of with-standing rated blocking/off-state voltageuntil triggered/gated to on state. Whentriggered to on state, thyristors becomea low-impedance current path until prin-ciple/main current either stops or drops

below a minimum holding level. Oncethyristor triggered to on state, the trig-ger/gate current can be removed with-out turning off until main current dropsbelow holding current of thyristor.

Littelfuse thyristor line includes components that range from 0.8Arms to 70Arms. These components are produced in surface mount packages as well as heat sinkable through-holepackages.

The heat sinkable devices are alsoavailable in isolated mounting tab pack-ages that are UL recognized. The higher

amperage devices are offered in TO220package, “industry workhorse”, pluslarger TO218 package.

Conventional TO220 and TO218packages are available, but where veryrobust component leads are required,Littelfuse offers a TO218X package vari-ation that has eyelet leads capable ofaccepting #8 gauge stranded wire. Thehigh volume production line TO218Xproducts are viable replacement devicefor similar amperage older Press-Fit and Stud products.

Figure 1. Photo of Standard TO-220. Figure 2. Photo of TO-218. Figure 3. Photo of more robust TO-218X.

TRANSIENT PROTECTION

http://www.eupec.com


Internal construction of LittelfuseTO220 and TO218 thyristor switchingdevices include robust copper alloy connections/contacts from externalleads/pins to semiconductor die. Thisprovides best thyristor surge currentcapabilities for wide variation of loads,such as, locked rotor currents in motorloads and initiating current for tungstenloads and end of life lamp burn-out. AlsoLittelfuse TO220 and TO218 thyristorcomponents are RoHS compliant.

Bidirectional Triac variations areoffered in the TO220 package with up to 35A rating. Triacs are also availablein TO218 package variations with ratingup through 40Arms. These higher cur-rent devices are offered with excellentcommutating and static dv/dt ratings are known as alternistor triacs. Thealternistor triac die design used byTeccor for many years has gate regionlocated in center of die offering excellentpropagation allowing current to spreadquickly for good turn-on and excellentsurge current capabilities. Alternistor tri-acs are excellent solid state switchesused to control a wide variety of loads.

The Alternistor triac was designedwith difficult inductive loads in mind. Itsdie design effectively offers the sameperformance as two Silicon ControlledRectifiers (SCRs) connected inverseparallel (back to back). The die con-struction provides the equivalent of twoelectrically separated SCR structures,but in a single die. The alternistor triac isoffered from 6Arms to 40Arms ratings.

Blocking voltage ratings for these triacsrange from 200 to 1000Volts. For ACloads requiring greater than 40Arms,inverse parallel SCRs such as 65Amp inisolated TO218X package may be usedproducing 92Arms full wave triac.

Typical full wave AC applicationsusing triacs are home and commerciallighting controls, motor controls (includ-ing Permanent Magnet motors), heatingcontrols, AC solenoid controls in appli-ances such as washing machines andlawn water sprinkler systems plus manyother AC solid state switch controls.

For unidirectional thyristor require-ments, Littelfuse offers up to 70ArmsSCRs with excellent characteristics tocontrol a wide variation of loads withvoltage capability up to 1000Volts. Astriacs, SCRs are offered in TO220 pack-age (to 55Arms) and up to 70Arms inTO218 package. Because of its unidi-rectional switching capability, the SCR is used in circuits where high surge cur-rents are present or where latchingaction is required. SCRs also used incircuits where half-wave gate controlledrectification is required.

Typical SCR applications includecrowbars for power supplies, capacitivedischarge control, motor controls, bat-tery chargers, welding machines, tank-less hot water heaters, electrical toolcontrols and many other solid stateswitch applications.

To compliment the SCRs with isolatedmounting tabs, Littelfuse also offers rec-tifiers with isolated mounting tab inTO220 package for full wave bridgecontrolled loads.

Typical applications for SCRs and rectifiers in a bridge configuration aretool controls for maximum motor torquecapability, in exercise equipment suchas treadmills, in conveyors, in food processing equipment and other spe-cialty machinery.

Littelfuse isolated TO220 and TO218packages are very robust for isolatedmounting protection. Unlike over moldedproducts (mounting tab covered withepoxy), each isolated Littelfuse part has an isolation element that is totallyenclosed and sealed with high pressuretransfer molded epoxy with no internalpackage metal singularization pointsexposed on package surface. Thisensures voltage breakdown well beyond2500Vrms rating that is UL recognized.The isolation element also has goodthermal characteristics to allow heat tobe removed from the thyristor package.The high-pressure transfer moldingepoxy/compound also has best UL 94V0flammability ratings and insures bestoperating and storage temperatures.

Detailed product information can befound at Rectifiers:http://www.littelfuse.com/cgi-bin/r.cgi/prod_series.html?LFSESSION=GKmXmvDBzZ&SeriesID=434

Figure 4. Isolated TO-218 construction. Figure 5. Typical TO-220 Overmolded Device.

SCRshttp://www.littelfuse.com/cgi-bin/r.cgi/prod_series.html?LFSESSION=GKmXmvDBzZ&SeriesID=433

Sensitive SCR datasheethttp://www.littelfuse.com/cgi-bin/r.cgi/prod_series.html?LFSESSION=GKmXmvDBzZ&SeriesID=432

Sensitive Triacshttp://www.littelfuse.com/cgi-bin/r.cgi/prod_series.html?LFSES-SION=EMoiBr66Ko&SeriesID=431

Alternistor Triachttp://www.littelfuse.com/cgi-bin/r.cgi/en/prod_series.html?SeriesID=593&LFSESSION=yPiD7w8EdK

Triachttp://www.littelfuse.com/cgi-bin/r.cgi/en/prod_series.html?SeriesID=592&LFSESSION=yPiD7w8EdK

The Littelfuse portfolio is backed byindustry leading technical support,design and manufacturing expertise.The products are vital components invirtually every product that uses electri-cal energy, including automobiles, com-puters, consumer electronics, handhelddevices, industrial equipment, and tele-com/datacom circuits. Littelfuse offersTeccor, Wickmann, Pudenz and Efenbrand circuit protection products.

www.littelfuse.com

TRANSIENT PROTECTION

http://www.epcos.com

Depending on the response timeand duration required, a varietyof technologies are suitable for

meeting power quality and reliabilityneeds. Ultracapacitors are ideally suit-ed for power needs from tenths of sec-onds up to as much as one minute—including uninterruptible power supply(UPS), voltage stabilization and energy recapture.

Ultracapacitors in UPSFor many years batteries have played

a major role in the design of industrialUPS systems, mainly due to their rela-tively low cost. Batteries have histori-cally required significant maintenance,frequent replacement and have provenunreliable. Reliability concerns haveprompted alternative technologies suchas ultracapacitors to be considered forUPS applications.

Ultracapacitors have proven capablefor maintenance-free operating lives inexcess of twenty years. For applicationsrequiring seconds to minutes of backuppower, ultracapacitors have proven tobe cost competitive. For backup times inexcess of minutes, ultracapacitors havebeen coupled with generators and fuelcells to provide seamless power transi-tion during start up of the primary power source.

A newer use of ultracapacitors withinUPS has been for diesel generator start-ing applications. In this example batter-ies have been either replaced or signifi-cantly downsized depending on theapplication needs. This has been espe-cially successful for cold weather appli-cations where batteries are less reliable.The ultracapacitor is used to provide thecold cranking amps, while the nowsmaller battery supports longer durationsustaining power, to keep the enginerpm maintained during sluggish cranks.

Voltage stabilization and energyrecapture

Voltage stabilization and energy recap-ture are two proven applications whereultracapacitors help the reliability ofpower, as demonstrated by several suc-cessful deployments at railway stations.

These systems operate in either avoltage stabilization or energy savingsmode. In the voltage stabilization mode the energy storage unit preventsa minimum line voltage on the systemas trains enter and leave the station. Inthe energy savings mode, braking ener-gy is stored and utilized for acceleratingtrains away from the station. One instal-lation has reported an annual energysaving of 320,000 kWh with a potentialof 500,000 kWh.

Another example of voltage stabiliza-tion is windmill farms. Sudden changesin wind speed cause significant voltagefluctuations leading into the main powergrid. To smooth out these fluctuations,ultracapacitor energy storage systemsare being deployed and evaluated.

Power conditioning and controlFor most applications, additional

power conditioning may be necessaryon the input/output of the ultracapaci-tors. Ultracapacitors are DC deviceswith time constants of approximatelyone second. High frequency AC or dirtyDC waveforms will require filtering toavoid overheating the ultracapacitors.This may be accomplished utilizingstandard electrolytics in parallel with the ultracapacitors. Proper convertersmay be required to either boost or buckregulate for providing proper voltage tothe system if constant output voltage is required.

Consideration should be made for thefrequency of charge/discharge and therate of recharge required.

For the voltage stabilization examplesat the rail stations, the ultracapacitorsprevent low voltage sags which mayrequire quick discharge capability. Theultracapacitors are then able to be


With ever increasing demands for cost competitiveness and conservation of resources,

industrial companies and public facilities are looking for improved power consumption

and reliable power systems.

By John Dispennette, Maxwell, Application Engineer

Improving reliability and power consumption

SUPER CAPACITORS

http://www.micrel.com


and low ultracapacitor voltage producesthe lowest switching frequency. Thehighest switching frequency occurs atthe peak of the AC power line at maxi-mum voltage and full charge voltage onthe ultracapacitor. Depending on theapplication the switching frequency cancover a range of 20:1. When C1 reach-es its maximum voltage, then the volt-age sensor will drive the control circuitinto discontinuous operation.

The output voltage of the ultracapaci-tor to supply the required electronicapplication will vary as current is drawnfrom the capacitor module. As previouslymentioned if the device is voltage sensi-tive a DC/DC boost converter may beadded enabling a constant input voltage to the device.

If the ultracapacitor module is allowedto operate between its full rated voltageand half rated voltage the module cansupply 75% of its available energy.

Selection of an appropriate DC/DC converter is dependent on the operatingvoltage range of the system and thewattage requirement for the application.

The example charging circuits andapplications show how ultracapacitorshave been designed or could be used in industrial applications. Specific circuitry is dependent on the needs ofeach application.

www.maxwell.com


slowly recharged with a DC source, see figure 1.

Other applications may need addition-al control to be maintained between theultracapacitors and the grid interface.This may be due to differences in volt-age between the ultracapacitor and themain line.

An example of this circuit is providedin figure 2. The first step takes the ultra-capacitor voltage to an intermediate DCvoltage through a reversible boost con-

verter. The second stage employs a DC buck converter to a high level DCvoltage to the main line. This topologyallows complete control of the ultraca-pacitor currents and the main line inde-pendent of the voltages.

Charging ultracapacitors from an ACor DC source

Ultracapacitors may be linked to eithera DC or AC source. To understand howultracapacitors may be integrated intoan industrial design, let’s look at exam-ple charging circuitries for both sources.

A DC/DC constant current regulator isthe simplest form of active charging. Itcan be operated as either a buck orboost regulator depending on the appli-cation. The buck regulator is the pre-ferred topology for larger fast charge,high duty cycle applications due to thecontinuous charge current. An examplecircuit is provided in figure 3 represent-ing a constant current charging sce-nario. A simple constant current chargermay be built with standard power supplyICs. The current limit is set to therequired charge current and the voltagelimit would be set to the maximumrequired voltage.

When charge time is critical, constantpower charging provides the fastestmethod. It can transfer all the availablepower from the charge source into theenergy storage capacitors. Drawing aconstant current from the source at aconstant voltage is a simple implemen-tation of constant power charging. Thisusually requires a maximum switchingcurrent of 2.5 times the nominal to pre-vent overloading the switching circuitrywhen the ultracapacitor voltage is below40% of maximum. An example is shownin figure 4 (Patent pending).

If a device is to be directly chargedfrom an AC source it is often difficult tocover the wide dynamic range require-ments for charging ultracapacitors froma varying AC power line. An example cir-cuit is provided in figure 5 (U.S Patent6,912,136) for AC charging scenarios.The circuit uses the L/V characteristicsof the switching transformer to set theswitching frequency. Full output currentat zero volts is capable with this designwith no risk of saturating the magnetics.

The switch, Q1, turns on charging theprimary of T1 to a preset current limit.Q1 then turns off, permitting the energystored in T1 to discharge through D1into the ultracapacitor, C1. When thesecondary current has discharged to apreset lower limit then Q1 will turn onagain repeating the cycle. The timerequired to charge T1 is inversely pro-portional to the instantaneous line volt-age and the ultracapacitor voltage atC1. The combination of low line voltage

Figure 1. Railway circuit interface example.

Figure 2. Two stage power converter.

Figure 3. Constant current charging circuit.

Figure 4. Constant power charging circuit.

Figure 5. AC charging circuit.

SUPER CAPACITORS SUPER CAPACITORS


Latest generation of low voltage power MOSFETs

The forward-biased safe operating area (FBSOA) in a MOSFET defines the permissible and

safety operation locus in terms of drain current (ID) and drain-to-source voltage (VDS).

This area is limited by maximum allowable voltage, current and power dissipation. In modern low

voltage power MOSFETs, experimental data have shown an anomalous failure mode which limits

the theoretical FBSOA at low drain current and its ability to operate in linear mode.

By G. Consentino and G. Bazzano, STMicroelectronics

A Mathematical Model andExperimental ResultsThe parameter identifying such thermalinstability is the thermal coefficient T.C.is mathematically defined as the deriva-tive of the drain current versus tempera-ture. A positive value of the thermalcoefficient means that as temperaturerises, due to power dissipation in linearoperation, drain current also rises.Every low voltage MOSFET exhibitsranges of drain current ID where thethermal coefficient can be either positiveor negative but in general safety opera-tion in linear mode depends on the cur-rent range over which the coefficient ispositive and also on its peak value.Sometimes, even if the thermal coeffi-cient is positive no failure occurs in thedevice because the junction- ambientthermal resistance Rthja can be suffi-ciently low to avoid thermal runaway.The failure mechanism observeddepends on certain intrinsic deviceparameters and its structure.

Theoretical derivation of the thermalcoefficient

Id in the linear zone can be written as:(1.1)

(1.1)

μ is the carriers’ mobility, COX the oxidecapacitance, VTH the threshold voltage,W the channel perimeter, L the channellength and VGS the applied gate-to-source voltage. In subsequent calcula-tions we will define K=1/2_W/L so wecan re-write as:(1.2)

Under these conditions, the T.C. has thefollowing general expression:(1.2)

The terms calculated forconstant VDS are both negative.

In order to compare the performanceof different MOSFET devices, the T.C. isnormalized versus active area or chan-nel perimeter as:(1.3)

We can calculate now the maximumvalue of the T.C.:

(1.4)

The drain current range for which thecoefficient is positive is expressed by:(1.5)

Now, we will take into account theeffect on the thermal coefficient by thecarriers’ saturation velocity when a highelectric field is applied across the chan-nel,see figure 1.

Two models can approximate thecurve of figure 1, the former being(1.7a)

and the latter(1.7b)

POWER SEMICONDUCTORS

http://www.apec-conf.org

38 Power Systems Design Europe December 2005 www.powersystemsdesign.com 39

Thus(2.5)

In other words, we also have a furtherreduction of the threshold voltagecaused by an increase in VDS.

Another factor to consider is the effectof the short channel (0.6μm and less)because the junction depletion widthbecomes comparable with the electricalchannel length. Such effects show upwith a further decrease in threshold voltage as VDS increases.

Conditions for thermal stabilityThe heat generated inside the silicon

junction is due to the electrical powergenerated during the device’s linearoperation, thus:

(3.1)

When an electrical power impulse isapplied to the device, the temperaturerises and the drain current changes itsvalues depending on the thermal coeffi-cient. The electrical power variation willbe written as:

(3.2)

Such electrical power pulse due to thetemperature rise can be supported fromthe device if:

(3.3)

Instead, if the following conditionoccurs, the electrical flux exceeds themaximum thermal flux involving the failure mechanism of the device.

(3.4)

By equaling the left-hand side term of (4.5) to the right-hand term, (4.5) canbe written as below and it establishesexactly the maximum value of the ther-mal coefficient where the failure of thedevice is avoided.

(3.5)

A real example will be taken into con-sideration. The main characteristics ofthe device tested are shown in table I.

Figure 4 depicts the normalized ther-mal coefficient curves versus ID for thesecond model shown earlier consideringVDS equal to 8V and 20V, respectively.From (3.5) we can establish the right IDpositive range where the device couldfail considering Rthja equal to 1.5°C/W.

Technology comparison between planar and Trench structures

In this section we will show a realexample which compares one of ST’splanar MOSFET (STripFET) versus oneof the competitor’s trench. The main dif-ference between planar and trenchstands essentially in the gate electrode

structure. In fact, in the planar technolo-gy, the gate electrode is placed on thesilicon surface while in the others thegate is placed inside the epitaxial layer.

Due to trench gate shape, the oxidethickness in the trench device is highercompared to the planar technology. Inorder to fix the right threshold voltage,considering that the trench technologyhas a higher gate oxide thickness, it istherefore necessary to decrease thecarrier concentrations in the channel.

In table 2 are summarized the mainparameters of the two devices undertest, the ST part being planar.

The T.C.s, considering VDS of 15V, forboth devices are shown in figure 6.

From this graphs it is possible to seethat, even if the positive thermal coeffi-cient range is equal in the two devices,its peak value in the trench device isaround three times higher than that ofthe planar device. Such behavior involvesthe best operation mode in linear zonefor the planar device.

Pspice SimulationTo validate the empirical results, the

Pspice simulator has been used. In fact,the T.C. and electrical characterists hasbeen obtained by Pspice simulator cre-ating an model that include also the suit-able thermal coefficients. The graphsshown in figure 7, figure 8 and figure 9compare the real and the simulated data.

Figure 5.

so the T.C. in case of (1.7b) can be written(1.8)

Impact on the T.C. induced by channel length

When VDS increases, with VGS con-stant, the device works in the linearzone and channel length modulationoccurs. In particular, the electrical effec-tive channel length becomes shortercompared to the physical channel lengthinvolving a little rise on ID (see figure 2).

When VDS exceeds (VGS-VTH), theeffective channel length, Leff, becomesshorter compared to L; the voltage appliedacross the channel remains fixed to

VGS-VTH, while, across (L-Leff ) a voltage equal to VDS-(VGS-VTH) is applied

From (1.1) after some manipulationswe can obtain: (2.1)

ID(Leff) and _L can be written as:(2.2)

(2.3)

Considering that the threshold voltagedepends on temperature also ΔL dependson this parameter as:(2.4)

Figure 1. Normalized Carriers Drift Velocity vs NormalizedElectric Field.

Figure 2. Channel Length Modulation.

Figure 3. Cross Section. Figure 4. Normalized Thermal Coefficient Curves.

Table 1. Device Characteristics

POWER SEMICONDUCTORS POWER SEMICONDUCTORS


Fast, safe and efficient solutions

Safety, performance and power consumption criteria are driving the development of highly efficient

photocoupler devices. Electrical isolation is a familiar requirement in the design of many systems.

Motor controllers, power supply electronics, data communications and general-purpose

switching-type applications all require components that form a reliable,

controllable electrical barrier between two parts of a circuit.

By Josef Abels, Toshiba

Isolation serves a number of purposes.For instance, the intrinsic function ofdevices such as relays and switches

is to toggle between on- and off-statesas part of a control system. Transistorcouplers and triac couplers, in contrast,can more often be regarded as signalconversion devices that allow logic-levelelectronics to control higher currentsand voltages. At the same time, thesedevices may also protect the sensitiveelectronic portion of the system from thepotential damage that can be caused byaccidental contact with power circuitry.Just as important is their role in protect-ing the human user of the equipmentfrom injury, and in preventing the trans-mission of noise from one part of a cir-cuit to another—in effect to protect thetwo parts of the circuit from “injuring” the signals of the other.

Optical couplingDesigners are increasingly choosing

to implement many of these isolationtasks, which have traditionally been per-formed by electromechanical or woundcomponents, using optically coupled

devices. This move is fuelled not only bythe increase in the sheer number of sys-tems—from refrigerators to factoryautomation—that are electronically con-trolled. It is also driven by the ability ofoptical devices to provide better per-formance, in terms of switching speed,isolation voltage, and power consump-tion, in ever-smaller packages.

The principle of operation of photo-coupler components is simple. Theyaccept an electrical input, which drives a light-generating device—typically anLED. The photons produced travel with-in the component packaging across aphysical gap (hence providing the elec-trical isolation), to a photodetector,which can be used to drive a range ofoutput circuitry, also integrated withinthe package. Component manufacturerssuch as Toshiba Electronics offer abroad range of such devices whose outputs are able to drive a variety of circuits, including inverters, medium orhigh-power IGBTs, MOSFETs or intelli-gent power modules (IPMs).

The widening variety of componentson the market today makes it increas-ingly possible for designers to choose a device that will directly drive theirrequired output circuitry and, therefore,eliminate the need for external compo-nents. Nevertheless, straightforwardtransistor couplers remain an importantelement in the product mix and continueto develop in terms of performance, for applications as wide-ranging asmodems, air conditioning units, andPLC control in factory automation.Toshiba’s TLP283 exemplifies the typi-cal trend: it features an input current aslow as 2mA - half that of previous gener-ations – and has a switching time that isless than 100μs. The devices contributeto system size reductions as they areavailable in an SOP package, with anoption of one or four couplers in a singlepackage. The combination of fast andreliable switching with small size andlow power make the TLP283 couplerssuitable for applications such as on-board power supplies, programmablecontrollers, I/O interface boards and ACadapters for PDAs.


ConclusionsThis paper has analyzed the anomalous failure mech-

anism encountered on the latest generation of low voltage low on-resistance power MOSFET devices con-sidering a detailed mathematical model of the thermalcoefficient. Experimental results have also been associ-ated to theory showing a good matching.

Furthermore, this article has performed a comparisonbetween two different MOSFET structures, competition’strench and ST’s planar technologies, in order to under-stand which one exhibits the best performance in linearoperation. To this purpose two devices with similar elec-trical characteristics have been chosen. Results confirmthat the planar STripFET is the most suitable device tooperate in linear mode owing to lowest thermal gradientof VTH and lowest oxide thickness tOX.

Finally, a Pspice simulation of the electrical character-istics and TC were performed. The comparison betweenthe real and simulated curves shows a good matching.

Table 2.

Figure 6. Therma Coefficient by Active Area Trench vsPlanar @ 15V.

Figure 7. Drain Current vs Gate Drain Voltage @ Tj 25 °C and@ Tj 125 °C.

Figure 8. Drain Current vs Gate Source Voltage @ Tj25 °C and @ Tj 125 °C.

Figure 9. Thermal Coefficient Simulated vs measured.

www.st.com

POWER SEMICONDUCTORS OPTOELECTRONICS


drive to data communications that combine high-speed switching withindustry-standard (VDE, UL, SEMKOand TUV) isolation.

The TLP705 device, for instance, isdesigned for high-speed driving ofIGBTs and MOSFETs. With a switchingtime of less than 200ns and 3mA currentdrive capability, the device can be usedas a high-speed inverter in factoryautomation applications, and to driveplasma display panels (PDPs).

For inverter drive and UPS applica-tions, the company’s TLP350/351devices rely on another innovation inmanufacturing techniques: they are fab-ricated in a BiCMOS process, reducingtheir current consumption requirementsby around 80% when compared withToshiba’s previous generation ofdevices. The TLP350 includes a UVLOfunction, and delay time is guaranteedat 0.5μs over an extended temperaturerange of -40°C to 100°C. The devicesare housed in space-saving 8-pin DIPpackages, but perhaps the most impor-tant contribution to miniaturisation inapplications such as motor drives isrooted in the TLP350’s lower power con-sumption. Such applications typicallyrequire the inclusion of relatively largebypass capacitors between power sup-ply and ground pins: by reducing powerconsumption so substantially, the device

allows the designer to choose smallerceramic capacitors over the bulky elec-trolytics that would otherwise have beenrequired. Figure 3 shows how such ICcouplers can be employed in a typicalIGBT-based inverter system.

The increased market acceptance ofIPMs has also led to the development ofIC coupler products specifically for usein this area. Devices such as Toshiba’sTLP102/106 provide totem pole outputsat voltages of up to 20V. Whilst reducingexternal component count by eliminatingthe need for a pull-up resistor, thesedevices also exemplify the overall mar-ket trend towards smaller packages,higher speed and better isolation: supplied in 5-pin MFC, their switchingtime is 0.4μs, and isolation voltage is3750V rms.

PhotorelaysThe final area where photocoupler

products are making inroads is inreplacing the once ubiquitous electro-magnetic relay. Photorelays consist ofan LED coupled with a photoMOSFETand offer many intrinsic advantagesover their older competitors, not leastthe ability to be directly driven by logicor a microcontroller. They switchapproximately five times faster; theirpower consumption when switching isas much as two orders of magnitudelower; they are small (and can be further

miniaturised); and they offer inherentlylow-noise operation as they suffer fromnone of the contact bounce and back-EMF generation problems of their elec-tromagnetic counterparts.

The key figure of merit for compo-nents in this area is the product of on-resistance and output capacitance: cur-rent state-of-the-art developments focuson reducing this figure to below 1pFμΩ,and on cutting package size. Componentsin SSOP4 packages with dimensions of2.04mm x 3.8mm are already available,whilst isolation voltages are up to 2.5kVand a wide range of current-handlingcapacities are available.

SummaryOverall, the move to optical isolation

is part of a wider trend towards semi-conductor-based operation in all aspectsof product manufacture. By eliminatingwound and electromagnetic components,designers can reap considerable bene-fits: by deploying one of the increasingrange of photocoupler componentsintended to directly serve their applica-tions, they can simplify their designseven more. Manufacturers such asToshiba will continue to increase theusefulness of their devices by integrat-ing more circuitry, by offering a greatervariety of options in terms of drive out-puts, by continuing to miniaturise and byfurther reducing power consumption.

www.toshiba-components.com

Figure 3. IC couplers in a typical IGBT-based inverter system..

Triac couplersFor AC systems, triac couplers are

increasingly used, as illustrated in the typical application circuit shown in Figure 1.

Devices such as Toshiba’s TLP36xcouplers use a gallium-arsenide (GaAs)infrared LED, which optically couples toa phototriac. Again the need for minia-turisation is a key driver in this area.However, it is also balanced with cost-sensitivity and the need for standardscompliance via reliable isolation in applications such as office automation,

in which the isolator provides safety pro-tection for the human user. The TLP36xdevices provide 5kV of isolation in 4-pinDIP packages, 35% smaller than thespace previously required to achievethis level of isolation. Furthermore, atjust 20V, the inhibit voltage of thesedevices is up to 50% lower than previ-ous generations of triac couplers, leading to significant reductions in turn-on noise.

Some of the advances in the phototri-ac field have been made possible by thedevelopment of new methods of device

construction. As its name suggests, thetraditional “face-to-face” topology putsthe light-emitting device opposite theoutput circuit within the package. In “single-mould” construction the two ele-ments are sited face-to-face, separatedby a single dielectric, whilst “double-mould” fabrication sees both parts giventheir own moulded insulation (althoughthey remain face-on).

The most recent space-saving alter-native, however, is reflective-type cou-pling, which uses a single transparentresin to achieve optical coupling. TheLED and phototriac are actually side-by-side in the package, but optical couplingis still achieved because the devices areencapsulated in the same resin mould-ing, which is reflective at its boundaries.This allows components such asToshiba’s TLP263 to offer up to 3kV ofisolation in small, cost-effective surfacemount packages, with full UL and VDEsafety approvals.

IC CouplersLike most components today, higher

levels of integration are being requiredof photocoupler devices, leading to thedevelopment of photo-IC couplers thatcan be used to drive IPMs, IGBTs andFETs. Applications such as communica-tions network equipment and drivingmotors or flat-panel displays requireadvanced functionality: for instancehigh-speed operation in the Megahertzrange or undervoltage lock-out (UVLO).Designers also require flexible options,for instance the choice of open-collectoror totem pole operation, depending onthe nature of the load and whether cur-rent sinking and sourcing is required.

All of these requirements, naturally,come along with the increasing expecta-tions now familiar in the semiconductorindustry: reduction in power consump-tion (and supply voltage); smaller size;and higher drive currents, sometimesreaching several Amps.

Pressures such as these have ledToshiba to develop its TLP70x andTLP71x series of photo-IC output cou-plers, which are the industry’s smallestsuch devices with 5kV isolation. Theyare aimed at applications from IGBT

Figure 1.Typical application of triac couplers.

Figure 2. TLP70x and TLP71x series of photo-IC output couplers.

OPTOELECTRONICS OPTOELECTRONICS


Adetailed consideration, account-ing for the worst case operatingconditions, is compellingly nec-

essary. By using Infineon´s dimension-ing program, IPOSIM, for loss and ther-mal calculation, this procedure can beconveniently accomplished.

Module SelectionAt present eupec’s product spectrum

covers over 300 module types in diversehousing designs and current ratings.

As a first step the characteristicparameters of a yet to be determinedoperation point has to be defined. Valueswhich are affected by these parametersare indicated in parentheses:

• DC link voltage and switching frequency fsw (switching losses)

• Modulation factor m and cos ϕ(conduction losses)

• RMS current (switching and conduc-tion losses)

• Output frequency f (temperature ripple)

• Case and max. Junction tempera-ture (thermal utilization)

After entering the parameters into the main input mask, IPOSIM will makea pre-selection of all suited modulesfrom the available product spectrum. Arestriction to a specific module housingis possible. The applied criteria to selecta module are:

• Suited module voltage class as pera given DC link voltage

• Adherence of the module’s peakcurrent capability which is limited byits RBSOA to Irms ≤ 2 Inom / √2

• Compliance with the thermal limits;due to the presetting of case andmaximum junction temperatures, thetotal losses of IGBT or diode mustfulfil Ptot*Rthjc ≤ Tj -Tcase

Loss calculationThe Infineon application note [1]

covers how to calculate the major IGBToperating parameters by using thedatasheet as a source for device char-acteristics. These calculations deliverpower dissipation, total power losses aswell as junction and heatsink temperature.

The presented methods are applica-ble for continuous and pulsed collectorcurrents but will not be adaptive for voltage source inverters with naturallysampled PWM and sinusoidal outputcurrents. For these common applica-tions [2] gives an approach to calculateaverage and total power dissipation ofIGBT and diode. It uses a closed solu-tion for the conduction and switchinglosses without summing up severalswitching energies of the single switch-ing instants.

The conduction losses of an IGBTwith sinusoidal current î sin(ωt) aregiven by

Where VCE0 and r are threshold voltageand slope of the output characteristic.

The conduction losses can bedescribed by

Find the right IGBT module for your inverter

There is no simple criterion or rule of thumb available to select a properly suited IGBT module.

This is due to the multiplicity of inverter applications in traction and industry, the individual condi-

tions regarding DC input and AC output voltage, switching and output frequency, inverter power

rating, overcurrent and overload capability as well as the influence of different cooling conditions.

By Dr. Thomas Schütze, Infineon

IGBT MODULES

http://www.e-emv.com


IPOSIM assumes a half-wave sinu-soidal shape for the average powerlosses Pav. Only with this assumptionthe distinct fluctuation of the junctiontemperature is considered properly. Theripple and therefore the peak junctiontemperature increases at lower frequen-cy and decreases with higher frequency.

The mentioned setting of a case tem-perature can only be estimated. This issufficient for a module comparison andas a first guess. By defining a heat sinkand with the easier to appraise maxi-mum ambient temperature (e.g. 60°Cfor water and 40°C for air cooled sys-tems) an accurate junction temperaturecalculation can be achieved.

Figure 3 visualizes the temperaturedistribution across the different layersjunction-to-case, case-to-heatsink and heatsink-to-ambient for differentload currents.

A detailed evaluation of the ripple(Figure 4) gives minimum and maximumjunction temperature in the consideredoperation point as well as the maximumachievable current while keeping thelimit of Tjmax e.g. below 125°C. Bychoosing between IGBT and diode thecalculation can be performed for theIGBT as well as for the diode part of themodule. In most cases, the highest temperatures for IGBTs will be reachedin motor operation while diodes willreach highest temperatures in regenera-tive operation.

Load cycle calculationSo far IPOSIM assumed steady state

operation with the primarily enteredparameters. Further worksheets offerthe option to not only calculate lossesand temperatures for single operationpoints, but also for complete loadcycles. With sampling points of chang-ing parameters it is possible to freelydefine application specific operationalcycles with up to fifty sets of data. Youhave the option to enter, save and lateron restore and if necessary modifythese cycles.

During the previous steady stateapproach, there was no need to definethe transient characteristic for the heatsink. Considering now fluctuating cyclesin the time frame of seconds, the tran-sient response of the heat sink must beconsidered by a time constant.

The calculation finally ends up in loadcycle diagrams which give a graphicpresentation of the input data, the calcu-lated losses and resulting temperaturesfor IGBT, diode and heat sink. Figure 5shows a typical example for a tractioncycle comprising acceleration, coastingand braking phase.

The most current version of IPOSIM,as well as an additional technical docu-

Figure 3. temperature distribution across junction-case-heatsink-ambient vs. loadcurrent.

Figure 4. losses and junction temperature ripple for one period.

The nominal switching losses (at datasheet conditions) are scaled by the cur-rent and DC voltage at the respectiveoperation point. For a more detailed der-ivation of these formulas see [2] or [3].The calculation results in a diagramshowing the average power losses Pav

for IGBT and diode at sinusoidal cur-rents versus the RMS phase leg current (Figure 1).

Thermal calculationIGBT as well as diode are conducting

for 50% of the cycle-time of the sinu-soidal output. Losses in these devicesappear for one half-period at a time,while the IGBT or diode in the opposite

inverter position is not active. The junc-tion temperature therefore oscillateswith the output frequency. There arethree approaches for its thermal calculation:

• By using the average losses Pav, we achieve an average junction temperature Tjav = Pav * Rthjc. Due to the ripple on the output, the peakjunction temperature will exceed thecalculated Tjav.

A simple approach is to assume thatthe average losses Pav are dissipated ina rectangular shaped block of two timesPav during one half period of the phasecurrent. The simulation (Figure 2) shows

a comparison of the junction tempera-ture rise for sinusoidal (blue) and rectan-gular (red) shaped power losses p(t) at50, 5 and 1 Hz respectively with thesame value for the average power Pav.

At 50 Hz the junction temperatures(bold curves) behave quite similar. At 1Hz it becomes obvious, that due to thelow output frequency and its relation tothe time constants of the transient ther-mal resistance junction-to-case, thetemperature closely follows the shape ofthe predetermined loss curve. Thereforea rectangular approximation is not gen-erally applicable for sinusoidal losses!

Figure 1. IPOSIM entry mask and ‘average power losses vs. RMS phase leg current’ diagram.

Figure 2. junction temperature for sinusoidal and rectangular shaped power loses at 50, 5 and 1 Hz.

IGBT MODULES IGBT MODULES


EPCOS has added the SpeedPowerII and III series to its SpeedPower rangeof tantalum chip capacitors because moreand more applications call for diversifi-cation of equivalent series resistance(ESR) as well as of capacitance andrated voltage in capacitors. ESR is oftenthe critical factor for the entire design.

The SpeedPower II series extendsthe range of capacitors in case sizes B

to E with several gradations in the ESRlimits. This series also introduces tanta-lum capacitors in case size A.

The SpeedPower III series featuresan optimized anode design with a largersurface area for even lower ESR.

Both series cover the ESR range from40 m_ to 15 _ at rated voltages of 2.5 to50 V in case sizes A to E. If the low-ESRtypes are used, connection of capacitors

in parallel can be avoided. This leads tomore compact designs, enhanced relia-bility and reduced placement costs.

The capacitors selected are usedthrough-out electronics. Typical applica-tions will be found in telecommunications,automotive, consumer and industrialelectronics, data systems, and all appli-cations where space is at a premium.

www.epcos.com/tantal_capacitors

Tantalum Chip Capacitors

Advanced Power Technology Europe ispleased to announce the addition of 100V

MOSFET modules to the existing prod-uct range in SP3, SP4, and SP6 packages.

These modules are offered in singleswitch, buck, boost, dual commonsource, phase leg, full bridge, asymmet-rical bridge, dual boost, and dual buckconfigurations. Current ratings rangefrom 50A to 600A @ TC = 80°C.

Single switch, phase leg, and fullbridge configurations integrate FREDFETdevices for improved body diode recoverycharacteristics and better dv/dt capability.High current single switch modules areoffered in standard with AlN substratesfor best thermal characteristics.

With 12mm height for the SP3 and17mm height for the SP4 and SP6 pack-

ages, these modules exhibit minimuminternal parasitic resistance and induc-tance, offering high frequency operationcapability. The use of PowerMOS VFREDFETs allows operating in hardswitching mode in the range of severaltens of kHz without the need for seriesand parallel diode combination. All mod-ules offer very low thermal resistancewhich leads to high current capability fora given RDS(on) switch resistance.

Main applications for this extendedfamily are motor control, UPS, powersupplies, and all other equipment oper-ating from battery voltages in the rangeof 24V to 48V.

www.advancedpower.com

ABB is producing the first VoltageDetector VD1500, adapted completelyto meet the major specifications of thetraction market. The new ABB VoltageDetector VD1500 has been designedincorporating a new ABB innovation, a100% electronic solution. This safety

product will be used in applications suchas on-board railway equipment (mainconverters, auxiliary converters) and inpower electronics systems (batterychargers, choppers, sub-stations, etc.).It allows the maintenance operator toverify the presence of a dangerous volt-age (AC or DC) up to 1500V on anyequipment. Dangerous voltage indica-tion is provided via 2 independent LEDsthat flash when the voltage detected ishigher than a specified limit (maximum50V according to standards such asCF60-100). The mechanical and elec-tronic concept ensures an excellent reliability of the product through twoindependent electronic boards (redun-dancy of the same function in only oneproduct). The Voltage Detector (VD)operates from –40°C up to +70°C. TheVD range has been designed to comply

with traction standards: EN50155 forhigh-tech electronic design and tests,EN50124-1 for electrical insulation andEN50163 for voltage. The VD range hasalso been designed in compliance withsafety standard EN50129 for communi-cation, signalling and processing systems.

The ABB Voltage Detector provides acost-effective solution that is easy toinstall and use.

ABB has been concerned for longwith the protection of the environmentand has been ISO 14001 certified in1998. This environmental approach isparticularly noticeable in the productionof the VD range with the reduction of thenumber of components, use of a low-energy manufacturing procedure andrecyclable packing.

www.abb.com/lowvoltage

ABB Voltage Detector

NEW PRODUCTS

100V MOSFET Modules

Figure 5. load cycle diagrams.

mentation, are available for downloadon eupec’s homepage („Products &Solutions” and then „Simulation Tools”).

References1 nfineon Technologies Application noteANIP9931E: Calculation of major IGBToperating parameters

2 D. Srajber, W. Lukasch: The calcula-tion of the power dissipation for theIGBT and the inverse diode in circuitswith the sinusoidal output voltage; electronica ´92 Proceedings, pp. 51-58

3 eupec Technical documentation:Dimensioning program IPOSIM for loss and thermal calculation of eupecIGBT modules

www.infineon.com

www.eupec.com

IGBT MODULES


NEW PRODUCTS

www.necel.com

To further strengthen its automotivesemiconductor business, NEC ElectronicsCorp. and its subsidiaries in the UnitedStates and Europe, NEC ElectronicsAmerica, Inc. and NEC Electronics(Europe) GmbH, announced theμPD166007, an intelligent power device(IPD) for use in on-board electronic con-trol units (ECUs) supporting applicationssuch as headlights, anti-lock brakingsystems and air conditioners. With this

new device, the mechanical switches or relays conventionally used to controlthese units have been replaced bysemiconductors to enable smaller andlighter ECUs with improved on/off controland high reliability. This innovation freescabin space and contributes to a lowerenvironmental impact by reducingengine load. Further environmental ben-efit is achieved through the μPD166007IPD’s lead-free design.

The device features a specific switch-ing control to limit rapid fluctuations inoutput current. This reduces the electro-magnetic noise and improves ECU per-formance. The μPD166007 IPD sup-ports a current sensing function to moni-tor current flow during normal operation,and a diagnosis function to detect over-current and overheating. Support forthese features allows an accurate grasp

of the output status and the detection of abnormalities.

The μPD166007 uses a stacked con-struction in a multi-chip package (MCP).The device also features vertical typefield-effect transistors (FET) using aUMOS trench cell structure. This struc-ture has a solid track record for lowresistance and low heat generation, and with an on-resistance of 10 mW(see note), the μPD166007 is ideal foruse in on-board control units. Thedevice comes in a compact TO-252package with 5 external pins in line with JEDEC standards.

Samples of the μPD166007 IPD areavailable now. Mass production ofapproximately 1 million units per monthis scheduled to start in the second quarter of FY2006.

Automotive Intelligent Power Device

Vishay announced that it is shippingthe first two power MOSFETs to beoffered in its innovative PolarPAK pack-age, which uses double-sided cooling toreducing thermal resistance, packageresistance, and package inductance for a more efficient, faster switchingpower MOSFET.

PolarPAK was specifically designedfor easy handling and mounting onto the PCB with high-speed assemblyequipment and thus to enable high

assembly yields in mass-volume produc-tion. This is one reason why PolarPAKhas already earned the distinction of beingthe first MOSFET package with double-sided cooling to be sourced by multiplemanufacturers, beginning with a licenseagreement concluded between Vishayand STMicroelectronics in March 2005.

The first PolarPAK power MOSFETsare the 30-V n-channel SiE802DF andSiE800DF. Optimized for the low-sidecontrol switch in synchronous rectifica-

tion dc-to-dc converters, the SiE802DFoffers exceptionally low on-resistance of1.9 milliohms maximum at a 10-V gatedrive (2.6 milliohms maximum at 4.5 V)and can handle current levels up to 60A. The SiE800DF, optimized to work asthe low-duty-cycle high-side MOSFET insynchronous dc-to-dc converterdesigns, features a very low typical gatecharge Qg of 12 nC, with on-resistanceof 7.2 milliohms maximum at 10 V and11.5 milliohms at 4.5 V.

The new PolarPAK power MOSFETs,which have the same footprint dimen-sions of the standard SO-8, dissipate 1°C/W from their top surface and 1 °C/Wfrom their bottom surface. This providesa dual heat dissipation path that givesthe devices twice the current density ofthe standard SO-8. With its improvedjunction-to-ambient thermal impedance,a PolarPAK power MOSFET can eitherhandle more power or operate with alower junction temperature. Typicalapplications for the new PolarPAKdevices will include voltage regulatormodules in notebook and desktop PCs, and other systems requiring high-efficiency dc-to-dc conversion.

MOSFETs With Double-Sided Cooling

www.vishay.com

STMicroelectronics has introduced aportfolio of application-specific analog,digital, and power switches, namedNEATSwitch, designed to address thegrowing demand for application-specificswitches that deliver higher performancewhile meeting increasingly demandingcost, space, and power constraints.

This demand is driven by the continu-ing rapid convergence of functions inmobile and digital consumer equipmentand the multiplicity of new, emergingstandards in high-speed data communi-

cation such as HDMI/DVI, 10/100/1000Gigabit Ethernet LAN, and PCI-Express.These functions require application-spe-cific signal switches built with state-of-the-art technology solutions such asvery low-capacitance transistors or copper metallization, combined withinnovations in circuit design and pack-aging technology.

The NEATSwitch products leverageST’s unsurpassed ability to combinedeep submicron geometries with rela-tively high-voltage (5V) operation, itslong-term leadership in mixed-signal cir-cuit design, and its expertise in innova-tive packaging. This unique combinationenables ST to design and fabricateswitches that offer very low on-resist-ance, low capacitance, high bandwidthand low power consumption and whichare housed in the smallest packages.

For example, analog switches in theNEATSwitch range include the world’ssmallest 16-pin SMD quad-SPDT

(single-pole, double throw) switch in a2.6 x 1.8 x 0.55mm QFN16 packageand the smallest dual-SPDT switch,housed in a 2.5 x 1.3 x 0.7mm DFN10package.

The NEATSwitch power-switch familyexploits ST’s long-standing leadership insmart power technologies, mixing ana-log and logic circuitry with integratedpower MOSFETs. These devices inte-grate two 80m-ohm N-Channel MOS-FET high-side power switches, eachcontrolled by an individual logic-enablesignal, and capable of continuous cur-rent intake of up to 500mA. Future offerings will integrate both start-upblanking features and reverse currentprotection, providing even higher levelsof performances and accuracy than arecurrently available.

Application-Specific Switches

www.st.com/neatswitch

Texas Instruments announced today a family of non-isolated, plug-in powermodules with a new technology that pro-vides ultra-fast transient response ratesand dramatically reduces customers’need for output capacitors. Featuring atight 1.5 percent output voltage regula-tion, the point-of-load modules offerincreased performance in a smaller foot-print for designers of 3G wireless infra-structure, networking and communica-tion systems that use advanced DSPs,FPGAs, ASICs and microprocessors.

Extending TI’s popular line of PTHdevices, the new T2 series of point-of-

load modules will support step-downDC/DC conversion from a wide 4.5 V to14 V input with adjustable output volt-ages down to 0.7 V at output currentsup to 50 A making them ideal for inter-mediate bus architecture

(IBA) applications. The power mod-ules offer several advanced features toreduce the overall power solution foot-print by as much as 50 percent com-pared to TI’s previous generation devices.

Faster Transient Response and LessOutput Capacitance, Advanced process-ing platforms, such as TI’s new 1-GHzTMS320TCI6482 DSP for 3G wirelessbase stations, requires extremely fasttransient responses for a system to per-form at a high-level. To achieve certaintransient targets,

thousands of microfarads of capaci-tors must be added to most power sup-ply systems. Even with higher volumet-ric efficiency, the added level of capaci-tance may expand the size of the overallpower solution to more than that of thepower supply alone—space that mayhave not been budgeted for or available.

The T2 power modules feature aninnovative new TurboTrans technology,which allows a power supply designer todynamically “tune” the modules using asingle external resistor to meet a specif-ic transient load requirement. The end-result is faster transient response with40 percent less output voltage deviationand a five to eight times reduction inoutput capacitance. In addition, systemstability is enhanced when using ultra-low equivalent series resistance (ESR)polymer tantalum or ceramic capacitors.

In addition to TurboTrans technology,the T2 modules incorporate an innova-tive SmartSync function that allows thedesigner to synchronize the switchingfrequencies of multiple T2 devices tomaximize power efficiency and minimizeelectromagnetic interference (EMI).

Point-of-Load Power Modules

www.ti.com

NEW PRODUCTS


NEW PRODUCTS

Companies in this issue

Company Page Company Page Company Page

ABB Lyon . . . . . . . . . . . . . . . . . . . . . . . . .1ABB Switzerland . . . . . . . . . . . . . . . . .1,49Actel . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8Advanced Power Technology . . . . . . .19Advanced Power Technology . . . . . . . .49Anagenesis . . . . . . . . . . . . . . . . . . . . . . .1Ansoft . . . . . . . . . . . . . . . . . . . . . . . . . .11Ansoft . . . . . . . . . . . . . . . . . . . . . . . . . . .1APEC . . . . . . . . . . . . . . . . . . . . . . . . . . .37Artesyn Technologies . . . . . . . . . . . . . . .1Bergquist . . . . . . . . . . . . . . . . . . . . . . . .14CT-Concept Technology . . . . . . . . . . .17Curamik . . . . . . . . . . . . . . . . . . . . . . . . .1Danfoss . . . . . . . . . . . . . . . . . . . . . . . .25Danfoss . . . . . . . . . . . . . . . . . . . . . . . .1,2EMV . . . . . . . . . . . . . . . . . . . . . . . . . . . .45EPCOS . . . . . . . . . . . . . . . . . . . . . . . . .31EPCOS . . . . . . . . . . . . . . . . . . . . . . . . .49

Eupec . . . . . . . . . . . . . . . . . . . . . . . . . .29Fairchild Semiconductor . . . . . . . . . .C2Fairchild Semiconductor . . . . . . . . . . . .1,4Infineon Technologies . . . . . . . . . . . . .1,44International Rectifier . . . . . . . . . . . . .C4International Rectifier . . . . . . . . . . . . .1,10Intersil . . . . . . . . . . . . . . . . . . . . . . . . . .1,4iSuppli . . . . . . . . . . . . . . . . . . . . . . . . . .12IXYS . . . . . . . . . . . . . . . . . . . . . . . . . . . .9LEM Components . . . . . . . . . . . . . . . . .3LEM Components . . . . . . . . . . . . . . . . . .1Linear Technology . . . . . . . . . . . . . . . . .5Linear Technology . . . . . . . . . . . . . . . . . .1Littelfuse . . . . . . . . . . . . . . . . . . . . . . . .28Maxwell . . . . . . . . . . . . . . . . . . . . . . . . .32Micrel . . . . . . . . . . . . . . . . . . . . . . . . . .33Microsemi . . . . . . . . . . . . . . . . . . . . . . .25Micrel . . . . . . . . . . . . . . . . . . . . . . . . . . .52

National Semiconductor . . . . . . . . . . . . .1NEC . . . . . . . . . . . . . . . . . . . . . . . . . . . .50Ohmite . . . . . . . . . . . . . . . . . . . . . . . . .15Ohmite . . . . . . . . . . . . . . . . . . . . . . . . . . .1On Semiconductor . . . . . . . . . . . . . . . . . .1Power Integrations . . . . . . . . . . . . . . . .7Power Integrations . . . . . . . . . . . . . . . . . .1Power Systems Design Franchise . .C3Semikron . . . . . . . . . . . . . . . . . . . . . . . . .1ST Microelectronics . . . . . . . . . . . . .36,51SynQor . . . . . . . . . . . . . . . . . . . . . . . . . .1Texas Instruments . . . . . . . . . . . . . . . .21Texas Instruments . . . . . . . . . . . . .1,24,51Toshiba . . . . . . . . . . . . . . . . . . . . . . . . .41Tyco Electronics . . . . . . . . . . . . . . . . . .1,6University of Brenan . . . . . . . . . . . . . . .20Vishay . . . . . . . . . . . . . . . . . . . . . . . . . .50

Please note: Bold—companies advertising in this issue

Micrel rolled out six new additions toits family of high-current power switch-es, the MIC2007/08/09 and theMIC2017/18/19. The MIC2007/08/09feature adjustable current limiting andthe MIC2017/18/19 offer both adjustablecurrent capability and Kickstart, a uniquenew feature that handles high currentsurge devices without the stalling orstuttering caused by standard currentlimiting devices. The chips are part ofthe expanding family of MIC20xx power

distribution switches introduced just lastquarter. The ICs are aimed at PCs, set-top boxes, docking stations, game con-trollers and USB host devices wherecurrent limiting as protection againstexcessive loads and short circuits isrequired. The MIC2007/08/09 and theMIC2017/18/19 are available in volumequantities and in two package options;SOT-23 and 2mm x 2mm ML. Pricingstarts at $0.69 in 1K quantity for SOT-23and $0.80 in 1K quantity for MLF packaging.

The MIC2007/08/09 andMIC2017/18/19 feature the industry’shighest power-to-footprint ratio: 0.5A/mm2.In addition, these ICs are among the firstto offer adjustable current limiting, up to2 Amps, in such small packages.

Adjustable current limiting allows thecustomer to set a limit of his or her ownchoosing rather than choosing from alist of pre-set factory limits. Adjustabilityalso helps inventory control as, withadjustable current capability, one IC can easily address the power needs ofmany designs.

The ICs all offer 100mohm maximumon-resistance and an adjustable currentlimit in the smallest form factors avail-able on the market today. These chipsare capable of serving multiple (USB)ports which provides both cost andspace savings. Standard featuresinclude a 2.5V to 5.5V operating range,output Enable control input, thermal protection, under voltage lock-out, lowquiescent current and fault masking,which prevents nuisance alarms on current surges or hot plug events.

High-Current Power Distribution Switch Family

www.micrel.com

The Power FranchiseThe Power Franchise

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

Power Systems Design Europe and China in print and online

Power Systems Design Europe and China in print and online

FREE Subscription!FREE Subscription!

Power Systems Design North America online

Power Systems Design North America online

http://www.powersystemsdesign.com

http://www.irf.com/eu

viewpoint - power systems design · 2013-10-16 · viewpoint it is toy train time ... germany,...

Documents