viewpoint - tope.com.twtope.com.tw/upload/files/lem/leaflet/solar inverter.pdf · 2 power systems...
TRANSCRIPT
CONTENTS
ViewpointNuremberg the Largest Power Electronics Show in Europe . . . . . . . . . . . . . . . . . . . . . 2
Industry NewsECPE Joint Booth with Competence Centers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Intelligent Motion at PCIM Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Artesyn Launches Chinese Web-site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Fairchild Appoints Marketing Director, Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Diehl Selects Maxwell Ultracapacitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Cree-Raytheon Team Awarded . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Better Power MOSFETs for Greener Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Learning from the Customer for Success. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Marketwatch“A Necessary Evil?”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Cover StoryEarth Leakage Control in Solar Inverters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
FeaturesDriver IC—High Noise Immunity High-Side Gate Driver IC. . . . . . . . . . . . . . . . . . . . . . 24
PCIM Show —Exhibitors at the Show . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Power Factor Correction—Power Factor Correction-on-a-Chip . . . . . . . . . . . . . . . . . 32
Power Management—Passive Current Sharing . . . . . . . . . . . . . . . . . . . . . . . . 36
Power Modules—Networking in Overdrive . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Power Management—Powering the Heart of the LCD Display . . . . . . . . . . . . 44
Transient Protection—Micro-Packaged ESD and EMISolutions . . . . . . . . . . . 50
Power Supply—Options for Secondary Post-regulation . . . . . . . . . . . . . . . . 52
New Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56-60
Member Representing
Eric Carroll ABB SwitzerlandPaul Löser Analog DevicesArnold Alderman AnagenesisTodd Hendrix Artesyn TechnologiesHeinz Rüedi CT-Concept TechnologyDr. Reinhold Bayerer eupecChris Rexer Fairchild SemiconductorDr. Leo Lorenz Infineon TechnologiesEric Lidow International RectifierDavid Bell Linear TechnologyRüdiger Bürkel LEM ComponentsPaul Greenland National SemiconductorKirk Schwiebert OhmiteChristophe Basso On SemiconductorDr. Thomas Stockmeier SemikronDr. Martin Schlecht SynQorUwe Mengelkamp Texas InstrumentsPeter Sontheimer Tyco Electronics
Power Systems Design Europe Steering Committee Members
PowerSystems DesignE U R O P EPowerSystems DesignE U R O P E
2 Power Systems Design Europe May 2005
The place to meet the power experts isNuremberg with the PCIM Europe show cele-brating it’s 26th anniversary this year from June6-9. All engineers are on their way toNuremberg to participate in one of the mostimportant events throughout Europe. Theirinterest is focused to power. Power is and hasbeen the heartbeat of any electronics.
The world economy is still diverted but powerelectronics is not a lifestyle trend, it is essentialto the support of today’s technology. The showand conference are always busy, a successattributed to all the hard working conferenceboard members. Of course, having had Gerdand Christine Zieroth driving the shows spirit fora quarter of a century has made a big differ-ence, creating an environment similar to a bigfamily event.
Becoming a tradition for the PCIM Exhibitionin Nuremberg, I will moderate each daysPodium discussions during the lunch break inthe arena on the exhibitors floor. This year Iexpanded our topics to include more systemlevel focused subjects for discussion betweenthe experts.
• Tuesday: Power Management controlledby analog or digital.
• Wednesday: Portable Power, charging andoptimizing operating time.
• Thursday: Thermal Management and thedesign and simulation tools.
The PCIM conference provides the platformfor the future view into technology. The strengthfor us is that technology can conserve our rap-idly depleting natural resources, improve livesand ensure our future. Progress is driven by innovation.
The technology for size and weight of amotor drive or of a power supply is influencedby other circuit requirements beside the powerelectronics. Power management starts at thedevice level with controller ICs that conserveenergy by putting non-required functions to sleep or by slowing clock cycles, followed by the distributed power at the board level.Additionally, system solutions are being incorporated to boost the design to higher lev-els of performance.
In the center of this issue of Power SystemsDesign Europe, you will find the exhibitors boothlayout, this should help you to get around theshow floor.
I am always looking forward to seeing myfriends at PCIM in Nuremberg, the place to meet the power experts. Stop by our booth 12-239 just across from the podiums dis-cussion arena.
Best regards,
Best regards,
Bodo ArltEditorial DirectorThe Power Systems Design Franchise
Nuremberg the largestPower ElectronicsShow in Europe
VIEWPOINT
AGS Media Group
Katzbek 17a
D-24235 Laboe, Germany
Phone: +49 4343 421790
Fax: +49 4343 421789
www.powersystemsdesign.com
Editor-in-Chief
Bodo Arlt, Dipl.-Ing.
Publishing Director
Jim Graham
Associate Publisher
Julia Stocks
Creative Direction & Production
Eyemotive
www.eyemotive.com
Circulation Management
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 4
4 Power Systems Design Europe May 2005
INDUSTRY NEWS
At PCIM Europe (7 – 9 June 2005), ECPEEuropean Center for Power Electronics e.V.organises a joint booth with leading EuropeanCompetence Centers active in PowerElectronics to present latest results fromEuropean Research. In this frame, the ECPE Demonstrator Programs will be presented as well.
The following university and research insti-tutes are partners on the ECPE joint booth:
• Aalborg University, Institute of EnergyTechnology, Power Electronic Systems
• Delft University of Technology, ElectricalPower Processing (EPP)
• ETH Zurich, Power Electronic SystemsLaboratory (PES)
• FAU Erlangen-Nuremberg, Center forPower Electronics
• Fraunhofer Institutes IISB (Erlangen-Nuremberg) and ISIT (Itzehoe)
• RWTH Aachen, Institute for PowerElectronics and Electrical Drives (ISEA)
In the forum program of the PCIM exhibi-tion, ECPE will give a presentation on itsPower Electronics Online Course (8 June2005, 12.00 h) which is built up in cooperationwith the Swiss Federal Institute of Technology
(ETH) Zurich, Power Electronic SystemsLaboratory. The course comprises interactiveand animated Java applets as well as a scripton fundamentals and theory of power electronics.
The Online Course is available on theECPE web site www.ecpe.org, the use is freefor ECPE Members as well as for universitiesand students.
You find the ECPE joint booth in hall 12,booth 551.
ECPE Joint Booth with Competence Centers
Intelligent Motion at PCIM Europe
IPCIM Conference Directors for IntelligentMotion, Prof. Helmut Knöll, FachhochschuleWürzburg-Schweinfurt, Germany and Prof.Mario Pacas, University Siegen, Germanyhave their impressions for Nuremberg sum-marized. In Intelligent Motion we see a
steady progress in all fieldsof drive technology. Only topoint out some trends andhighlights of our conference:
The trend of this year inmotors and actuators arefault-tolerant systems,direct drives and improveddesign methods.
Concerning variablespeed drives we concen-trated on low power drivesas they are today veryimportant in automotive andin white goods applications.Vorwerk, Germany presents
a high speed switched reluctance motor drivefor kitchen machines. EBM-Papst, Germanyreports on high speed PM drives for automo-tive applications.
Mechatronics continues to be one of themain driving forces for novel and improveddrive technology. We observe an increase in
both, applications as well as new solutions inmachine building. Siemens, Germany reportson improved control of elastically coupledmachine tool axis as well as compensation ofstructural oscillations of large gantry machines.ELAU, Germany presents flexible mechatron-ic machines in the packaging industry.
Furthermore, sensorless drives still repre-sent a challenge to our engineers. We invitedProf. Holtz, Germany to report on perspec-tives and concrete solutions of sensorless acdrive technology. International Rectifier,Canada speaks on a novel position sensor-less PMSM drive.
In the topics covering components for sys-tem integration as well as control and meas-urement you will find applications dedicatedto all of our specialities.
www.ecpe.org
Artesyn Launches Chinese Web-site
www.pcim.de
In a development that reflects the growingimportance of the domestic power supplymarket in China, Artesyn Technologies haslaunched a Chinese version of its web site.Most of the navigation and search tools onthe new site - http://www.artesyn.com/china/ -use Chinese nomenclature to help new andexisting customers in the region find whatthey're looking for very quickly and easily.The site presents all top level information inChinese, and effectively mirrors Artesyn'smain English-language web site for detailedproduct information. The site also lists com-pany job vacancies in China: Artesyn has anumber of facilities in the country, including
sales and support offices in Beijing andShanghai, Asia-Pacific headquarters in HongKong, and a large manufacturing and R&Dcentre in Zhongshan City. The Chinese mar-ket for power conversion products has out-performed most other regions of the worldover the past few years, and is currentlygrowing at a rate in excess of 10%. It is beingdriven by a combination of increased domes-tic demand for electronics goods and servic-es, especially in the telecoms sector, and bycompanies continuing to transfer electronicsmanufacturing from other parts of the world.Our new Chinese web site provides cus-tomers in the region with fast access to a
fully-maintained information resource. Thesite is likely to prove extremely popular withdesign engineers, who need to be confidentthat they are referencing the most up-to-datedata when specifying AC/DC power suppliesand DC/DC converters for their products.
In order to support and further enhancegrowth in the Asia-Pacific, Artesyn is develop-ing dedicated local language sites for othermajor countries within the region. These sites will be brought on-line during the courseof 2005.
www.artesyn.com/china/
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
6 Power Systems Design Europe May 2005
INDUSTRY NEWS
Power Events
• PCIM Europe 2005, June 7-9, Nuremberg,
www.pcim.de
• Automation Seminare, June, Germany,
www.mesago.de/automationseminare
• GEMV Energietechnisches Forum,
June 14-15, Kiel, www.gemv.de
• EPE 2005, September 11-14, Dresden,
www.epe2005.com
• H2Expo 2005, Aug. 31 - Sep. 1, Hamburg,
www.h2expo.de
• INTELEC2005, September 18-22, Berlin,
www.intelec.org
• SPS/IPC/DRIVES, 22-24 November,
Nuremberg, www.mesago.de/sps
Cree announced that a team led byRaytheon, as the prime contractor, and Creeas a subcontractor, has won a program awardfrom the Defense Advanced ResearchProjects Agency (DARPA), under its WideBandgap Semiconductor TechnologyInitiative. The team will be developing nextgeneration gallium nitride (GaN) semiconduc-tors for military and commercial products. The program is focused on acceleratingprogress and targeting the insertion of GaNinto military and commercial programs starting in 2006. The Cree-Raytheon teamwas formed to combine the efforts of two ofthe premier GaN RF teams in the UnitedStates to allow even faster developmenttowards system needs.
John Palmour, Cree’s Executive VicePresident, Advanced Devices, stated, “Thisprogram is perfectly aligned with Cree’s strat-egy to be a supplier not only to the militarybut also to a wide variety of commercial appli-cations. The same improvements that willmake GaN components viable for military
systems insertion we believe will also makethem viable for insertion into cellular infra-structure, as well as other burgeoning wire-less applications. Combining our efforts withRaytheon is intended to accelerate the poten-tial deployment of this important enablingtechnology.” Cree has been developing GaN-on-silicon carbide (SiC) RF devices since1996. The semiconductor work for this pro-gram will be conducted at Raytheon RF com-ponents (Raytheon’s MMIC foundry) inAndover, Massachusetts and in the WideBandgap MMIC foundry of Cree, Inc. inDurham, North Carolina, as well as in Cree’s Santa Barbara Technology Center inGoleta, California.
Cree-Raytheon Team Awarded
www.cree.com
FairchildSemiconductorannounced that it hasrecruited AlfredHesener to the newly-created position ofMarketing Director,Europe. He will reportto Ole-Petter Brusdal,Vice President of Salesand Marketing for
Europe, and will be based in Fuerstenfeldbruck,Germany.
Hesener joins Fairchild from Infineon, where
he was Senior Manager Technical Marketingfor power semiconductors and control ICs forpower conversion.
“Hesener brings to Fairchild a combinationof a strong background in worldwide productmarketing and definition and experience of theEuropean market and customer base,” com-mented Brusdal. “To have these two experi-ence sets combined in one individual is rare,and Hesener will add considerable strength toour expanding presence in the European mar-ket. With a focus on automotive, industrial,consumer and communications segments, ourEuropean organisation provides strong design
expertise and customer support, from designto delivery.”
Alfred Hesener spent a total of over 9 yearsat Infineon and Siemens AG, SemiconductorDivision in Technical Marketing and R&D. Hestarted his career with Atmel, where he was aField Applications Engineer. Hesener has adegree in Electrical Engineering fromDarmstadt Technical University, with a speciali-sation in Microelectronics especially analogueand digital design, semiconductor physics andprocess technology.
www.maxwell.com
www.diehl-gruppe.de
Fairchild Appoints Marketing Director, Europe
Diehl Selects Maxwell UltracapacitorsMaxwell announced today that Diehl
Luftfahrt Elektronik GmbH, a leading producerof aerospace systems and controls, has selected Maxwell’s BOOSTCAP ultracapaci-tors to power
emergency actuation systems for doors andevacuation slides in passenger aircraft, includ-ing the new Airbus 380 jumbo jet. These sys-tems are designed to provide fail-safe backuppower to open aircraft doors and deploy evacu-
ation slides in an emergency, so Diehl thoroughly evaluated and tested all availableenergy storage alternatives before settling onultracapacitors. Each of the 16 Door and
Slide Management System (DSMS) actua-tors on the Airbus 380 incorporates ultracapaci-tor cells in a redundant configuration to ensurereliable operation even if individual cells shouldbe damaged or fail for any reason.
Ultracapacitor-based systems are lighter
than batteries and are more reliable becausethey remain fully charged off the main powersystem during normal operation, and they storeup to eight hours of standby power to operate theactuators in the event of a main system failure.
www.fairchildsemi.com
8 Power Systems Design Europe May 2005
Better Power MOSFETsfor Greener Power
When examining power circuits,there is the need for increasedefficiency due to better topolo-
gies, better magnetics and other passivecomponents, superior control ICs andswitching regulators as well as betterdiscrete semiconductors. The majorityof the power losses in power circuitsoccur in discrete power semiconductorsand magnetics.
Fairchild recently introducedSuperFETTM MOSFETs, such as theFCP7N60. These new high voltageMOSFETs specifically address demand-ing high voltage fast switching applica-tions requiring high efficiency such asActive Power Factor Correction (PFC),lighting and AC/DC power supplies.These devices allow applications toachieve a very impressive RDS(ON)and small die size resulting in increasedefficiency in applications (Figure 1).
As you can see above, the orange bar represents the RDS(ON) of theSuperFET and the green bars representconventional technology MOSFETs.
Fairchild’s SuperFET technology(FCxxNxx) creates more efficient switch-es by changing the relationship betweenthe RDS(ON) of the MOSFET and itsbreakdown voltage from an exponentialrelationship to a linear one. This solutionis achieved through Fairchild’s SuperJunction FET process technology thatgrows pillars of P-type material (seeFigure 2). The high break down voltage(600V) and industry leading Figure ofMerit (FOM) parameters such as RDS(ON)
per unit silicon area can significantlyimprove the efficiency and reliability ofend applications like AC/DC power sup-plies, motor drives and lamp ballasts. As an example, a common 30W flybackAC/DC Switching Mode Power Supply
(SMPS) with 110VAC input was testedin our lab. A maximum duty cycle of 0.77was measured and at the system effi-ciency of 75%, the peak drain currentwas 1.22A. At this current, the differencein consumed power based on theRDS(ON) of the SuperFET and standardMOSFET was 0.97W. This change inwasted energy directly correlates to anincrease in efficiency and a reduction ofheat dissipation that could reduce theheat sink size, reducing weight and cost.The efficiency gained may seem to besmall but the advantages of more effi-cient power supplies everywhere willsave money in different ways. One hasto only look at typical office building tosee that it is populated with variouselectronic office equipment. Each ofthese has an AC/DC power supply and ifevery device dissipates 1 Watt less, thatis one less Watt that the Air-conditioningsystem needs to remove.
As can be seen from Figure 1, theconventional MOSFETs have evolvedinto more rugged, faster switching, easier to drive and less expensive
devices over time. The high voltageMOSFETs of today, with a breakdownvoltage of 500-600V or higher have over85% of the total RDS(ON) coming from the lightly doped epitaxial (Repi)region—a component that cannot beeasily decreased since the epitaxialregion of the mosfet is needed to blockthe high voltage in the off state.Knowing this, it is evident that theadvances in conventional MOSFETtechnology are impressive and the con-ventional MOSFETs of today can com-pete in critical areas such as the specificRDS(ON) [ Rds(on) per unit area] withsuper junction MOSFETs of three yearsago at a very competitive price, due toits simpler structure.
By far, the most impressive newpower MOSFET technology introducedin the last decade has been the SuperJunction MOSFETs like Fairchild’sSuperFETs. This modern family ofMOSFETs has a structure distinctlydifferent from that of the conventionalMOSFETs. Figure 2 compares a cross-section of the conventional MOSFET
Fig. 1. Decrease of RDS(ON) of 600V MOSFETs Over Time.
10 Power Systems Design Europe May 2005
with that of the SuperFET. Like all dis-crete MOSFETs, the SuperFET carriescurrent vertically through the bulk of the silicon (see Figure 2).
As you can see, the most dramaticdifference between the cross-section of the SuperFET and the conventionalMOSFET is the deep P-type pillar in thebody of the MOSFET. The effect of thedeep P-type pillars is to confine theelectric field in the lightly dopedEpitaxial region of the mosfet.
To study the advantage of the con-fined electric fields of the SuperFET, letus look at the effect of the rise of the on-resistance of the MOSFETs as thebreakdown voltage of the MOSFETs isincreased. In standard MOSFETs, theRDS(ON) is directly proportional to the diesize but the RDS(ON) increases exponen-tially as the breakdown voltage increas-es. In fact, as the breakdown voltageincreases, the RDS(ON) increases with anexponent of roughly 2.5. The 1000VMOSFET, therefore, has a resistance offive times that of a 500V MOSFET ofequivalent die size. In other words, a1000V MOSFET should have roughlyfive times the die area as a 500V MOS-FET in order to have the same RDS(ON).This means that higher voltageMOSFETs have very large die sizes thatincreases their cost and package sizeas well as decreasing manufacturingyield, etc. Hence, this presents a seri-ous design challenge. The SuperFETreplaces this exponential relationshipbetween voltage and RDS(ON) with a lin-ear one—that is the 1000V SuperFETshall only be about twice the area of a500V SuperFET with the same RDS(ON).
The reduced specific on-resistance ofthe SuperFET, however, comes at theprice of greater design complexity com-pared to the conventional MOSFET. The idea of the Super Junction MOS-FET has been well known to designersfor several decades but it is only recent-ly that the costs of processing the deepP-type implants has decreased to makethe SuperFET a viable option. Presently,the manufacture of the SuperFETsinvolves several additional epitaxial layers that comprise the P-type pillar.
This involves many masks approximate-ly doubling the number of masks used in a 600V SuperFET compared to theconventional MOSFET. The number ofmasks and therefore, the complexity,manufacturing challenges and price ofthe SuperFET goes up as the break-down voltage is increased. These pro-cessing challenges make the manufac-ture of say, a 1000V SuperFET muchmore difficult than the 600V SuperFET,even though the 1000V SuperFET hasdecidedly greater advantage in RDS(ON)
over its conventional counterpart.
This design constraint can make thedie size larger than the package at thesame RDS(ON) and increases die cost.The package and cost constraints forcestandard MOSFET manufacturers toincrease the RDS(ON) to produce a prod-uct at high breakdown voltages.
The real advantage of the SuperJunction MOSFETs is the way they have increased the efficiency of powercircuits not only in the high power appli-cations, but the smaller die sizes ofthese Super Junction MOSFETs hasresulted in smaller packages such asthe DPAK being used in many lowerpower, higher volume applications suchas lamp ballasts. The 600V, 600mWmosfet such as the FCP7N60 for exam-ple has become very popular in variousapplications replacing conventional600V, 1.2 ohm TO-220 MOSFETs.
The manufacture of the SuperFETtakes advantage of the various tech-niques developed over time to make theMOSFETs more rugged, better suitedfor bridge-type high power applications.As the power supply industry makesmore area-efficient MOSFETs, the siliconarea needed for the MOSFETs reduces.It is increasingly important for thesemodern mosfets to have avalancheenergy and anti-parallel diode dV/dt rat-ings per unit silicon area to be higherthan those of previous generations.
The SuperFET is also designed tohandle a high rate of change of current(dI/dt) during switching. This rugged-ness is important in many transientprone applications such as HID ballastsand plasma televisions. A guaranteedmaximum gate source voltage of +/-30Vhelps the device withstand drive tran-sients. Careful design of the cell struc-ture also results in this newly releasedSuperFET family having a repetitiveavalanche energy rating significantlyhigher than competitive solutions. The SuperFET manufacturing processalso lends itself well to fast body diodeversions that are optimized for use inbridge circuits.
www.fairchildsemi.com
Fig. 2. Comparison.
12 Power Systems Design Europe May 2005
Today’s market for Power Modulesis put at about US$ 800 millionworldwide. High quantities are a
characteristic of the lower power catego-ry, while in the medium to high powerrange you find greater focus on valueadded and high quality. Well-knownmarket researchers expect to see con-tinuing substantial growth for bothranges in the next ten years, anassumption supported by the generaltrend and the need to economize onenergy. This is where producers ofPower Modules make an extremelyimportant contribution, though they pursue very different strategies.
A share of the innovative drive natu-rally comes from the R&D labs of thesemiconductor producers. We haveseen ground-breaking technologies likenon-punch through (NPT) and trenchfield stop result in regular revamping ifnot complete redesign of Power Modulegenerations from all suppliers. Voltage-controlled power MOSFETs, more ofwhich are now being developed for highvoltages too, are finding their way intonew applications that call for very highswitching frequencies for example.Zero-voltage-switching (ZVS) circuitscan now be implemented in integratedPower Modules in a similar way tothyristor/diode combinations for very different applications.
In all cases the customer drives thedevelopment, and enables in particularthe manufacturers of Power Modules, ineffect mechatronic subsystems, to bettermeet the requirements. As a conse-quence we saw the appearance of twostrong trends in the low-power drivetechnology segment between the middleand end of the 1990s. On the one handthere are cost-efficient solder moduleslike flowPIM modules, characterized by
unique power flow-through. Then thereare press-fit modules like MiniSKiiPmodules, enabling the customer to dis-pense entirely with soldering, and thus,on the European market at least, creat-ing cost advantages. CIB (converterinverter brake) topologies rule the roostin the industrial drive technology sector.Sixpacks round off the palette. Suppliershave an ongoing mandate to presentmodules that will cut the cost and BoMof customer’s application design andspeed up their production further. Oneapproach is Power Modules whosedirection of insertion is identical to thatof other through hole components likerelays or DC link capacitors. The bestway is still to learn from the customer.The new product families of the flow90°PIM series from Tyco Electronics makea decisive contribution. Here too the factwas considered that the case not onlyfunctions as a shock-hazard guard but,through its molding, enables a definitemechanical attachment of the circuitboard to the heatsink, as a cost-freeextra as it were.
The customer expects even more –complete power range from one source.Seeing as the majority of customers do
not concentrate solely on designing andmarketing low-power devices, the sup-plier of Power Modules must respond byconstantly expanding and improving hisproduct portfolio. That also applies toTyco Electronics of course. A major con-tribution was last year’s launch offlowPHASE 0 modules, unique on themarket in terms of power density andpower cycle capability. The next move isflowPHASE 2 modules, extending theoffered power spectrum up to an IGBTcurrent of 300 A for 80°C heatsink tem-perature and 1200 V reverse voltage.This is a further important step towardshigh power, and certainly not the last.The high-power technologies needed forthe purpose were already developed awhile back for application-specific mod-ules, and their reliability has been quali-fied and tested in the field.
The platforms on which these technolo-gies set up allow the development ofPower Module solutions tailored to spe-cific applications. The resulting productsare destined for applications worldwidein industrial drive technology, pump andfan controls, in welding and for soft starters.
Summarizing it can be said that thechallenges of the market are exception-al, but its growth curves are too.Customers can only be offered the rightsolutions for their markets if one listenscarefully to what their needs are.Generating standard products from custom solutions means letting all cus-tomers benefit from the latest chip andmodule technologies, true to the mottoyour success is our success.
Green products that meet RoHSdirectives will be qualified and availablefrom Tyco from mid-2005.
http://em.tycoelectronics.com
By Peter Sontheimer, Tyco Electronics
14 Power Systems Design Europe May 2005
By Chris Ambarian, iSuppli Corporation
The 20th annual Applied PowerElectronics Conference andExhibition (APEC) in Austin this
year might well be remembered as theturning point for market recognition ofthe importance of power electronics.
I recently had the occasion to behanging out with a couple thousandpower electronics industry people inTexas, when I had an almost out-of-body observation of us as a group, andthe observation has some pretty signifi-cant implications for us. So if you’llindulge me, I’ll share it.
I was sitting in one of several techni-cal sessions, when I realized that I’dheard something about 3 times at theconference already: “The power supplyis a necessary evil.” It was a powerengineer, casually parroting the atti-tude(s) of his customer base—attitudesthat fundamentally drove his activities. I thought about that. “A necessary evil.”Hmm. It sounded like things I’ve heardfor years in the consumer part of themotor drive market as well. “A neces-sary evil…”
How did we get to this point, a placewhere what we’re working on is “a nec-essary evil” ? And by the way, whowants to work on a necessary evil???How exciting is that?
I have developed products for and sold into this market for about twodecades now, so I can certainly under-stand where this attitude was devel-oped. We developed it in speaking withcustomers who knew little about power,and for whom it was not a critical func-tion. In other words, they didn’t valuepower. They told us that it was a neces-sary evil, and in the absence of anycompelling evidence to the contrary, we believed it. Heck, it may have evenbeen true.
Now mind you, I’m a realist. I’m notgoing to tell you that the emperor hassome great clothes, and ask you to justbelieve. It probably hurts the brain justto imagine going to your customer(s)and saying anything other than “okay,here’s what we’re doing to drive cost outof our function.” But in the same breath,I’d like to ask a question: How valuablecan “a necessary evil” ever be?
Business is all about knowing whatyour value is, and then getting paid thatvalue. You create value and articulateyour value to your customers based onthis self-image of what your value is –it’s the basis of what we direct our R&Dengineers to create, it’s what drives ourmarketing and business managementpeople, it’s all that becomes available astools and messages to drive our salespeople. If you accept the notion that youare nothing more than “a necessaryevil,” what kind of conversation do youthink you’re going to be able to havewith potential customers?
Ironically, an equipment systemdesigner’s freedom to design is nowbounded by (or better, enabled by) thekind of power systems available to himor her. This is truer today than it hasbeen for the last 20 years that I know of.One significant change is that power is
now a critical function, a rate-determin-ing step (or at least a cost- and space-determining one). Another fundamentalchange in the landscape is the set ofpossibilities created by the convergenceof power, data processing, and commu-nication. The competitiveness and differ-entiation of the end system are tightlyinterwoven with the power system. Youcan’t easily design a smart network ofdrives to run your factory in a revolution-ary way if your drives can’t be communi-cated with easily. The more work thatwe do to facilitate new possibilities forthe designer, the more they are going tobe able to value our product. Of course,this requires that we seek out designersat a level above the “same thing for 10%less than last year” level.
I would propose that it would be muchmore powerful for us as suppliers (andfor our customers as well) if we were tostop blindly accepting the “necessaryevil” image as our own, and to adopt anew self-image: “Power supply asenabler.” “Motor drive as enabler.” “Byusing this product or this approach, youcan make a much more valuable endproduct.” This is a much broader per-spective, where “value” is a subject tobe explored between you and the cus-tomer, not just dictated to us as being"lower cost" (which is only one side ofwhat’s potentially valuable). Sure, “morevaluable end product” could simply beone that costs 10% less next year – butit could also be a product that doesthings not possible before.
In the end, this is all just my conversa-tion to create a new possibility for howour portion of the industry could work,and how we might become much moreprofitable. You don’t have to accept it.You can continue to do business in theexisting paradigm. At least we knowwhat that’s like.
MARKETWATCH
www.iSuppli.com
16 Power Systems Design Europe May 2005
COVER STORY
Evolution of photovoltaic cell productionKyoto protocol committed different national governments to increase the production of
green energy and to make a grant to every initiative for this promotion.
By Stéphane Rollier, Bernard Richard, Martin Keller, Lem
In collaboration with Hans Welschen, Philips Lighting
Governments have decided tosupport every effort oriented for energy saving, and for
sustainable energy. This can explain the revived interest in solar energy.
Solar energy had already been asource of interest some years ago dueto profitability analysis. Indeed, theinvestments incurred were significantlyhigher, as compared to the usable energy generated. That was until thisrenewed commitment and today’s elec-tronic technologies, which made thisenergy source a more profitable one.
For instance, in Europe, 410.5 solarMWp have been installed during 2004representing 69.2 % of growth vs. 2003.Photovoltaic market is led by Germanyfollowed by Japan and USA. Spain con-firms nevertheless, the increase inimportance of its photovoltaic market(11.8 MWp installed in 2004 vs. 6.5MWp in 2003).
Confirming this general trend is theevolution of the worldwide photovoltaiccell production. It has been produced1194 MWp last year (representing about400,000 systems with a mean capacityof 3 kWp), that is to say 450 MWp more
than in 2003: A growth rate of more than60%! More than half of the photovoltaiccell production has been realised inJapan, the remaining coming mainlyfrom Europe (26%), USA (12 %) andrest of the world (12%) in 2004.
A solar panel is comprised of an arrayof solar cells, connected together inseries and parallel. These cells are thenencapsulated in glass and plastic. Toallow for mounting on rooftops, usuallythe panels themselves are enclosed inaluminium or steel framing (Figure 1 and 2).
Figure 1. Cells. Figure 2. Solar Panels.
18 Power Systems Design Europe May 2005 www.powersystemsdesign.com 19
COVER STORY
A variety of cells are available today,and in general today’s panels generateDC current between 7 and 7.5 A. Othermodels also exist (Thin-film solar mod-ules for example) generating differentcurrent levels. The maximum solar cellpower is defined by the working point ofthe cells, which correspond to a certainvoltage (Vm) and current (Im) (Figure 3).
When cells are short-circuited, a constant current is provided (with avalue depending of the light intensity). A voltage expressed as Voc, approxi-mately 0.6 Volt, is present when cellsare open circuited. The total voltage Vocin a full solar panel is dependent on thenumber of cells used in the panel. Oftenused numbers are 36, 54 or 72 cells,resulting in 22, 33 or 44 Volts Voc.
Several panels can then be connectedin series and/or parallel until reachingthe required power and/or the maximumallowed voltage. A Voc of <120 V (atstandard conditions +25°C) is consid-ered as safe to touch.
During normal operation a specialcontrol routine keeps the solar panel inits maximum power operating point. Apanel of e.g. 36 cells will then produce avoltage of approximately 14 -18 Volt,depending on the temperature.
To optimise the power delivered, it iscommon to use special software anddedicated electronics to control theoperating point of the cells.
This generated electricity can then beused in 2 ways: For autonomous instal-lations, remote from the distribution grid,charging a battery. This is commonlyreferred to as the off-Grid system (30 %of the market in 2002) For feeding backinto the grid as green electricity. This“Grid connected” system will be describedhereafter (70 % of the market in 2002).
The connection of the solar arraythrough the inverter to the grid can berealized either by using a transformer or directly without transformer, so calledtransformer-less, meaning without gal-vanic isolation (Figure 4).
The transformer can be a convention-al 50/60 Hz type between the inverterand the grid or a high frequency trans-former as part of the DC part of theinverter. Transformer-less designs havebecome a general trend in such applica-tions to enhance overall efficiency andlower cost.
High frequency transformers have theadvantage of lower weight and a smallervolume. Compact and lightweight designsplay an equally important role. There isalso a question of power. Higher gener-ated power requires large transformers,leading to larger installations and highercosts, and this explains the interest intransformer-less configuration.
Figure 3.Cell I-U characteristics/Solar cell power.
Figure 4. Solar panel system.
20 Power Systems Design Europe May 2005
Whatever the system used, be it withtransformer or transformer-less, themain concern is to ensure the safety ofthe entire system, and more importantlythe safety of any human contact with theentire system.
As a starting point, the metal frame ofthe solar panel can be grounded. Forlarger systems that is mandatory fromthe building installation code but alsouseful in mountainous areas due to thelikelihood of lightning strikes(Switzerland, Austria for example).
In the US, connection to ground forthe DC part of the system is mandatoryand when an electrical fault occurs, thisconnection must be interrupted as wellas the entire installation disconnectedfrom the grid. There is a chance that theground connection will not be mandatoryafter the next installation code release.
In Europe it is free to connect or notthe DC system to ground, however, witha transformer-less inverter the DC sys-tem will be connected to ground (vianeutral of the grid) through the inverterelectronics. The array may not haveanother connection to ground to avoidDC ground currents.
In Germany and some other coun-tries, the insulation from ground must betested before the inverter may connectto the grid and start operation. By hav-ing the solar panel DC voltage floating inrespect to ground, a first layer of safetyis insured; touching at a single point willnot immediately create a hazard.
A system is defined as floating when
its resistance to ground is > 1kOhm/Voltwith a minimum of 500 kOhm (Figure 5).
Although the Photovoltaic (PV) arraycan be connected as floating, the wholesystem will be floating depending on themaximum possible voltage of the PVarray (for a resistance to ground already installed).
Measuring a possible leakage toground is not possible without closingthe current path e.g. by means of aresistor. This is true for both, inverterswith transformer and transformer-lessinverters because the latter has to bedisconnected from the grid (by means ofrelay contacts) during the measurement.
Building installation code requires aRCD (Residual Current Device) type Bfor transformer-less PV-systems. TypeB, the version that is also sensitive toDC currents, is necessary because aground fault in the PV array may generatea DC current. Disadvantage of suchRCD is the high sensitivity for disturbingpulses and the fact that they have to bereset manually. Incorporating this functionin the inverter will have some advantages.
1) The function can be combined withthe required array insulation measure-ment. 2) The sensitive measurement ofsmall DC currents can be done beforethe inverter starts operating and highfrequency switching signals could dis-turb the measurement. 3) The functioncan be made with automatic reset aftera nuisance trip. 4) Last but not least, theac+dc ground current protection levelcan be set for person safety (30 mA)
while accepting a larger stationary acground current due to the capacitancebetween the solar cells and a groundplane in the vicinity. This current isallowed to be as large as 300 mA. Asudden change of 30 mA can lead todisconnection. A differential currentmeasurement can be used for this function.
Another requirement for the grid con-nection is not to supply DC current intothe grid. The accepted level differs fromcountry to country but often seenrequirements are 0,5 % or 1 % of thenominal current output. Thusly, the DCoffset of the current transducer used inthe control loop of the inverter should beas low as possible. Also, DC offset as aresult of the IGBT’s switching delay inthe inverter is then to be avoided or assmall as possible.
The consequence of this DC offsetcan be a saturation of the network distri-bution transformers. To decrease thisDC offset, new topologies of invertersare ongoing. Also, the Total HarmonicDistortion (THD) of the output currenthas to be limited to a value defined bythe different utilities. This is differentaccording to the country concerned asno real value has yet been agreed on.
When these problems occur, a circuitbreaker is usually used to disconnectthe solar installation from the grid.
LEM’s CT range (Figure 6) has beenspecifically designed to meet those criteria in modern solar topologies, inresponse to increasing market demandin providing a compact, low cost andreliable solution based on a currenttransducer.
COVER STORY
Figure 5. Photovoltaic (PV) array floating with transformer configuration.
Figure 6. LEM CT series of differentialcurrent transducers.
www.powersystemsdesign.com 23
Traditionally used devices (such asRCD) are commonly known devices, butare rather bulky, and not really adaptedto the new requirements in solar invert-ers. They also might fail, consideringthat both DC and AC currents must bemonitored with spectral components up to 30 kHz induced by high speedIGBT switching.
Impedance measurement can also bea way to check the insulation level andto detect an earth fault in the solarpanel. To achieve this, three measure-ments are mandatory; Impedancemeasurement, resistance ratio change + impedance measurement (to detectsymmetrical fault to the earth in thesolar panel), and voltage measurement.But this is not easy to do.
For a PV floating, and linked to thegrid through a transformer, specificrequirements do not yet exist. However,
the impedance measurement betweenthe PV array and the ground beforestart-up could be the tool to prove thereal floating point (requirement is for 1 kOhm/V with a minimum of 500kOhm/V). To measure this value, a volt-age or current transducer can be used.
For a PV connected to earth, andlinked to the grid through a transformer,impedance measurement and/or differ-ential current measurement are possibleto prove the ground connection.
The LEM CT series differential currenttransducers measure safely, currentswith nominal values of 100mA, 200mAand 400mA, providing a linear voltageoutput of 5V at nominal current.Response time is better than 20ms at80% and 60ms at 90% of peak current.
The use of a high technology (“flux-gate”) has been the solution to these
requirements, especially to achieve theaccurate measurement of very small DCor AC currents.
DC and AC components up to 30 kHzcan be measured. The CT products arePCB mountable, small size, lightweightcomponents with an aperture for inser-tion of earth leakage wires.
Generally, the CT range is also suit-able in other applications, includingmedium power inverter applications.Supported by international agreementsto reduce the volume of carbon dioxideproduced by fossil fuels, and by thegrants allocated by the various govern-ments, statistics forecast some 4500MWp which could be produced by solarenergy by 2010 in Europe. This assureselectrical measurement requirementswill be practiced more and more toensure quality and safety.
COVER STORY
www.lem.com
24 Power Systems Design Europe May 2005
The monolithic high-side drive poses a fascinating challenge for system designers. Their goal is
to find a way to drive the switching devices that is as simple as possible—without suffering from
lack of speed or, wasting large device area—while maintaining low-voltage control capability.
And they must do this using only the outputs of a micro controller or any other control
circuits working in the 3.3V or 5V supply.
By Jong-Tae Hwang, Senior Engineering Manager, Fairchild Semiconductor
The HVIC can replace pulse transformer
This article introduces a high-volt-age process for the high-side gatedriver IC, the conventional high-side
driver topology, and addresses the dri-ver’s problems. Also, it will briefly explaina new proposed circuit technique thatcan solve common challenges faced bydesigners by the conventional high-sidegate driver IC Such as critical malfunc-tion of the HVIC. In addition to explain-ing how to solve such a malfunction, therobustness issue will be exploredthrough the design and some experi-mental results will be presented.
High Voltage Junction TerminatedProcess
It is difficult, and not area-effective tothe design, to make every device in thehigh-side gate driver circuit endure highvoltage up to 600V. Actually, only fewtens of volts are required to drive thegate of the power MOSFET and IGBT.Thus, the high-side driver itself does not require a full high-voltage rangeoperation. The offset voltage of the high-side driver, however, must be guaran-teed up to the maximum allowable volt-age. To satisfy this high offset voltage
constraint, the high-side circuit is incor-porated in the cathode area of the high-voltage diode.
The high voltage diode of a devicesuch as Fairchild’s HDG4 625V ratehigh-voltage process is comprised by P-substrate, which has high resistivity,N-type buried layer and low doped N-type epitaxial layer. These layers aredesigned to meet the 625V break-downvoltage requirement. To horizontally isolate this diode from other devices,some isolation layers are needed. Since the cathode of the diode is madeof N-epitaxial layer, P-type isolation isrequired. Unfortunately, P-type isolation(PISO) layer cannot guarantee 625Visolation rate.
The HDG4 process uses additionalhigh-voltage isolation (HVI) structure,which is based on the RESURF tech-nique and N-epitaxial layer. Owing tothe HVI structure, the PISO layerbetween high-side circuit region (thecathode of the high voltage diode) andthe HVI structure is fully depleted and it is possible to prohibit a high electricfield from forming in the PSIO layer.
Of course, it is not an easy job toaccomplish the high-voltage junction termination. This phenomenon dependson the N-buried layer concentration, onits location in the HVI, and on the depthand the resistivity of N-epitaxial layer.
Once this design difficulty is over-come, the process gives a good area-effective LDMOS. The HVI structure inthis example can be the drain of theLDMOS, and the MOS channel can beimplemented in the PISO layer. Sincethe drain of the LDMOS has the samestructure with the HVI, LDMOS can beimplemented without need for extraarea. This is a dramatic leap of the mod-ern high-side gate driver IC processsince the conventional process requiresan additional area for the LDMOS.
High-Side Driver ConfigurationFigure 1 shows a conventional high-
side gate driver circuit configuration.The conventional high-side gate driverIC typically is composed of following six parts:
Input detection logicEdge pulse generator
DRIVER IC
www.powersystemsdesign.com 27
In the case (B), the allowable negativeVS voltage is –VTH since the HVICdoes not respond below this level. Thismalfunction can be frequently observedin the real application since there are somany chances for the HVIC to driveinductive loads. Figure 3(A) is very
familiar half-bridge circuit using HVIC(where FAN7382 is a HVIC, which has alow-side driving circuit as well).
This circuit drives an inductive load, L,and also has a current-sensing resistor.The switching inputs are shown inFigure 3(B). When HIN is high, a high-
side external power MOSFET, whosedrain is connected to 600V line, isturned on and supplies current to theinductor load. Even if LIN is high, andthe low-side MOS is turned on sometimes later, the current flows fromground level to the load. Accordingly,output voltage (VS) becomes negative.Furthermore, the sensing resistor makesthe output into a deep negative voltage.Following this, HIN becomes high again.What is wanted from the HVIC is tomake the output become high. TheHVIC, however, misses the command ifthe allowable negative VS voltage ishigher than –V, as depicted in the figure.
Proposed HVIC CircuitToday there are two goals to achieve
with the HVIC: (1) to enhance the noiseimmunity for the high dv/dt noise, and(2) to extend allowable negative VS volt-age. To achieve these goals, a newreshaper circuit is proposed (as shownFigure 4.)As explained previously, theallowable negative VS limitation comesfrom the predetermined threshold levelof the reshaper. It is needless to say thatit is possible to increase that limitationby increasing VTH level.
Such design, however, makes HVICtoo sensitive to the noise.The output ofthe level shifter swings from VB toground direction, whereas the conven-tional reshaper is operating in the VBSsupply voltage. Accordingly, the detec-tion of the level shifter’s output by thereshaper is determined by the VB volt-age level and VTH of the reshaper.Inthis new approach, a V/I (Voltage-to-Current) converter is used to detouraround the VTH problem. The V/I con-verter converts the level shifter’s outputsto the current information. Since the onlyrole of this circuit is conversion, VTH isnot considered. Next, an I/V converterreconstructs the voltage signal, whichswings between VB and VS supply railsfrom the current outputs of the V/I con-verter. After these sequences, the ampli-fied signal is sent to the S-R latch.Owing to the V/I converter, the allowablenegative VS voltage is no longer gov-erned by the threshold level, VTH, of thereshaper. As long as VB voltages arelarge enough to meet the requirement
Figure 3. Allowable negative VS problem at Inductive Load Application: (A) A typical half-bridge configuration, and (B) Command missing due to the lack of allowable negative VS.
26 Power Systems Design Europe May 2005
Level shifterReshaperS-R latchGate driver
Before transferring the input signal tothe high-side drive circuit, through thelevel shifter circuit by high-voltageLDMOSs with R1 and R2, the inputdetection circuit determines the inputsignal with the logic triggering levels,VIH and VIL. From this signal, two kindsof short pulses are generated by theedge pulse generation circuit withrespect to the rising and falling edges.The high voltage LDMOSs convertsthese short-pulse voltage-signals to thecurrent-signals, and R1 and R2; remak-ing short-pulse voltage-signals at thehigh-voltage region. The “reshapers,”which are a type of a comparator with agiven threshold level, amplify the signals
to swing between VB and VS. Thesesignals determine the state of the S-Rlatch. By virtue of the S-R latch, the low-side input command can be transferredto the high-side region with only edgeinformation. Finally, the output drivercomposed by M1 and M2 drives anexternal high voltage power device withappropriate current driving capability.
High Dv/Dt Noise WarningA pulsating noise, however, also
affects the state of the S-R latch. Highdv/dt noise, especially, is the most dan-gerous type for this kind of pulse drivenHVIC. Assuming the parasitic capaci-tance, Cp, of the LDMOS seen at thedrain is 2pF and a noise which has dv/dtof 50V/nsec is applied to the VS node.This noise is coupled to VB by the boot-strapping capacitor, CBOOT. For theLDMOS drain voltage to follow the fast
changing VB voltage, Cp must becharged with large current. The currentcan be estimated using the followingequation:
I=Cp*dv/dt
The charging peak current due to highdv/dt noise reaches 100mA. The largecharging current makes enough voltagedrop on R1 and R2 to abnormally triggerthe S-R latch. If there is easy and effec-tive method to remove this noise per-fectly, that would be the best solution tomake the HVIC free from the malfunc-tion. However, it is impossible to sup-press the noise to be coupled to theHVIC. Therefore, it is very challengingwork to increase the immunity of theHVIC on the noise.
Allowable Negative Output VoltageVB and VS are the highest and lowest
voltage of the high-side driver, respec-tively. The voltage difference, VBS, isalmost constant by the bootstrappingcapacitor, CBOOT, and a bootstrappingtechnique. VS, however, depends on thehalf-bridge output, since VS is supposedto be connected to the source of theexternal MOSFET or the emitter of theexternal IGBT. An ideal HVIC must oper-ate regardless of the VS voltage situa-tion. For a real HVIC, however, VSheavily affects the operation. Someexamples of the HVIC operation versusVB are depicted in Figure 2 as describedin the following (A) and (B):
(A) VS > 0: The reshaper has the threshold level,
VTH. Thus, only when the level shifter’soutput swing is below this level, is thesignal recognized by the reshaper. If Vsis higher than 0, the level shifter’s outputswings enough to touch VTH level.Consequently, there is no problem fornormal HVIC’s operation.
(B) VS<-VTH:In this case, VS is negative. Accordingly,
the absolute VB voltage is lower thancase of (A). Therefore, the maximumoutput swing of the level shifter is limitedwithin VB and ground and,as a result,the reshaper cannot catch the levelshifter’s output and the S-R latch statecannot be updated.
Figure 1. Conventional high-side driver configuration.
Figure 2. Allowable negative VS problem: (A) VS >0, and (B) VS < -VTH.
Figure 4. Proposed reshaper configuration which provides a noise-canceling function and enhances allowable negative VS range.
DRIVER IC DRIVER IC
voltage for the V/I converter and tomake the level shifter to work, HIN input is never lost.
Another merit of V/I and I/V conver-sion is that it is easy to sense noise and to attach a noise canceling circuit.Because high dv/dt noise is added on to VB line via the CBOOT, noise is com-monly applied to all the elementsattached to the VB line. Thus, for thecommon-mode noise, which has highdv/dt, the V/I converter gives same out-puts. Whereas, the V/I converter outputsare different from each other for normaloperation, as only one of two LDMOSsoperates at a normal level shifter opera-tion. Thus, it is not difficult to determinewhether the V/I converter output is dueto noise or not. Once the noise cancellerrecognizes a common-mode noise intru-sion, it absorbs the current outputs ofthe V/I converter.
Owing to its unique topology, FairchildHVIC demonstrates good noise immuni-ty against high dv/dt noise up to 50V/nsecand guarantees an extended negativeoperation. At VBS=VCC=15V, the pro-posed method satisfies approximately–10V operation. Figure 5 shows themicro-photos of two new HVICs alsofrom Fairchild: the FAN7380 andFAN7382, which are designed using the HDG4 process.
RobustnessIn general, the HVIC drives high-volt-
age and high-current switching devices.Resulting high-voltage peaks, noise andEMI can cause the HVIC to malfunction,sometimes leading to destruction.Therefore, considering the actual usageof the HVIC, the robustness is the key issue.
To test the robustness of the designedHVIC, positively pulsating noise isadded to VBS in series and negativepulse is applied to VB line using theHP8114A’s pulse generator. During thetest, VBS is constantly biased to 15V.The test was performed by changing the output of an HP8114A programma-ble pulse generator from 0V to 80V. TheHVIC responses for the applied noisepower can be sorted into the followingfour categories:
(1) Normal Operation: The HVIC isnormally controlled by the command asa user intends.
(2) Abnormal Operation: The HVICskips and misses the input command ina medium noise peak. For a high noisepeak, the HVIC output is fixed to thelowest level.
(3) Latch Operation: The HVIC outputis fixed to the lowest level and large cur-rent flows toward the HVIC. But, ifreducing the applied noise power, theHVIC is recovered and works normally.
(4) Destruction: The HVIC is damaged
permanently. The HVIC is never recov-ered; in a worse situation, the packagegets cracked or explodes.
As shown in Figure 6, the FAN7380and FAN7382 show an abnormal opera-tion at high-voltage peak. However, the test vehicles are not destroyed anddo not fall into the latch state. Thismeans that the designed of these twodevices is very robust for high-energypulsating noise.
The HVIC design based the new tech-nique presented in this article demon-strates good noise immunity andextended allowable negative VS voltageswing. Furthermore, this effective HVICprocess offers a robustness which canovercome even difficult situations.
The HVIC itself is significancebecause it can replace the bulky pulsetransformer using silicon technology. Butthe more important factor to rememberis that a power-system-on-a-package ispossible with HVIC because it is built ona cheap familiar silicon wafer. For thesuccess of the power-system-on-a-package such as SPM (Smart PowerModule) and IPM (Intelligent PowerModule), the high-side and low-sidegates driver IC’s role is very importantand the proposed HVIC offers a work-able solution for designers.
28 Power Systems Design Europe May 2005
www.fairchildsemi.com
Figure 5. Micro-photo of the designed half-bridge driver: (A)FAN8380 which provides internally fixed 100nsec dead-time and is suitable for the ballast and low frequency PWMbased half-bridge control, and (B) FAN7382 whose currentdriving capability is 350mA/650mA (sourcing/sinking cur-rent) is a general purpose high/low-side gate driver IC.
Figure 6. Robustness test results: (A) Positive Pulse NoiseTest, and (B) Negative Pulsating Noise Test.
DRIVER IC
Visit us at PCIM Europe Booth 12-304
www.powersystemsdesign.com 3130 Power Systems Design Europe May 2005
Ferraz Shawmut, F Booth 12-230URL: www.ferrazshawmut.comProtection, Cooling
Floeth Electronic GmbH, DBooth 12-324URL: www.floeth-electronic.comIGBT-Driver-Controller
Fuji Electric GmbH, DBooth 12-304URL: www.fujielectric.comIGBT Modules
Infineon Technologies AG, DBooth 12-501URL: www.infineon.com/powerMOSFET, IGBT, ICs,
International Rectifier, GBBooth 12-201 URL: www.irf.com
MOSFET, IGBT, ICs, Modules, PowerSolutions
Isabellenhütte Heusler GmbH, DBooth 12-627URL: www.isabellenhüette.comPrecision Resistors, ASIC for monitoring
IXYS Semiconductor GmbH, DBooth 12-302 URL: www.ixys.comMOSFET, IGBT, ICs, Modules
LEM SA, CHBooth 12-401 URL: www.lem.comCurrent/Voltage Transducers
Microsemi Corp., USABooth 12-416URL: www.microsemi.comICs for power control
MITSUBISHI ELECTRIC EUROPE B.V., DBooth 12-320URL: www.mitsubishichips.comIGBT Modules; GTO
National Semiconductor, DBooth 12-349URL: www.national.comPower Management ICs
NEC Electronics (Europe) GmbH, DBooth 12-411URL: www.ee.nec.deMOSFETs
Ohmite Manufacturing Co., USABooth 12-606 URL: www.ohmite.comHigh Voltage Resistors
Power Electronic Measurement, GBBooth 12-546URL: www.pemuk.comRogowski Current Transducers
Power Integrations, USABooth 12-345URL: www.powerint.comPower Management ICs
Powersem GmbH, DBooth 12-218URL: www.powersem.dePower Modules
Power Systems Design MagazinesBooth 12-239, URL: www.powersystemsdesign.comThe European Power Magazine
SanRex Europe GmbH, DBooth 12-429 / 12-237URL: www.sanrex.comPower Modules
SEMIKRON International GmbH, DBooth 12-310URL: www.semikron.comMOSFET, IGBT, GTO, Modules, ICs,Power Solutions
Spoerle, DBooth 12-424URL: www.spoerle.deDistributor
STMicroelectronics GmbH, DBooth 12-106URL: www.st.comMOSFETs, Rectifiers, IGBTs
Texas Instruments, F Booth 12-229URL: www.ti.comICs, Power Management, Modules
Thermacore Europe Ltd., GBBooth 12-622URL: www.thermacore-europe.comThermal Management
Transim Technology Corporation, USABooth 12-444URL: www.transim.comDesign and Simulation Software
Tyco Electronics, DBooth 12-421URL: http://em.tycoelectronics.comPower Supply, Power Modules
Vishay Electronic GmbH, DBooth 12-547URL: www.vishay.comMOSFETs, Ceramic Components
Würth Elektronik eiSos GmbH & Co.KG, DBooth 12-628URL: www.we-online.deEMC and Inductive Components
Zetex PLC, GBBooth 12-208URL: www.zetex.comICs, MOSFET, Power Management
ABB Switzerland Ltd., CHBooth 12-408 / 12-503URL: www.abb.com/semiconductorsIGBT modules
Advanced Power Technology, USA;Europe, FBooth 12-416URL: www.advancedpower.comMOSFETs, IGBTs, Rectifier
Ansoft GmbH & Co. KG, DBooth 12-308URL: www.ansoft.comSimulation software
Apex Microtechnology Corporation,USABooth 12-526URL: www.apexmicrotech.comPower Operational Amplifiers, PWMAmplifiers
Coilcraft Europe Ltd. GBBooth 12-111URL : www.coilcraft.comInductors, Transformer, surface mount
CT Concept Technologie AG, CHBooth 12-102URL: www.IGBT-Driver.comIGBT driver boards
curamik electronics gmbh, DBooth 12-118URL: www.curamik.comCeramic substrates
Danfoss Silicon Power GmbH, DBooth 12-120URL: www.danfoss.com/siliconpowerCustomized Power Modules
EPCOS AG, D Booth 12-436URL: www.epcos.comCapacitors, Inductors, Varistors, Passives
Eupec, DBooth 12-501URL: www.eupec.comMOSFET Modules, IGBT, GTO, PowerSolutions
Eurocomp Electronik GmbH, DBooth 12-416URL: www.eurocomp.deDistributor for APT and Microsemi
Fairchild Semiconductor, USABooth 12-601URL: www.fairchildsemi.comMOSFET, IGBT, ICs, Opto
EXHIBITION FLOOR PLANEXHIBITION FLOOR PLAN
32 Power Systems Design Europe May 2005
From industrial applications to con-sumer electronics, and from homeappliances to office equipment,
designers of modern AC-DC SMPS cir-cuits are under pressure to develop highperformance, high power density cir-cuits. And with commercial, legislativeand environmental issues placing powerfactor correction (PFC) implementationat the top of the agenda, these design-ers are looking for solutions that willminimise component count, simplifyboard layout, and reduce board space. It is in the context of these demands thatInternational Rectifier has developed itsnew IR1150 family of µPFC (microPFC)devices for AC-DC power factor correc-tion (PFC) circuits.
The concept of the IR1150 is to pro-vide an integrated PFC controller, on asingle, compact SO-8 IC, that can beused across a wide application voltagerange. The key to the new device is theOne Cycle Control (OCC) techniqueemployed. OCC is a patented techniquedesigned to combine the performance ofthe continuous conduction mode (CCM)technologies traditionally employed inhigher power PFC designs with the sim-plicity, reliability and low componentcount of the discontinuous current mode(DCM) approach conventionally usedwith lower power applications. Powerfactor is the relationship between AC
voltage and current waveforms and is ameasure of “power quality” that affectsthe efficiency of electrical transmissionnetworks. With unity PF as the ultimategoal, the IR1150 is designed to deliver a0.999 power factor with only 4% totalharmonic distortion.
IR1150 OverviewFigure 1 shows a schematic diagram
of the new IR1150 IC, which is intendedfor boost converters for PFC operatingat fixed frequency in continuous currentmode. By using the OCC method, theneed for an analogue multiplier, ACinput voltage sensing or a fixed oscillator
Saving board space and reducing component count
A new power factor correction IC that can help designers to halve the size of their control boards
in applications with power ratings from 75W to 4kW
By Stephen Oliver, International Rectifier
Figure 1. Shematic diagram of the IR1150 IC.
POWER FACTOR CORRECTION
www.powersystemsdesign.com34
Finally an Output Under Voltage pro-tection OUV is provided. In the case ofoverload or brown out, the converter will automatically limit the current: as aresult the output voltage will drop. If thedrop exceeds 50% of the nominal outputvoltage, the controller will shut downand restart.
An additional feature of the IR1150 isthe ability to force the IC into a “sleep”mode. In the sleep mode, the internalblocks of the IC are disabled and the ICdraws a very low quiescent current of120µA (typ.). This is a desirable featuredesigned to reduce system power dissi-pation to an absolute minimum duringstandby, or to shut down the converterat the discretion of the system designer.The sleep mode is activated any time theOVP pin is at a voltage lower than 0.62V.
Designing PFC applications with the new ICs
Figure 3 shows an IR1150 IC beingused in a typical PFC circuit. Despite itssmall size, the IR1150 still provides afast, 1.5A gate drive. For applicationswith higher powers (over and above1.5kW, depending on switching frequen-cy), the only major addition needed tothe application design is the substitutionof this gate drive for large, paralleledpower semiconductors (FETs or IGBTs).
In order to further simplify the imple-mentation of PFC designs based on thenew IC, IR has also developed a num-ber of support materials and tools,including application notes and ademonstration board. In addition to theIR1150 device itself, this demo boardincorporates a high-efficiency powerswitch, a hyper fast recovery boostdiode, and all of the other componentsneeded to evaluate the performance ofthe IR1150 control IC in a 300W contin-uous conduction mode boost converterdesign. Using the board, designers canprogram frequency (between 50kHz and100kHz), evaluate input, output andpower factor performance for differentload and line conditions, identify compo-nent requirements for their specificapplications, and conduct a range ofother tests including EMI measurement.
ramp (as used in conventional CCM-based designs) is completely eliminated.
Instead, IR’s OCC uses a proprietary‘integrator with reset’ circuit. In thiscase, the output of the error amplifier isintegrated over each clock cycle to gen-erate a variable-slope ramp. This rampis then compared with the error voltageand subtracted from the current sensesignal to generate the PWM gate drive.In a 1kW example, this control methodrequires 40% fewer resistors and 50%fewer capacitors than traditional multipli-er techniques and delivers a unified
control solution that is adaptable to vari-ous topologies for both leading and trail-ing edge modulation.
Figure 2 shows the new IR1150 imple-mentation for a 1kW PFC compared tothe traditional multiplier-based solution.The IC operates with essentially twoloops, an inner current loop and an outervoltage loop. The inner current loop sus-tains the sinusoidal profile of the aver-age input current based on the depend-ency of the pulse width modulator dutycycle on the input line voltage, to deter-mine the analogous input line current.
Thus, the current loop exploits theembedded input voltage signal to com-mand the average input current follow-ing the input voltage. This is true so longas operation in continuous conductionmode is maintained.
The outer voltage loop controls theoutput voltage of the boost converterand the output voltage error amplifierproduces a voltage at its output. Thisvoltage directly controls the slope of the integrator ramp, and therefore theamplitude of the average input current.The combination of the two control elements controls the amplitude andshape of the input current so as to beproportional to and in phase with theinput voltage.
Integrated protection andsleep functions
In any PFC design, implementing pro-tection must be a design priority, and in the case of the 1150, much of therequired protection has been integratedinto the IC. This integrated functionalityincludes protection from system levelovercurrent, overvoltage, undervoltage,and brownout conditions.
Open Loop Protection (OLP), forexample, prevents the controller fromoperating if the voltage on the feedbackpin has not exceeded 20% of its nomi-nal value. If for some reason the voltagecontrol loop is open, the IC will not start,avoiding a potentially catastrophic fail-ure. As soon as the VCC voltageexceeds this threshold, provided thatthe VFB pin voltage is greater than20%VREF, the gate drive will beginswitching. In the event that the voltageat the VCC pin should drop below thatof the UVLO turn off threshold, VCCUV-LO , the IC then turns off, gate drive isterminated, and the turn on thresholdmust again be exceeded in order to restart the process.
A dedicated programmable OverVoltage Protection pin (OVP) is avail-able to protect the output from overvolt-age. If the output voltage exceeds theset OVP limit, the gate drive will be disabled until the output voltage againapproaches its nominal value.
Figure 2. IR1150 implementation for a 1kW PFC compared to the traditional multiplier-based solution.
Figure 3. IThe IR1150 IC being used in a typical PFC circuit.
Power Systems Design Europe May 2005
www.irf.com/eu
POWER FACTOR CORRECTION
www.powersystemsdesign.com 3736 Power Systems Design Europe May 2005
A low cost solution for increasing deliverable power
Passive current sharing can also improve the reliability of N+1 power module configurations,
by reducing the stress on each converter in the system without the need for
any additional active circuitry.
By Barry Ehrman, Artesyn Technologies
Connecting two or more DC/DCconverter modules in parallel toincrease current availability is a
perfectly reasonable design aim. Butfew low-cost converters offer currentsharing facilities, so designers need toimplement their own circuits. One optionis to add active circuitry to the output of each converter, but this tends to beexpensive and takes up valuable boardspace. Another approach is to use pas-sive current sharing, which involves putting droop resistance in series withthe converters’ outputs to equalise theirvoltages. This is much simpler to imple-ment, costs next to nothing, and takesup hardly any space. This article isintended to clear up the mysteries ofpassive current sharing, using a fullyworked real-life example to show theadvantages and disadvantages of the technique.
Although passive current sharing can-not be used to obtain double the currentoutput you would get from a single con-verter (because one of the converterswill always try to output more than halfthe total load current, and thereforeexceed its maximum rating), it providesa highly scalable means of accommo-dating demands for more power, whichtypically result from increases in systemsize or functionality over time. Passivecurrent sharing can also improve thereliability of N+1 power module configu-rations, by reducing the stress on each
converter in the system without theneed for any additional active circuitry.
Unfortunately, the simplicity of thismethod of paralleling is not without itstradeoffs – the biggest being loss ofsystem efficiency and load regulation.Whether these tradeoffs are acceptableis obviously a design decision, and to alarge degree depend upon the applica-tion. In the example presented in thisarticle, load regulation is less of anissue, because the paralleled convert-ers are feeding an on-board intermedi-ate bus, to supply multiple point-of-load(POL) converters which provide furtherdown-conversion and regulation fortheir various silicon loads.
We have chosen to illustrate theadvantages and disadvantages by paralleling two Artesyn TQW14A-48S12 intermediate bus converters(IBCs). These are wide-input 168 WattDC/DC converters, primarily intended
for telecom applications, which convert anominal 48V DC input into 12V DC out-put. TQW14A-48S12 IBCs can outputup to 14A with a typical efficiency of95%, and are not equipped with activecurrent sharing facilities. For the purpos-es of illustration all calculations arebased on worst-case component toler-ances. Figure 1 shows the two IBCs inan N+1 redundant passive current share configuration.
In addition to the two converters,there are two Schottky ORing diodes,D1 and D2, to decouple the outputs.
These are assumed to have a forwarddrop of 0.2V, plus a resistive componentequivalent to 7 milliOhms, which models the datasheet curves for theSTPS20L15 devices manufactured bySTMicroelectronics.
In order to implement current sharingbetween the two converters using Oringdiodes, their output voltages ideallyneed to be adjusted to be perfectlymatched under all conditions. However,in the real world it is nearly alwaysimpossible to achieve this degreeof trim accuracy; furthermore, inthe example we are using, for rea-sons of economy the IBCs areonly designed to produce looselyregulated outputs, and do not offera voltage trim facility.Consequently, there are twooptions. One is to implement anactive circuit on the output of theconverters, to force them to cur-rent share: this is relatively expen-sive, and takes up valuable boardspace. The second option is toemploy passive current sharing,which involves putting droopresistance in series with the out-puts. This droop resistance willcreate enough voltage drop underload to cause the two converters’voltages to equalise, and the converters will consequently current share.
To complete the circuit in Figure1, we need to determine the valueof the droop resistors R1 and R2.The major tradeoffs are the follow-ing: if the droop resistors are toosmall then there will not be enoughvoltage drop under load to causethe converters to share the load.Conversely, if the droop resistorsare too large, the final voltageunder full load will drop too low tobe useful. To determine the idealvalue, we need to determine themaximum variation allowed in thevoltage to the load.
The first consideration is theminimum voltage the TQW14AIBCs will output under worst-caseconditions. This will occur when
their input voltage is at the bottom endof its permissible range, i.e., 36V.According to the datasheet, the outputvoltage could then be as low as 12Vminus 10%, which is 10.8V.
The second consideration is the mini-mum voltage which the load can toler-ate. The TQW14A IBC is primarilydesigned to drive POL converters, sowe’ll assume that these constitute theload in this case. The nominally 12V
input POL converters that Artesyn pro-duces fall into three groups, with inputranges of 10.8 to 13.2V, 10.2 to 13.2V,and 10 to 14V. Obviously we cannotdrive the POL converters with a 10.8 to13.2V input range, because there’s nomargin. So for the purposes of example,we’ll use the second group, and limit thedroop to 600mV.
To determine the value of R1 and R2,we first need to subtract the voltage
POWER MANAGEMNENT
Figure 1. Typical N+1 RedundantPassive Current Share Configuration.
POWER MANAGEMNENT
38 Power Systems Design Europe May 2005
drop caused by the isolating diodesfrom the 600mV, as shown below:600mV - 200mV - ((14 Amps x 0.007Ohms) x 1000) = 302mV
By using Ohms law: voltage = current xresistance
R1 = R2 = 0.302 Volts / 14 Amps =0.0215 Ohms or 21.5 milliohms.
For the circuit we will choose the nextlowest standard value, which is 0.020Ohms. Assuming 1% tolerance resis-tors, the minimum value will be 0.0198Ohms, and the maximum will be 0.0202Ohms.
The circuit is now designed. Thequestions are: how well does it work,and what is the theoretical loss of effi-ciency? We also need to bear in mindthat the resistance of the PCB conduc-tor traces will affect the results. Becausethis resistance varies from application toapplication, we will assume a value ofzero Ohms for the purposes of thisexample. The resistance of the PCB
traces will tend to improve current shar-ing while decreasing system efficiency.
Through circuit analysis the outputvoltage = Vout1 - Iout1 x R1 = Vout2 -Iout2 x R2, and the load current =Ioutload = Iout1 and Iout2.
The individual output currents Iout1and Iout2 can be calculated by the fol-lowing formulae:
Iout1 = (Vout1 - Vout2 + (R2 x Ioutload))/ (R1+R2)
Iout2 = Ioutload - Iout1
Vout = Vout1- (Iout1 x R1)
Table 1 shows Iout1 and Iout2 undervarious load current conditions:
Note that the formulae for Iout1 andIout2 indicate a negative current forIout2 for load currents of 5 Amps orless. Because of the ORing diodes, neg-ative currents are blocked, which resultsin 0 Amps for Iout2. At the other end ofthe current output scale, note that pas-
sive current sharing can only be usedup to 22A – beyond this the maximumoutput capability of one of the IBCs isexceeded, as shown in red.
In addition, Table 1 shows the powerloss due to the resistors and the ORingdiodes, together with the overall effecton efficiency. As can be seen from thetable, passive current sharing is far fromperfect. With the circuit constraints con-veyed by the load and 1% standardparts, the worst case load sharing istheoretically 24.4% (based on the 0.02Ohm droop resistors) with a 22A loadshared between the two converters.However, this load sharing is achievedwith an efficiency loss of only 4.05%.
It is important to note that we areusing worst-case figures to illustratepassive current sharing. Based on actu-al Cpk (Process Capability Index) sam-ple measured data, using extreme val-ues, the worst case output voltage val-ues for the TQW14A IBC are 12.098Vmaximum and 11.957V minimum. Afterallowing for the voltage drop of the
Table 1. Worst-case current sharing scenario for two TQW14A IBCs fitted with 0.02 Ohm droop resistors.
ORing diode, these reduce to 11.898Vand 11.757V respectively. A more rea-sonable scenario would be to use theactual Cpk sample measured data, butwith values equivalent to the StandardDeviation from the average. This wouldproduce converter output voltages of12.076V maximum and 12.006V mini-mum, giving post ORing diode values of 11.876V and 11.806V respectively.Although overall efficiencies remainlargely unchanged, the effect of usingthe more reasonable output voltage figures is an improvement in currentsharing accuracy to 11%, and the paral-leled IBCs will now deliver up to 25Abefore the output rating of a converter is exceeded.
Passive current sharing offers aninexpensive means of accommodatingincreased on-board power demandswithout major redesign, provided theslight loss in conversion efficiency canbe tolerated. Although we have chosento illustrate the technique using two
IBCs feeding POL converters via anintermediate bus, the approach is also suitable for use with conventionalbrick type converters with tightly regulated outputs.
Eliminating the ORing diodes wouldobviously increase overall efficiency, butit would be necessary to guarantee afairly high minimum load at all times.However, this approach is not withoutrisk, because the FETs used by the con-verters’ synchronous rectification stagescan sink or source current when active,and power could consequently circulatebetween the two converters.
Other reasons for adopting passivecurrent sharing are to increase reliabilityof N+1 power module configurations,and to secure better performance onboards that do not use an intermediatevoltage bus and point-of-load convert-ers. If the board contains widely distrib-uted loads, better voltage regulation willbe achieved by placing the converters
as closely as possible to the loads—perhaps using both sides of the board—and the amount of copper can bereduced because the board traces willbe carrying less current.
About the author:Barry Ehrman is Marketing ApplicationEngineer for Artesyn Technologies'Power Group. He is based at the com-pany’s facilities in Tucson, Arizona, andcan be contacted via email [email protected].
Company contact:Jackie Day, Marketing CommunicationsManagerArtesyn Technologies, SpringfieldIndustrial Estate, Youghal, Co. Cork,Ireland.
Tel: +353 24 25388 Fax: +353 24 25407 Email: [email protected]
www.artesyn.com
POWER MANAGEMNENT POWER MANAGEMNENT
40 Power Systems Design Europe May 2005
International centers for complete power sub-assemblies
The modern power electronics solution for markets such as wind, solar, traction, welding,
elevators and electric vehicles is a complete sub-assembly. The decision to select an
all-round solution which includes a complete sub-assembly, the specification and design,
production, arduous testing and after-sales support and service is time-saving,
cost-effective and so simple with an experienced partner.
By Fréderic Sargos, International Coordinator Solution Centers Network and Paul Newman,
Managing Director, Solution Center United Kingdom
Semikron combined the expertiseof nine Solution Centers fromeach continent in South Korea,
Australia, South Africa, USA, France,United Kingdom, Brazil, India andSlovenia into a unique Solution CentersNetwork. The main goal is the develop-ment of cost-effective platform productswhich are marketed in high quantities inthe global network. Additionally thecapacities in production, purchase, engineering and logistics are united.The hands-on organization makes thenetwork even more competitive.Specifically global customers profit fromthe local service and support as well asthe swiftness with which the networkoperates. The speciality of over 40design and application engineers is tofind the best possible solution for a specific customer inquiry.
Joined forces Over 5 decades ago Semikron started
integrating semiconductors into powerassemblies, called SEMISTACK® .These completely tested systems con-sist of diode/thyristor, MOSFET or IGBTmodules with gate driver, protection,cooling, DC link capacitors and sen-sorics. More than 15 000 differentdesigns have been successfully imple-mented in the industry. Semikron spe-cializes in the most arduous applicationssuch as wind plants or traffic engineer-ing and produce customized as well as
standard SEMISTACK`s ranging fromsimple press fit diode plates for batterychargers to specific solutions for eleva-tors, wind, solar plants, traction electricvehicles and marine. For example, 57%of globally installed power capacity invariable speed generators is poweredby Semikron, a significant part of itbeing made of assemblies.
The customer benefitsGlobal support for local service is the
philosophy of the Semikron SolutionCenters by providing development andproduction capabilities as close as pos-sible to the customers’ applications.Wherever the intended application, thecustomer is offered local expertise andhighly qualified service of one of the
Solution Centers’ team of applicationengineers, as well as local develop-ment and production, for a high qualityproduct and support.
Every Solution Center has its ownteam of application engineers, expert-ise as well as knowledge transfer withinthe network. This allows tried-and-test-ed systems from one center to betransferred to another, closer to its mar-ket, for manufacture. Semikron is theonly company in this market to offersuch a uniform worldwide design, sup-port and build network, combining glob-al design resources and local customersupport. The development, design and
experiences of different projects areshared in the network. Customer specif-ic designs are handled with the utmostconfidentiality and loyalty.
The testing policy of the SolutionCenters is a quest for zero defect ratesin the field. All SEMISTACK`s are visual-ly and electrically checked. The inverterunits are tested under full load andshort-circuit conditions.
The first platform solution The SEMIKUBE B6CI is a power elec-
tronics platform solution meeting marketdemands for a standardized, compact,maintainable, flexible and above allcost-effective system for forced air-cooled inverters from 300A up to 1550A.
Figure 1. Solution Center Network.
POWER MODULES
42 Power Systems Design Europe May 2005
www.semikron.com
It is the first standard platform solutiondeveloped by the Semikron SolutionCenters Network and is the result of 45years of experience in the stack business.
The ground-breaking cubic form witha low inductivity connector for 220kW –900 kW drives allows for similar SEMI-KUBE blocks to be connected in variousmechanical and electrical arrangementsto achieve the desired electrical andmechanical performance. The lowinductive connector makes a fast con-nection and disconnection possible. The cubic form of the blocks also offersthe possibility to install the products hor-izontally or vertically into the cabinet.
To achieve the most cost-effectiveindustrial platform, parts used within therange from 385A – 1550A, (at 3kHzswitching frequency) are reduced to theminimum, the whole range can be cov-ered with only one reference of IGBTmodule, capacitor, heat sink, fan, andcurrent sensor. The SEMITRANS mod-ule family with the latest IGBT chip gen-eration is optimal for the SEMIKUBErange. The Semikron SKHI gate driverfamily along with patented close protec-tion scheme gate boards protect thedevices with electrical isolation, Vcemonitoring, short pulse suppression, under-voltage monitoring and interlocking.
The SEMIKUBE range has forced air-cooling of the capacitors guaranteeing along lifetime expectancy and avoidingthe occurrence of hot spots. The excel-lent symmetry of the busbar system be-tween the capacitor blocks and the IGBTmodules ensure a current sharing be-
tween the modules and between thecapacitors. All five sizes use the samespare parts. This ensures easy integra-tion, plug and play and easy maintenance.Three of these sizes have a half controlledrectifier with a pre-charge system included.
All removable parts weigh less than30 kg and all screwing points are acces-sible from the front side. For a largeinstallation with multiple power ratingsless spares are required due to themodularity of the SEMIKUBE. The sameparts are used through the whole range,which simplifies training of the after-sales team and allows an improvedavailability of the parts.
All technical performances result in apower/volume ratio, which was previ-ously only achievable through water-cooling. Although SEMIKUBE B6CI wasdesigned as a standard platform it wasalso designed to allow customization,such as changing the power modules,the capacitors length or the type and orientation of the fans without changingthe whole concept.
The customer specific configurable“black box” solution
The SEMISTACK range is furtherexpanded with an intelligent power solu-tion, integrating the SEMiX IGBT half-bridge modules and SKYPER half-bridge driver core containing all basicdriver functions. It takes the form of anoff-the-shelf or configurable “black box”that offers all the functionality that isneeded for a wide range of applications—all the user needs to do is provide acontrol scheme. This SEMISTACK solu-tion can be expanded as a platform withbuilt-in intelligence, incorporating thenewest advances in power electronics
while using common piece parts wher-ever possible. It can be based onSEMiX 2, 3 or 4 IGBT half-bridge mod-ules. These are the latest-generation,lowest-loss, highest efficiency IGBTmodules, which are housed in a verylow inductance, 17mm high package. Arange of current ratings is available with-in each SEMiX family, providing an over-all choice ranging from 250 to 1300A.Therefore, the stack can be configuredfor low, through medium, to high powerto suit a wide range of applicationsincluding motor drives, inverters, unin-terruptible power supplies and renew-able energy generation. This SEMI-STACK platform is believed to be themost compact and volumetric-efficientsolution on the market. It is fully protect-ed by the incorporation of the SKYPERintelligent gate driver circuits which, inaddition to basic driver functions andelectrical isolation, also include VCE
monitoring, short pulse suppression,under-voltage monitoring and interlocking.
Other features of this SEMISTACKwith SEMIX and SKYPER include low-inductance co-planar bus barring; DC-linked voltage monitoring; highly-accu-rate, closed-loop current sensors andthermal measurement; high-frequencysnubber capacitors; and DC-linked bulk capacitance.
The units have been designed to beplug-and-play and are fully electricallyand thermally tested. They are availablein air-cooled and, for even more com-pact solutions, water-cooled variants.
Figure 2. SEMITOP inverter for ele-vators. SEMITOP is based on baseplate-free technology and offers thebest reliability in very high cyclingapplications such as elevators in the lower kW range.
Figure 3. SEMIKUBE B6CI, the cubicdesign realizes the most efficient modularity.
Figure 4. SEMISTACK intelligentmodular solution incorporating SEMiX IGBT half-bridge modules and SKYPER half-bridge driver core.
POWER MODULES
44 Power Systems Design Europe May 2005
The wLEDs visible spectrum appears whiteThanks to liquid crystals, electronic displays continue to shrink in size. Wrist watches, cell phones,
PDAs and laptop computers have LCD-based displays. LCD and other, smaller display
technologies are also slowly replacing the cathrode-ray tube (CRT) based
desktop monitors and televisions.
By Jeff Falin, Texas Instruments
The electronics behind an LCD arevery different than those of a CRT.A CRT has three rear-mounted
electron guns whose beams convergeon a material at the front of the displaythat emits light. However, regardless ofwhether an LCD display is the olderdual-scan twisted nematic (DSTN) typeor the high-image quality thin film tran-sistor (TFT) type, LCD is a transmissivetechnology. As shown in Figure 1, anLCD works by varying the amount offixed-intensity white “backlight” as itpasses through an active filter.
The actual liquid crystals are placedbetween two finely grooved surfaces:the alignment layers and polarizing fil-ters. Since the two alignment layers areperpendicular to each other, the crystalsbegin with the orientation of one layerand then twist to accommodate the per-pendicular orientation of the other layer.Light follows the alignment of the liquidcrystals as it passes through. However,if a voltage is applied across the liquidcrystals, they rearrange themselves ver-tically. This forces the light to passthrough vertically, only to be stopped bythe second polarizing filter.
For small- to medium-sized displays,this backlight is provided from white-light emitting diodes (wLEDs). LargerLCD displays, like those found in a lap-
top computer, typically use cold cathodefluorescent lamps (CCFL) to provide thebacklight. This first article of the threepart series details what to considerwhen powering wLEDs and providesexample circuits for powering them. Thesecond article explains the requirementsand gives examples of how to providethe various voltage and currents used inthin-film-transistor (TFT)-based activematrix displays. The last article in theseries explains the options available forand complexities of powering CCFLs.
wLEDAn LED is a semiconductor diode with
an inorganic chemical compound thatgives off light when an electric current
passes from the diode’s anode to itscathode. LEDs emit light proportional tothe amount of current being driventhrough them. For wLEDs, the emittedlight covers the entire visible spectrumso the light appears white. All LEDshave a non-linear V-I curve, so even thesmallest change in the forward voltageof the wLED creates a change in thewLED current. In addition, the forwardvoltage tolerance can vary by 1V ormore. Usually, multiple wLEDs config-ured in parallel or in series are neces-sary to produce a uniform intensity overthe display. A forward current for wLEDscan be generated by either a constantvoltage source (Figure 2) or constantcurrent source (Figure 3).
Constant Voltage vs. Constant CurrentUsing a constant voltage source to
drive multiple, parallel wLEDs as shownin Figure 2 is perhaps the simplest andleast expensive solution. The ballastresistors are required to promote evencurrent sharing among the diodes inparallel as well as to provide over-cur-rent protection. The primary weaknessof this method is that the feedback loopregulates the voltage across both theseries ballast resistor and wLED insteadof the current through the wLED. In thisconfiguration, the actual voltage dropacross each wLED, and therefore eachindividual wLED’s light intensity may vary.
POWER MANAGEMENT
Figure 1. Simplified LCD cell.
46 Power Systems Design Europe May 2005
So, wLED-1 may provide light intensityat VF = 3.6V and IF = 20mA while wLED-2 might require VF = 3.8V to provide thesame light intensity with IF = 20mA. Inaddition, the ballast resistors wastepower, thereby lowering the overall effi-ciency of the system and decreasingbattery life. This is unacceptable formany applications.
The preferred method for drivingwLEDs is from a constant currentsource as shown in Figure 3. The con-stant current source eliminates changesin current due to variations in thewLED’s forward voltage. A constant,controllable forward current deliversconstant, controllable display bright-ness. Rather than regulating the outputvoltage of a supply, the controller regu-lates the voltage across a single currentsense resistor as shown in Figure 3.The converter varies its output voltageuntil the wLED current being sensed
reaches regulation. The converter’s reference voltage and the current senseresistor value determine the wLED cur-rent. Most displays require more thanone wLED. The wLEDs should be con-nected in a manner that ensures identi-cal current flow through each to providebalanced brightness between them.
Special Requirements when Driving wLEDs
Many applications using a displaypanel require a backlight dimming func-tion. With products such as PDAs, theuser adjusts the brightness for ambientlight conditions. With other products,such as cell phones, the processorautomatically dims and turns off thebacklight after a period of inactivity. Twomethods are used for implementing adimming function: analog and PWM.With analog dimming, the display isdimmed to 50 percent brightness byapplying 50 percent of the maximum
current to the wLEDs. Drawbacks to thismethod include a wLED color shift andthe need for an analog control signal,which is not usually readily available.PWM dimming is achieved by applyingfull current to the wLED at a reducedduty cycle. For 50 percent brightness,full current is applied at a 50 percentduty cycle. The PWM signal frequencymust be above 100Hz to ensure that thepulsing is not visible to the human eye.The maximum PWM frequency dependson the power supply start-up andresponse times. For maximum flexibili-ty and ease of integration, the wLEDdriver should be able to accept PWMfrequencies as high as 50kHz. The
dimming signal is usually derived from a GPIO on the system processor.
Constant current wLED driversrequire over-voltage protection in theevent of an open circuit fault condition.The wLEDs are often on a separatePWM than their driver; therefore, anopen circuit condition is created if theconnector pins become dislodged.Another fault condition exists if a wLEDfails and becomes an open circuit. Ineither case, the driver increases its out-put voltage in an attempt to provide aconstant current. Without protection,the output voltage can raise highenough to damage the IC or output
capacitors. The easiest way to protectthe driver is to choose an IC that has abuilt-in over-voltage comparator thatlimits the maximum output voltage.Alternatively, a Zener diode can be usedto clamp the maximum output voltage;however, this is very inefficient becausethe maximum programmed currentflows through the Zener diode during afault condition.
An often overlooked feature in awLED driver supply is load disconnect,which electrically removes the wLEDsfrom the input source when the supplyis disabled. This feature is important intwo situations: shutdown and PWM dim-ming. Even when an inductive boostDC/DC converter is turned off, the loadis still connected to the input throughthe inductor and catch diode. Since theinput voltage is still connected to thewLEDs, a small current continues to
Figure 2. Constant voltage source powering wLEDs in parallel with ballast resistors.
Figure 3. Constant current source powering wLEDs in series.
POWER MANAGEMENT
www.powersystemsdesign.com 4948 Power Systems Design Europe May 2005
www.ti.com
The backlight driver is often on a differ-ent PWM than the actual LEDs them-selves, so power must be routed fromone board to another. Driving fourwLEDs in parallel requires a total of sixconnector pins, while driving them inseries requires only two pins.
The TPS61043 wLED driver shown inFigure 6 is ideally suited to meet these
requirements. It is a fully-integratedinductive mode boost converter with an integrated power FET that has beencustomized to operate as a wLED driver. The TPS61043 has built-in loaddisconnect, over-voltage protection andPWM dimming in a 3x3 mm QFN pack-age. Its 1 MHz switching frequency min-imizes the size of the inductor. In a typi-cal application, the TPS61043 can drive
four wLEDs in series with 15 mA from aLi-Ion battery at 83 percent efficiency.The datasheet provides several typicalapplication circuits and recommendedcomponents as a starting point for this design.
Because LCD is a transmissive tech-nology, wLEDs are critical componentsin today’s small to medium LCD dis-plays. While wLEDs can be poweredfrom any DC/DC converter configuredas a constant voltage source, poweringthem from a constant current sourceprovides the most uniform light intensi-ty. Features, such as over-voltage pro-tection, load disconnect and dimming,
not found in many DC/DC convertersare also desirable when poweringwLEDs. The TPS62030 charge pumpand TPS61043 boost converter are constant current power solutions withthe additional features needed to power wLEDs.
Figure 6. Typical wLED Driver Circuit Using the TPS61043.
POWER MANAGEMENT
flow even when the supply is disabled.Portable products may spend more than95 percent of their time in stand-bymode, so even small leakage currentssignificantly reduce battery life. Loaddisconnect is also important while thePWM is dimming. During the dimmingperiod off-time, the supply is disabledbut the output capacitor is still connect-ed across the wLEDs.
Without load disconnect, the outputcapacitor discharges through the wLEDsuntil the dimming pulse turns the supplyon again. Since the capacitor is partiallydischarged at the beginning of eachdimming cycle, the supply must chargeup the output capacitor at the start ofeach cycle. This creates a spike ofinrush current each cycle, which lowerssystem efficiency and creates voltagetransients on the input bus. With loaddisconnect, the wLEDs are electricallyremoved from the circuit, so there is noleakage current when the supply is dis-abled and the output capacitor remainsfully charged between intervals of PWMdimming. A load disconnect circuit isbest implemented by placing a MOSFETbetween the wLEDs and the currentsense resistor. Alternatively, placing theMOSFET between the current senseresistor and ground provides load dis-connect but inadvertently creates anadditional voltage drop that appears asan error in the output current set-point.
wLEDs in ParallelAll ICs designed specifically to drive
wLEDs provide a constant current.
Most solutions are eitherinductive-based orcharge-pump-based. The primary advantage of a charge pump, orswitched capacitor, solu-tion is that this does notrequire an inductor. Power is transferredfrom the input to the output by discretecapacitors. Charge pump power sup-plies are easy to design since the com-ponent selection is usually limited tochoosing the correct capacitor from adatasheet list. The main disadvantage of a charge pump solution is its limitedoutput voltage. Most charge pump ICscan only provide an output voltage up totwice the input voltage. Therefore, driv-ing more than one wLED with a chargepump requires them to be driven in par-allel. When driving multiple wLEDs witha charge pump that only regulates theoutput voltage, ballast resistors arerequired to improve current matching.
However, the latest charge pump ICsthat are specifically designed to drivewLEDs have incorporated current shar-ing circuitry directly into the IC. As shown inFigure 4, the TPS60230 is a chargepump solution that generates a voltagein parallel across the anode of multipleLEDs. The resistor to ground from theISET pin determines the current.
Each wLED cathode is directly con-nected to a different pin on the IC.Internal circuitry regulates the forwardcurrent through each wLED to within 0.3percent without the need for external
resistors. The device has over-voltageprotection and being a charge pump,has inherent load disconnect. The EN2pin is used to implement PWM dimming.Analog dimming is implemented byapplying an analog signal through aseries resistor to ISET instead of tying it to ground through a resistor. Mostcharge pump based solutions have anacceptable efficiency between 60 to 85percent. The more advanced chargepump ICs have fractional conversionmodes that automatically switch to themost efficient conversion mode as theinput voltage changes. Figure 5 showsthe efficiency graph of a fractional con-version charge pump as it switches fromone mode to another.
wLEDs in SeriesWith their small size and high efficien-
cy, inductive DC/DC converter solutionsare ideally suited for most portable con-sumer products to provide longer batterylife. Through careful selection of a boostconverter’s inductor and catch diode,the designer can tailor an inductive converter’s efficiency to optimize thetrade-off between size and efficiency.Because boost converters are capableof providing an output voltage muchlarger than their input voltage, they can drive many wLEDs in series.
Figure 4. Typical wLED driver circuit using the TPS60230.
Figure 5. Fractional Charge Pump Efficiency CurveShowing 1x to 1.5x Mode Change.
POWER MANAGEMENT
www.powersystemsdesign.com 5150 Power Systems Design Europe May 2005
of modules and some of these modulesmay have load switching or inductivedevices such as motors that can createsurge pulses and high frequency noiseon the power lines. Furthermore, noiseon the power lines is easily coupled intodata line signals because the power anddata lines are often located inside thesame wire bundle. The coupled noiseproduced on the data lines can producea surge voltage that can permanentlydamage an IC.
The immunity level of a module torepetitive high frequency voltage EMIsurges, also known as electrical fasttransients (EFT), can be measured withthe IEC 61000-4-4 test. Repetitivesurges are modeled by a recurring pat-tern of high voltage spikes of 50 ns tran-sient pulses in bursts of 15 to 300 mslong. This test is used to verify a mod-ule’s immunity from noise sources suchas inductive load switching and relaycontact chatter. The switching transientsare coupled into the data line cablesbecause of the parasitic capacitanceand inductance inherent in the wiringharness. The ESD5Z5.0T1 TVS diodesprovide an IEC 61000-4-4 rating of 50 A.
The IEC 61000_4_5 test serves as astandard test to verify the immunity of asystem to a non-repetitive surge. Thesurge voltage waveform is defined by adouble exponential pulse with a risetime of 8 µs and duration of 20 µs. The 8 µs x 20 µs test is often used toquantify the power rating of a TVSdevice and is representative of a singlesurge event such as the voltage inducedby lightning. An ESD5Z5.0T1 TVS diode with a break down voltage of 6.2 V has an IEC 61000-4-5 rating of9.4 A and 174 W. Figure 2 shows theESD5Z5.0T1’s response to the 8 x 20 µs surge test.
TVS Selection and PCB GuidelinesTVS diodes provide a simple, low cost
solution for surge protection. Listed beloware recommendations on selecting anappropriate TVS device, along with PCBlayout suggestions that maximize theeffectiveness of the protection circuit.
TVS Selection GuidelinesSelect a device with a working voltage
that is greater than the maximum busvoltage. Select a device with a clampingvoltage less than the maximum speci-fied voltage for the protected circuit.
Verify that the TVS device’s power rat-ing can dissipate the energy of thesurge pulses. The capacitance of theTVS device should be minimized forhigh speed circuits.
PCB RecommendationsProtect all I/O signals entering or leav-
ing the PCB with a TVS device. Locatethe protection devices as close as pos-sible to the I/O connector. This allowsthe devices to absorb the energy of thetransient voltage before the surge pulsecan arc and propagate between adja-cent PCB traces.
Minimize the loop area for the high-speed data lines, as well as the powerand ground lines, to reduce the radiatedemissions and the susceptibility to RF noise.
Use ground planes to reduce the par-asitic capacitance and inductance. Thiswill prevent voltage overshoot and ring-ing on high speed data lines. Minimizethe PCB trace lengths for the groundreturn connection.
Figure 1. EC 61000-4-2 ESD Test. Figure 2. IEC 61000-4-5 8µ x 20µ Surge Test.
www.onsemi.com
TRANSIENT PROTECTION
Low-cost for portable applications
Designers are being challenged to develop reliable systems that meet stringent Electrostatic
Discharge (ESD) and Electromagnetic Interference (EMI) standards, while reducing
the size and cost of their designs.
By Jim Lepkowski, ON Semiconductor
Anew series of Transient VoltageSuppressors (TVS) avalanchediodes have been developed to
protect sensitive electronic componentsfrom the surge pulses that arise fromESD and EMI. The low cost, low clamp-ing voltage, fast response time, andsmall SOD-523 package of theESD5Z2.5T1 family of TVS diodesmakes this an ideal device for ESD andEMI protection in space constrainedproducts. Keypad and panel switchesinterface circuits are examples of appli-cations that can benefit from the fea-tures of the ESD5Z2.5T1.
ESD and conducted EMI surge volt-ages are a major design concern fordesigners because the surges can dis-turb the operation of the system, producepermanent damage, or cause latentdamage that will eventually cause a fail-ure. In particular, the latent damage ofsurge pulses is difficult to detect becausethe circuit may continue to operate foran extended period of time before acomplete failure occurs. Electronic prod-ucts are typically connected to othersystems through input/output (I/O)cables; thus, the I/O interface hasbecome a major source and entry pointfor both ESD and EMI. TVS devices arean effective tool to reduce field failuresand improve the reliability of a systemby suppressing high voltage surge puls-es at the I/O connector.
ESD ProtectionThe ESD immunity level can be speci-
fied by several different tests includingthe Human Body Model (HBM) andMachine Model (MM) specifications.The majority of IC bus transceivershave a HBM rating of 2 to 6 kV and MMrating of 100 to 200 V. In contrast, theESD5Z2.5T1 has a 16 kV HBM and a400 V MM specification. Although theHBM and MM tests provide a good rep-resentation of the typical ESD energylevels that occur during the assemblyand handling of the printed circuit board (PCB), the energy level of the test pulses is often too low to simulateESD events that occur in normal product usage.
The International ElectromechanicalCommission (IEC) 61000-4-2 specifica-tion is emerging as the preferred systemlevel test to measure the ESD immunityof a product. The IEC 61000-4-2 andHBM ESD specifications are designedto simulate the direct contact of a per-son to an object such as the I/O pin of a connector; however, the IEC test ismore severe than the HBM test. TheIEC test is defined by the discharge of a 150 pF capacitor through a 330 Ωresistor, while the HBM uses a 100 pFcapacitor and 1500 Ohm resistor. TheESD5Z2.5T1 TVS diodes have a 30 kVrating for both the contact and non-con-tact (air) IEC 61000-4-2 ESD tests. In
contrast, most IC data sheets provideonly a HBM and MM rating, and do notlist a rating for the IEC 61000-4-2 test.Figure 1 shows the low clamping volt-age output of the ESD5Z5.0T1 diode forthe IEC 61000-4-2 ESD test.
The surge ability of a diode is directlyrelated to its size. External TVS devicesare typically a factor of at least ten timeslarger than the size of IC ESD diodes.Since it is typically not practical for anIC to have large ESD diodes, the TVSdiodes will provide a higher level of pro-tection. Internal IC protection circuitsfunction well at preventing assemblyfailures; however, they often are inade-quate for ESD events that occur in nor-mal product usage. In addition, manyICs have an internal ESD protection cir-cuit that is designed to handle only afew transient events while TVS devicessuch as the ESD5Z2.5T1 provide immu-nity for an indefinite amount of surges.
EMI ProtectionEMI protection is important because
products must function in close proximi-ty to a wide range of other electronicdevices. These products must be capa-ble of operating without either becomingeffected by or adversely effecting theoperation of electronic devices. EMI’scommon entry points are the powersupply and data lines. The powersource is typically shared by a number
TRANSIENT PROTECTION
52 Power Systems Design Europe May 2005
POWER SUPPLY
Conventional DC-DC converters representthe highest performance
The increasing number and complexity of loads poses a problem to the power supply engineer.
Traditionally an off-line power supply would have one main high current output supplying
the core voltage to the microprocessor or microcontroller and auxiliary outputs
supplying I/O circuits and peripherals.
By Paul Greenland, National Semiconductor
The control loop would be closedaround the main high current out-put and auxiliary outputs would
be loosely regulated by means of aweighted-sum feedback network at theinput of a programmable shunt regula-tor. This approach is simple but the reg-ulation of the auxiliary outputs is poor,particularly if the secondary windingsare not tightly coupled to each other andthe main regulated output winding.Furthermore, there are cases where asecondary load is significant comparedto the main output and has a switchingcurrent waveform, for example, a harddisk drive. In such cases some form ofsecondary regulator is required.Fortunately there is a choice of post-reg-ulator; each form has its merits andsweet-spot of applicability.
Linear post-regulationLinear regulators are the simplest
option for post-regulation. However, theyare rarely employed at load currentsgreater than 2 A due to the impact onthe overall efficiency of the power sup-ply. For many years, a simple linearregulator such as the LM317 wasemployed with the secondary windingdelivering approximately 2.5 V of head-room at maximum load and minimum
line. Additional high frequency rejectionwas afforded by bypassing the adjustpin with a 10uF tantalum capacitor toreturn. Recently linear regulators suchas the LP38843, shown in figure 1, with an additional bias rail have been introduced.
These regulators are capable of verylow dropout provided an auxiliary or biasrail is available to power the drive for thepass transistor. Furthermore, they arestable with multi-layer ceramic capaci-tors, which are favored for filtering highswitching frequency. Generally speak-ing, linear post-regulation offers a fastload transient response, together withexcellent high frequency rejection. Alsoa typical monolithic linear post-regulatorusually includes overload and short-cir-cuit protection. These advantages haveto be traded off against efficiency andthe provision of heat sinks or printed cir-cuit board copper area for heat removal.
Magnetic Amplifier Post-regulationMagnetic amplifiers, shown in figure 2,
became popular in the 1980s for highpower post-regulation. A magneticamplifier is a coil wound on a core with arelatively square B-H characteristic. This assembly has two distinct operating
states. When the core is saturated, theimpedance of the winding drops close tozero allowing high current to flow unim-peded. In the unsaturated state, the coilacts as a high inductance capable ofsupporting a high voltage with negligiblecurrent flow.
In essence we have a “magneticswitch”. Regulation is achieved by eitherpassing current back through the mag-netic amplifier winding while the second-ary is not conducting, or applying a volt-age to a second winding on the magnet-ic amplifier core during the same portionof the main converter’s switching period.Each technique programs the volt*sec-onds that the magnetic amplifier willblock when the secondary next startsconducting. Magnetic amplifiers utilizeleading edge modulation, this techniqueis particularly beneficial in current-moderegulated power supplies, it ensures thatno matter how the individual outputloading varies, the maximum peak cur-rent seen in the primary always occursas the pulse is terminated. Magneticamplifiers are accurate and efficientpost-regulators. However, the extrainductive components, limited operatingfrequency, slow dynamic response,cumbersome short-circuit protection and
Multiple output power supplies arechallenging to design, particularly if theauxiliary rails represent a significant por-tion of the throughput power. Traditionallow current post-regulation techniques,such as master-slave and conventionallinear regulation are limited due to ineffi-
ciency or poor load regulation. Magneticamplifiers scale with the volt*secondblocking capability rather than currentand are complex to protect under over-load and short circuit conditions.Conventional DC-DC converters repre-sent the highest performance, at the
expense of component count. The syn-chronous secondary side switching post-regulator is the optimum solution for effi-cient high power auxiliary regulation.
www.national.com
Figure 4. Synchronous Secondary Side Post Regulation. Figure 5. SSPR Representative Waveforms.
POWER SUPPLY
54 Power Systems Design Europe May 2005
no load operation limit useful applica-tion. The European power supply manu-facturer Lambda, formerly calledCoutant, manufactured an innovativemodular AC-DC converter series thatemployed magnetic amplifier post-regu-lation before its re-emergence whensquare loop cobalt-enriched coresentered the market. The Coutant tech-
nique consisted of a constant volt*sec-ond primary converter feeding an arrayof secondary regulators with an inputvoltage head-room of approximately20%. There was no feedback across theprimary-secondary isolation barrier, onlyfault/diagnostic signals. This approachled to a significantly faster time-to-mar-ket for semi-custom AC-DC converters.
Switching Post-regulationSwitching regulation on the secondary
takes two forms, conventional switchingregulators and synchronous switchingpost-regulators. The former is usually aDC-DC converter with its own free-wheeling diode and LC output filter. Thisapproach, illustrated in figure 3, hashigh efficiency and accuracy. Outputvoltage regulation is independent of thepre-converter’s duty ratio.
DC-DC converters often have inde-pendent protection against output over-voltage and short-circuit, which willapply to the post-regulated secondary.On the downside, if steps are not takento synchronize the switching frequencyof the post-regulator with the main sup-ply, beating or interference may occur.The component count is higher with theadditional filter, synchronization circuitryand the extra rectifier.
The synchronous secondary side post-regulator (SSPR), shown in figure 4 isconsidered to be the best secondaryregulator for high power auxiliary outputs.
Its behavior is not unlike the magneticamplifier; consequently it is often calleda "solid state mag amp". The LM5115-based SSPR is synchronized to the pri-mary converter. It uses leading-edgepulse width modulation and is compati-ble with either current or voltage modecontrolled primary power source.Synchronization of the LM5115 comesfrom the main secondary winding of thepower transformer. A resistor dividersamples the pulsating power voltagewaveform creating a synchronizationsignal and setting an internal currentsource to create the PWM ramp. Thisaffects input voltage feed-forward, regu-lating against line voltage variation. TheLM5115 controls the buck power stagewith leading edge pulse width modula-tion holding off the high side driver untilthe necessary volt*seconds is estab-lished for regulation. Representativewaveforms are shown in Figure 5.
Bias to the part comes from a rectifiedpulse signal. Adaptive dead time controldelays the top and bottom drivers to avoidefficiency-sapping shoot through currents.
Figure 2. Magnetic Amplifier Post-regulator.
Figure 1. LP38843 Internal Schematic.
POWER SUPPLY
Figure 3. DC-DC Post-regulation.
56 Power Systems Design Europe May 2005 www.powersystemsdesign.com 57
Get your own copy of Get your own copy of
your trade journal for the power engineer!your trade journal for the power engineer!
FREE Subscription!FREE Subscription!
Sign up today online at www.powersystemsdesign.com
Sign up today online at www.powersystemsdesign.com
NEW PRODUCTS
Linear Technology’s LTC6101 high sidecurrent sense amplifier combines veryquick response times with an input volt-age range of 4V to 60V. The LTC6101can withstand voltages up to 70V, a valu-able feature when power supply failuresor catastrophic load changes can lead to
transient overvolt-age. With aresponse timeunder 1us, theLTC6101 is ideal forautomatic powershutdown underfault conditions.
The LTC6101 isdesigned to extracta small differentialsignal from a highcommon mode volt-age, and thenamplify and trans-late this into a
ground-referenced signal. Achieving thisfunctionality with precision has typicallyrequired a combination of discrete com-ponents and op-amps, differential ampli-fiers or instrumentation amplifiers. TheLTC6101 simplifies this task by requiring
only the LTC6101 ThinSOT package andtwo resistors. Not only is this part muchsmaller and simpler, it consumes only250uA.
The LTC6101 includes outstanding per-formance: input bias current is 170nAMax and input offset voltage is 300uVMax. Better than 1% accuracy can beachieved using precision gain resistors,allowing for excellent measurement reso-lution and the ability to accommodateinputs with large dynamic range.
For more information, contact:Doug Dickinson, Media RelationsManagerLinear Technology Corporation1630 McCarthy BoulevardMilpitas, CA [email protected]
60V High Side Current Sense Amplifier
www.linear.com
For variable or fixed speed control ofsingle-phase brushless DC fans, blow-ers and pumps, the ZXBM1015 pre-driv-er from Zetex Semiconductors includesa current monitor that enables supplycurrent on start-up and stall to be keptpermanently within OEM specifications.
The ZXBM1015 produces a PWM out-put to drive an external H-bridge that inturn drives the motor windings. Its inte-gral current monitoring circuit measuresthe voltage across a low value external
resistor in the low sideof the H-bridge anddeduces how much cur-rent the motor is takingduring each PWM cycle. If it takes toomuch, the pre-driverbacks off the PWM drive to reduce the cur-rent drawn.
The pre-driver IC isalso the first of its kindto offer a configurablephase commutationdelay, which allows
OEMs to accurately meet the require-ments of different motor sizes and sofurther optimise efficiency. TheZXBM1015 has a fixed internal commu-tation delay of 100µs, which can be eas-ily extended or reduced by the simple addition of an external resistor.
The flexibility of the ZXBM1015 solu-tion is demonstrated by its compatibilitywith thermistor, voltage and PWM control inputs, which support fully
adjustable motor speed control in a widerange of equipment, from desktop com-puters to central heating systems.
With a built-in Hall amplifier, the pre-driver is compatible with any type of Hall-effect sensor and can provide asa result, pulsed rotor speed and lockedrotor alarm outputs. For a locked rotorcondition, the ZXBM1015 puts motor driveoutputs into a safe operating mode toprotect the H-bridge and motor windings.Visit Zetex at PCIM Europe Booth 12- 208
For further information and readerenquiries: Lin CollierZetex Semiconductors plcLansdowne RoadChaddertonOldham OL9 9TYUnited Kingdom
Tel: +44 (0)161 622 4444Fax: +44 (0)161 622 4469E-mail: [email protected]
Single-Phase Motor Pre-Driver
www.zetex.com
Tyco Electronics Power Systemsannounced the PIM200, the industry’sfirst power input modules designed tocomply with the Advanced TelecomComputing Architecture (AdvancedTCA)specifications, providing TelecomEquipment Manufacturers a completepower solution for their AdvancedTCAboards. As a packaged solution, thesestandard, off-the-shelf modules greatly
reduce product time tomarket, thus meetingthe needs of telecomservice providers in amore efficient manner.The packaged solutionalso reduces productcomplexity as a “one-stop shop” for managingall functions required byAdvancedTCA stan-dards.PIM modules are ratedfor 200W of maximumoutput power with a
wide input voltage range of –38Vdc to -75Vdc, suitable for both –48Vdc and–60Vdc power distribution. The PIMmodules incorporate all of the powermanagement features required by theAdvancedTCA PICMG 3.0 specification.Some features of the modules are:Inrush Current Protection; ORingFunctionality with MOSFETs; EMI filter-ing meeting Class B CISPR Limits; Hot
Plug capabilities; A/B Feed loss alarm;8W of Auxiliary power for IPMI andhouse keeping functions and Inputundervoltage, overvoltage, overcurrentand thermal protectionsIn addition, the PIM200 modules providea charging current at 72Vdc for externalholdup/energy storage bulk capacitorsto reduce real estate on telecom boards.Sized at 2.78 in x 1.45 in x 0.5 in (70.6mm x 36.8 mm x 12.7 mm), the moduleshave a high efficiency of 97% at–48Vdc. When combined with TycoElectronics’ isolated DC/DC and busconverters and point of load modules,the PIM200 modules provide elegantpowering solutions while complying withAdvancedTCA board power require-ments, thus enabling automated andhighly efficient board-level managementcapabilities.Visit Tyco at PCIM Europe Booth 12- 421
www.power.tycoelectronics.com
Power Input Modules
NEW PRODUCTS
Vis i t us atHal l 12
Booth 239
58 Power Systems Design Europe May 2005
Unlike traditional packaging materials,A family of -150-V and -200-V p-channelpower MOSFETs that offers space-savingsolutions for active clamp configurationswere released by Siliconix / Vishay .
While p-channel MOSFETs for activeclamps offer a simpler alternative to n-channel clamp implementations, until nowthe majority of devices suitable for thisapplication were available only in largerpackages such as the SO-8 and DPAK.With the new devices in the SOT-23 andSC-70, Siliconix brings together small
size and on-resistance values at least80% better than the nearest competingdevices.
Built on Siliconix’s advanced p-channelTrenchFET technology, the compactdimensions of these new p-channelpower MOSFETs reduce board spacerequirements to enable smaller overallproduct design without sacrificing per-formance. Because p-channel drive cir-cuitry is less complicated than that of n-channel solutions, these new devices alsoallow designers to implement smaller,lower-cost active clamp designs in smalland intermediate-sized power converters.
The power MOSFETs released todayare intended for primary-side active clampcircuits in dc-to-dc converters for telecom,datacom, and industrial products. Withon-resistance values ranging from 1.2ohms to 2.35 ohms, these powerMOSFETs represent industry-best on-resistance for their respective footprints.
The Company's facilities include a com-pany-owned Class 1 wafer fab dedicatedto the manufacture of power products inSanta Clara, California, and a Class 1wafer fab located in Itzehoe, Germany uti-lized under a lease arrangement. Vishayis headquartered in Malvern, Pennsylvania,and has operations in 17 countriesemploying over 25,000 people. Visit Vishay at PCIM Europe Booth 12- 547
150-V and -200-V p-channel Power MOSFETs
www.vishay.com
NEW PRODUCTSNEW PRODUCTS
International Rectifier introduced theIR5001S universal high-speed con-troller/N-channel power MOSFET driverfor high-performance, active ORing cir-cuits. The active ORing IC, housed in anSO-8 package, is used with an externalMOSFET, replacing traditional diode ORingto increase efficiency and reduce powerdissipation. Active ORing is a key require-ment for many high-end systems requir-ing maximum up-time, such as carrier-class communication equipment and
telecom and datacom system servers.Active ORing combines two or more
power sources to create a redundantpower source, preserving the inputpower supply in case one of the inputsources fails. In the event of an inputpower failure, the active ORing circuitdisconnects the non-functioning powersource as quickly as possible to preventthe system bus voltage from falling, andto prevent large peak reverse currents.
Technical highlightsThe IR5001S active ORing controller
IC is suitable for a wide range of activeORing circuits, including -48V/-24V inputactive ORing for carrier-class communi-cation equipment, 24V/48V output activeORing for redundant
AC-DC rectifiers, 12V output activeORing for multiple-output DC-DC andAC-DC power, and for low voltage out-put redundant VRM DC-DC processorpower. In 12V output systems, ORing
circuits capable of handling currents of100A can be made using four IRF6609DirectFET MOSFETs in parallel. TheIR5001S can also be used in reversepolarity applications for 48V/24V sys-tems, replacing large D2Pak style diodesand an expensive relay.
Visit IR at PCIM Europe Booth 12- 201http://www.irf.comIR response management:Mike Wigg, International Rectifier Co Ltd, European Regional Centre 439/445 Godstone RoadWhyteleafeSurrey. CR3 0BLTel: +44 (0)20 8645 8003Fax: +44 (0)20 8645 8077e-mail: [email protected]
www.irf.com
Universal Active ORing Controller IC
59www.powersystemsdesign.com
www.powersystemsdesign.com 6160 Power Systems Design Europe May 2005
Texas Instruments introduced a 4mmx 4mm DFN package version of thehigh-precision OPA277 operationalamplifier from the company’s Burr-Brownproduct line. The OPA277 features anexcellent combination of offset, drift and
noise characteristics. This leadless,near-chip-scale package is less than1mm thick and features contacts on onlytwo sides, thereby minimizing boardspace requirements and enhancingthermal characteristics through anexposed pad. The DFN package can beeasily mounted using standard PCBassembly techniques. (Seewww.ti.com/sc05093)
The OPA277’s reduced package sizeallows compact designs in a variety ofapplications including industrial elec-tronics, temperature measurement, bat-tery-powered instruments and testequipment, and helps preserve preci-
sion by keeping the chip temperaturelow as the power is easily dissipatedthrough the exposed pad. Additionally,the OPA277 features ultra low offsetvoltage (35uV typical) and drift, low biascurrent, high common-mode rejectionand higher power supply rejection. Thedevice also offers high open-loop gainand a wide supply range (to +/-18V).
For more information on TI´s completeanalog design support, and to downloadthe latest Amplifier Selection Guide, visitwww.ti.com/analog.Visit Texas Instruments at PCIMEurope Booth 12- 229
High-Precision Industrial Op Amp
www.ti.com
To meet the growing needs of manu-facturers who must meet worldwidelead-free manufacturing requirements,Littelfuse has introduced lead-free ver-sions of its 2AG, 3AG, 3AB and 5 x20mm form factor fuses.
The company's goal is to allow end-product manufacturers to comply withemerging worldwide lead-free initiativessuch as Europe's Restriction of the useof Certain Hazardous Substances inElectronic and Electronic Equipment,
(RoHS) and the Japanese ElectronicsIndustry Development Association lead-free product roadmap.
The new fuses also comply with elec-tronic industry manufacturing standardsfor lead-free circuit board content includ-ing IEC 60068-2-20 as well as the EIA-IS-722 (Low Voltage Supplemental FuseQualification Specification), CSA and UL approvals.
Designed to withstand the higher tem-peratures associated with lead-free
wave soldering, the new fuses offer thesame performance specifications as the company’s conventional 5 x 20mm,2AG, 3AG and 3AB fuses, but in a lead-free package.
5 x 20mm series are currently avail-able in full production quantities. 2AG,3AG, and 3AB series are available forsampling and will be available June 2005.
www.littelfuse.com
Lead-Free Fuses
NEW PRODUCTS
Companies in this issue
Company Page Company Page Company Page
ABB Switzerland . . . . . . . . . . . . . . . . .C3ABB Switzerland . . . . . . . . . . . . . . . . . . .1Advanced Power Technology . . . . . . .57Allegro MicroSystems . . . . . . . . . . . . . . .1Anagenesis . . . . . . . . . . . . . . . . . . . . . . .1Analog Devices . . . . . . . . . . . . . . . . . . . .1Ansoft . . . . . . . . . . . . . . . . . . . . . . . . . .13Ansoft . . . . . . . . . . . . . . . . . . . . . . . . . . .1APEX Microtechnology . . . . . . . . . . . .45Artesyn Technologies . . . . . . . . . . .1,4,36Bergquist . . . . . . . . . . . . . . . . . . . . . . .15CT-Concept Technology . . . . . . . . . . .23CT-Concept Technology . . . . . . . . . . . . .1Cree . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6Danfoss . . . . . . . . . . . . . . . . . . . . . .58,59ecpe . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4Epcos . . . . . . . . . . . . . . . . . . . . . . . . . .35eupec . . . . . . . . . . . . . . . . . . . . . . . . . .41eupec . . . . . . . . . . . . . . . . . . . . . . . . . . . .1Fairchild Semiconductor . . . . . . . . . .C2Fairchild Semiconductor . . . . . . . . . . . . .1
Ferraz Shawmut . . . . . . . . . . . . . . . . . .55Fuji . . . . . . . . . . . . . . . . . . . . . . . . . . . .29Hitachi . . . . . . . . . . . . . . . . . . . . . . . . . .37Infineon . . . . . . . . . . . . . . . . . . . . . . . . . .1International Rectifier . . . . . . . . . . . . .C4International Rectifier . . . . . . . . . . . . . . . .1iSuppli . . . . . . . . . . . . . . . . . . . . . . . . . .14IXYS . . . . . . . . . . . . . . . . . . . . . . . . . . .39LEM Components . . . . . . . . . . . . . . . . .3LEM Components . . . . . . . . . . . . . . . .1,16Littelfuse . . . . . . . . . . . . . . . . . . . . . . . .60Linear Technology . . . . . . . . . . . . . . . . .5Linear Technology . . . . . . . . . . . . . . .1,57Maxwell . . . . . . . . . . . . . . . . . . . . . . . . .49Maxwell . . . . . . . . . . . . . . . . . . . . . . . . . .4Micrel . . . . . . . . . . . . . . . . . . . . . . . . . . .9Microsemi . . . . . . . . . . . . . . . . . . . . . . .47Mitsubishi . . . . . . . . . . . . . . . . . . . . . . .11National Semiconductor . . . . . . . . . . .1,52Ohmite . . . . . . . . . . . . . . . . . . . . . . . . .21Ohmite . . . . . . . . . . . . . . . . . . . . . . . . . . .1
On Semiconductor . . . . . . . . . . . . . . .1,50PCIM Europe 2005 . . . . . . . . . . . . . . . .53PCIM Europe 2005 . . . . . . . . . . . . . . . . .4PCIM Exhibitor Floor Plan . . . . . . . . . . .30Philips Semicondactors . . . . . . . . . . . . .16Power Integration . . . . . . . . . . . . . . . .17Power Systems Design Europe . . . . .57Sanrex . . . . . . . . . . . . . . . . . . . . . . . . . .51Semikron . . . . . . . . . . . . . . . . . . . . . . .18Semikron . . . . . . . . . . . . . . . . . . . . . .1,40ST Microelectronics . . . . . . . . . . . . . .25Spoerle . . . . . . . . . . . . . . . . . . . . . . . . .33SynQor . . . . . . . . . . . . . . . . . . . . . . . . . .1Texas Instruments . . . . . . . . . . . . . . . . .7Texas Instruments . . . . . . . . . . . . .1,44,60Tyco Electronics . . . . . . . . . . . . . . . . .22Tyco Electronics . . . . . . . . . . . . . . . . .1,56Vishay . . . . . . . . . . . . . . . . . . . . . . . . . .43Vishay . . . . . . . . . . . . . . . . . . . . . . . . . .59Zetex . . . . . . . . . . . . . . . . . . . . . . . . . . .56
Please note: Bold—companies advertising in this issue
