VOICE OF THE ENGINEER - ElectronicsAndBooks

VOICE OF THE ENGINEERFEB3Issue 3/2011www.edn.comA new twist ondesign chain Pg 50The design-to-costimperative andcustomer value Pg 10Signal Integrity:Take the fifth Pg 17Design Ideas Pg 43Tales from the Cube:Tower of babble Pg 54

YOUR SMART PHONE PERFORMANCE DEPENDS ONGLOBALFOUNDRIES. THAT’S WHY THE WORLD’S LEADING HANDSETCHIP SUPPLIERS DEPEND ON US FOR A WIDE RANGE OF DEVICESSUCH AS POWER MANAGEMENT IC’S ON OUR AWARD-WINNING200mm 0.18μm TECHNOLOGY, TRANSCEIVERS AND WIRELESSCONNECTIVITY IC’S ON BOTH 200 AND 300mm TECHNOLOGIES,AS WELL AS BASEBAND AND APPLICATION PROCESSORSAT 45 AND 40nm, WITH NEW DESIGNS UNDERWAY AT 28nm.GLOBALFOUNDRIES DELIVERS. The success of the next generation of mobile productsdepends on their ability to deliver PC-like performance,a highly integrated rich media experience and longerbattery life. GLOBALFOUNDRIES presents a revolutionarysystem-on-chip (SoC) platform, enabling breakthroughchip performance with lowest cost and lowest power.This significant platform for mobile products is enabled byGLOBALFOUNDRIES’ leadership in a wide array of foundrytechnologies from our award-winning 200mm technologies,to leading edge 28nm semiconductor technology.GLOBALFOUNDRIES MANUFACTURES THEFULL ARRAY OF HANDSET IC’S SUCH AS:Power Management ICs made that integrateRF components, High voltage BCD transistors,and non-volatile memory in a single chip.Slim modems and RF IC’s integratedwith baseband processors.Connectivity peripheral chips that integrateWiFi, Bluetooth, FM, GPS …Application processors that power theworld’s smart phones and tablets.With its 28nm platform, GLOBALFOUNDRIES not only leadsthe foundry industry, but is leaping ahead of the world’slargest semiconductor integrated device manufacturer,providing an advanced SoC platform for developersthat will deliver new performance levels and ease-of-usefeatures far ahead of the competition. The transition to this28nm technology platform is an inflection point for wirelesstechnology and products.From baseband and application processors, to transceivers,power management, bluetooth, WiFi, GPS and much, muchmore, GLOBALFOUNDRIES is revolutionizing next generationmobile products.www.globalfoundries.com/mwc

VOICE OF THE ENGINEERFEB3Issue 3/2011www.edn.comA new twist ondesign chain Pg 50The design-to-costimperative andcustomer value Pg 10Signal Integrity:Take the fifth Pg 17Design Ideas Pg 43Tales from the Cube:Tower of babble Pg 54ADVANCESIN ENERGY-STORAGETECHNOLOGYPOWERWIRELESSDEVICESPage 26UNDERSTANDINGEMBEDDED-SYSTEM-BOOTTECHNIQUESPage 18LOWERING THE COST OFMEDICAL-IMAGING R&DPage 36

Agilent InfiniiVision 7000B Series (MSO and DSO models) Bandwidth: 100 MHz - 1 GHz40% bigger screen.Tektronix 4000 Series475,000 times faster…Yes, really.Waveform updaterate wvfm/s100,00080,00060,00040,00020,0000Agilent MSO7104BTektronix MSO4104Waveform Update Rate Comparison95,00050,000Baseline: Analogchannels only,timebase20 ns/div.60,0002,300Changetimebaseto 10 ns/div95,000Turn onMSOchannels35Enabledeepmemory0.2Turn onserialdecodeNotes: · Measurements taken on same signal using Agilent MSO7104B and Tektronix MSO4104 .· Memory depth = display window X sample rate with up to 8 Mpts on Agilent.· Tektronix measurements taken with version 2.13 firmware.· Screen images are actual screen captures, and scopes are shown to scale.12595,00095,000A scope’s real world performance depends on morethan just bandwidth and sample rates. It takes anultra fast update rate to capture infrequent events anddisplay the most accurate signal detail. That’s whereAgilent 7000B Series scopes really stand out. Thanksto a proprietary chipset, you get faster update ratesnot only in analog mode, but in every mode. That’swhy you see greater signal detail. That’s Agilent.© 2010 Agilent Technologies, Inc.Agilent and ourDistributor NetworkRight Instrument.Right Expertise.Delivered Right Now.866-436-0887www.metrictest.com/agilentAre you using the best scope?Take the 5-minute Scope Challenge.Visit www.metrictest.com/agilent/scope.jsp

2.3.11contentsAdvances inenergy-storagetechnology powerwireless devices26Energy harvesting relieson intermittent energysources and requiresenergy storage, such as a capacitoror a battery. New improvementsin these components point towardsmaller, longer-lived wireless sensornodes, but a long-lived primary batterymay be your simplest energysourcechoice.by Margery Conner, Technical EditorUnderstandingembedded-systemboottechniquesThe boot-up process is simple18 in theory but often complexin reality. The main job of a bootloader is to load the operating system,but software and hardwareengineers view this process in differentways.by Mohit Arora and Varun Jain,Freescale SemiconductorSENSORSORARRAYRAYADCMULTI-CHANNELADCFPGADATA-ACQUISITION ACQUISITION SUBSYSTEMADCMULTI-CHANNELADCMULTI-PLEXERSLIP RINGBUFFERSEDESDATA-AGGREGATION CpulseDilbert12 DSP-based analyzers boast85-MHz bandwidth14 Wireless-power technology,development kit target Chevy Volt14 High-voltage gate-driver ICsimprove noise immunity15 Class D amplifier suppressesEMIDESIGNIDEASV 112VV IN V OUTIC 1LM317TV ADJR 472IC613 51 10kR 2 41415 Freescale unveilsCortex-A9-based devices16 Step-down dc/dc controllerfeatures fast transient response16 Solid-state drives are one-eighththe size of their siblings16 Tool performsFPGA-power designs43 Compute a histogram in an FPGA with one clockLowering the cost ofmedical-imaging R&DMedical-equipment companiesare shifting from a verti-36cal-integration model to a systemintegrationmodel to do more withless research and development.by Allan Evans,Samplify Systems IncCOVER: COMPOSITE ILLUSTRATION BY TIM BURNS.VAULT: MARK EVANS/ISTOCKPHOTO.COM;CIRCUIT PATTERN: MON5TER/ISTOCKPHOTO.COM44 Protect MOSFETs in heavy-duty inductive switch-mode circuits46 Control an LM317T with a PWM signal47 High-speed buffer comprises discrete transistors48 Limit inrush current in high-power applicationsFEBRUARY 3, 2011 | EDN 7

Improving RF Transmission EfciencyTiny RF PA Power Management ICOUTPUT VOLTAGE (V)4.54.03.53.02.52.01.51.00.5DC/DC Output Voltagevs. DAC VoltageV IN= 4.2VI OUT= 100mAV OUT= V DACx 3Bypass ModeDACEN1EN2MIC28082.0mm x 2.5mmRF PAPower SupplyRF TransceiverPA Reference00.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5DAC VOLTAGE (V)RF Power amplifiers used in cell phoneapplications are typically powered directly from thebattery. This creates a challenge for the designers:as RF devices, the power amplifier adjusts thevoltage and current based on the transmission.Inefficiencies are created when the PA supplyvoltage and the output voltage are different.The MIC2808’s DAC controlled adjustable voltageoutput feature minimizes the input to outputvoltage differential, thus improving efficiency andbattery life. The voltage is adjusted based on realtimeneeds of the PA to achieve the most efficientoperation at any given transmit power level.For more information, contact your local Micrelsales representative or visit Micrel at: www.micrel.com/ad/mic2808.Ideal for use in: CDMA2000 mobile phones UMTS/WCDMA mobile phones Power amplier modules (PAMs) with linear PAs WiMAX/Wibro modules WiFi modules DECT handsets© 2010 Micrel, Inc. All rights reserved.Micrel and Innovation Through Technology are registered trademarks of Micrel, Inc.www.micrel.com

contents2.3.11Simpler PowerConversion5254IR’s SmartRectifier chipsetfor AC-DC power convertersdramatically simplifies designand improves efficiency.DEPARTMENTS & COLUMNS10 EDN.comment: The design-to-cost imperative and customer value17 Signal Integrity: Take the fifth50 Supply Chain: A new twist on design chain:EBV collaborates to launch chips52 Product Roundup: Passives54 Tales from the Cube: Tower of babbleonline contentsONLINE ONLYCheck out these Web-exclusive articles:Photovoltaic market to see sunny 2011Photovoltaic solar installations are expectedto rise in 2011, but rollbacks in other countriesadd clouds to iSuppli’s positive outlook.➔www.edn.com/110203tocaEnergy Star turning into black hole,technology companies fearAlthough IT manufacturers have long supportedand would like to continue supportingEnergy Star, additional time and moneyrequirements may drive some companies outof the program.➔www.edn.com/110203tocbHOT 100EDN proudly presents its list of the Hot100 products that in 2010 heatedup the electronics world andgrabbed the attention of oureditors and our readers.➔www.edn.com/110203toccwww.edn.comWANTED: Design IdeasR 184 +180kμFD 11N4148R 17 R 190k 100kR 206.8kD 283+ IC 42 LM358–41Want to see yourwork featuredin EDN? Submityour own DesignIdea—a short,compact article that helps solve problemsor shows innovative ways to accomplishdesign tasks. We review articles that fallinto virtually any technology area: analogor digital circuits, programmable logic,hardware-design languages, systems,programming tips, useful utilities, testtechniques, and so on. The idea should beuseful, innovative, or tricky. And if we acceptyour Design Idea, we pay you $150.To learn what makes a good DesignIdea entry, spend some time browsingour archive of Design Ideas at www.edn.com/designideas. For information onhow to submit a Design Idea, see theDesign Ideas Writers’ Guide: www.edn.com/110203tocd.SmartRectifier ICsPartNumberPackageIR1166SPBFIR1167A IR1167B IR1168 IR11662 IR11672A IR11682SPbF SPbF SPbF SPbF SPbF SPbFSO-8V CC(V) 20V FET(V) 200Sw Freq.max (kHz)500 400Gate Drive+1/-4±(A)+2/-7 +1/-4 +1/-4 +2/-7 +1/-4V GATEClamp (V)10.7 10.7 14.5 10.7 10.7 10.7 10.7Min. OnTime (ns)Program. 250 -3000 750 Program. 250 -3000 850Enable Pin Yes Yes Yes No Yes Yes NoChannel 1 2 1 2AutomaticMOTProtectionNo No No No Yes Yes YesFor more information call1.800.981.8699 or visitwww.irf.comEDN® (ISSN#0012-7515) is published semimonthly, 24 times per year, by UBM Electronics, 11444 W. Olympic Blvd., Los Angeles, CA 90064-1549; 310/445-4200; FAX 310/445-4299. Periodicals postage paid at Los Angeles, CA, and at additional mailing offices. SUBSCRIPTIONS—Free to qualified subscribers as defined on the subscription card. Rates for nonqualified subscriptions, including all issues: US, $150 one year;$250 two years; $300 three years. Except for special issues where price changes are indicated, single copies are available for $10 US and $15foreign. For telephone inquiries regarding subscriptions, call 763/746-2792. E-mail: EDN@kmpsgroup.com. CHANGE OF ADDRESS—Noticesshould be sent promptly to PO Box 47461, Plymouth, MN 55447. Please provide old mailing label as well as new address. Allow two monthsfor change. NOTICE—Every precaution is taken to ensure accuracy of content; however, the publisher cannot accept responsibility for the correctnessof the information supplied or advertised or for any opinion expressed herein. POSTMASTER—Send address changes to EDN, POBox 47461, Plymouth, MN 55447. Canada Post: Publications Mail Agreement 40612608. Return undeliverable Canadian addresses to PitneyBowes Inc, PO Box 25542, London, ON N6C 6B2. Copyright 2011 by United Business Media. All rights reserved. Reproduction in whole orpart without written permission is prohibited. Volume 56, Number 3 (Printed in USA).THE POWER MANAGEMENT LEADERVisit us at Booth 409Fort Worth Convention Center07-09 March 2011

EDN.COMMENTBY MIKE DEMLER, TECHNICAL EDITORThe design-to-cost imperativeand customer valueEngineers are constantly challenged to meet aggressive performancespecifications and to produce designs that don’t exceedmanufacturing cost targets. That approach may be insufficientto ensure product success, however. I hold this opinion because,early in my career as an IC designer for Texas Instruments, thecompany adopted a “design-to-cost” philosophy. J Fred Bucy,then TI’s president, was clearly intent on driving design-to-cost consciousnessinto the culture of the corporation. “The application of the design-tocostconcept must become second nature not only to those who design butalso to those who manufacture and market,” he said (Reference 1).Looking back at TI’s product strategyin 1977 (Reference 2), you’ll see thatthings really haven’t changed much.As a novice IC designer, the design-tocostimperative struck home for me duringwhat appeared to be a simple designproject: a thermal print-head driver.The driver that I designed (photo)required three chipswith a fixed printhead. With this approach,we could simultaneouslyburnthe dot matrix rowby row as the paperscrolled past.The only problemwas that thedesign-to-cost targetshad determinedthe chip size before I even started thedesign. With the best layout draftsmenthat we had working on it, we produceda beautifully symmetrical die. Unfortunately,the constrained chip area limitedthe size of the output drivers, and theon-resistance was too high to get theprint head hot.After I convinced my boss that therewas nothing more to squeeze out of thelayout, he let me in on a simple fix:Shrink it! The new fabrication processwas not yet in volume production, butit would meet the design-to-cost goals.All of these memories came back tome during the 2011 Consumer ElectronicsShow in Las Vegas last month. Multicoreprocessors for smartphones areamong the hottestnew IC-design and-application topics,and a battlefor leadership isunder way. Nvid-ia has laid claim to“the world’s mostadvancedmobileprocessor,” the Tegra2 (Reference 3),which it built fromthe ground up as a heterogeneous multicoreSOC (system on chip). Meanwhile,TI (Reference 4), Samsung, and Qualcommare also competing for next-generationdesign wins.Broadcom recently announced its second-generationdual-core BCM2157SOC (Reference 5). It integrates twoARM11 cores with an HSDPA (highspeed-downlink-packet-access)basebandprocessor and an Android applicationprocessor. With the dual-core design,Broadcom is targeting the low-endmarket for $100 Android smartphones.More than 30 years ago, TI pursued thesame strategy for market penetration.Let’s not get carried away with thisdesign-to-cost concept, though. It is alsocritical to keep in mind the customerside of the cost equation—that is, thevalue proposition: Value equals benefitdivided by cost. My second-generationthermal printer design wasn’t just lessexpensive; it also provided added benefitin higher reliability. Both Broadcomand Nvidia use the free Android operatingsystem but design to deliver valueto consumers—from low-end phones tohigh-end “superphones.” Through eachgeneration of Moore’s Law, the semiconductorindustry has provided more customervalue, lowering cost but also increasingbenefits.Too many companies in “mature”competitive industries blame customersor the tough economy for driving downprices. The lesson from the most successfulcompanies is to continually delivergreater value. Companies shouldfocus more on how their engineers candesign for value rather than obsess overbalance-sheet-driven cost-cutting strategies,in which layoffs and outsourcingare all too prevalent.EDNREFERENCES1 White, John A, Production Handbook,Wiley, 1987.2 Texas Monthly, July 1977, http://bit.ly/feXMDW.3 Demler, Mike, “CES Day-2: Nvidia,Motorola, and you say you want to jointhe press corps?” EDN, Jan 6, 2011,http://bit.ly/eXUH31.4 Demler, Mike “More from CES:Blackberry PlayBook enters the battleof the 4G tablets,” EDN, Jan 11, 2011,http://bit.ly/f2eNgy.5 “Broadcom Announces New AndroidPlatform to Enable Mass MarketSmartphones,” Broadcom, Dec 14,2010, http://bit.ly/fG8SZr.Contact me at mike.demler@ubm.com.10 EDN | FEBRUARY 3, 2011

ASSOCIATE PUBLISHER,EDN WORLDWIDEJudy Hayes,1-925-736-7617;judy.hayes@ubm.comEDITORIAL DIRECTORRon Wilson1-415-947-6317;ron.wilson@ubm.comMANAGING EDITORAmy NorcrossContributed technical articles1-781-869-7971;amy.norcross@ubm.comMANAGING EDITOR⎯NEWSSuzanne DeffreeElectronic Business, Distribution1-631-266-3433;suzanne.deffree@ubm.comSENIOR TECHNICAL EDITORBrian DipertConsumer Electronics,Multimedia, PCs, Mass Storage1-916-548-1225;brian.dipert@ubm.comTECHNICAL EDITORMargery ConnerPower Sources, Components,Green Engineering1-805-461-8242;margery.conner@ubm.comTECHNICAL EDITORMike DemlerEDA, IC Design and Application1-408-384-8336;michael.demler@ubm.comTECHNICAL EDITORPaul RakoAnalog, RF, PCB Design1-408-745-1994;paul.rako@ubm.comDESIGN IDEAS EDITORMartin Rowe, Senior Technical Editor,Test & Measurement Worldedndesignideas@ubm.comSENIOR ASSOCIATE EDITORFrances T Granville, 1-781-869-7969;frances.granville@ubm.comASSOCIATE EDITORJessica MacNeil, 1-781-869-7983;jessica.macneil@ubm.comCONSULTING EDITORJim Williams,Staff Scientist, Linear Technologyedn.editor@ubm.comCONTRIBUTING TECHNICAL EDITORSDan Strassberg,strassbergedn@att.netNicholas Cravotta,editor@nicholascravotta.comRobert Cravottarobert.cravotta@embeddedinsights.comCOLUMNISTSHoward Johnson, PhD, Signal ConsultingBonnie Baker, Texas InstrumentsPallab Chatterjee, SiliconMapKevin C Craig, PhD, Marquette UniversityLEAD ART DIRECTORMarco AguileraASSOCIATE ART DIRECTORTim BurnsPRODUCTIONMichael Ciardiello,Director of Premedia TechnologiesJeff Tade, Production DirectorBrian Wu, Publications Production ManagerJeff Polman, Derric Treece,Senior Production ArtistsWilliam Baughman, Lucia Corona,Ricardo Esparza,Production ArtistsEDN EUROPEGraham Prophet,Editor, Reed Publishinggprophet@reedbusiness.frEDN ASIAWai-Chun Chen,Group Publisher, Asiawaichun.chen@ubm.comKirtimaya Varma,Editor-in-Chiefkirti.varma@ubm.comEDN CHINAWilliam Zhang,Publisher and Editorial Directorwilliam.zhang@ubm.comJeff Lu,Executive Editorjeff.lu@ubm.comEDN JAPANKatsuya Watanabe,Publisherkatsuya.watanabe@ubm.comKen Amemoto,Editor-in-Chiefken.amemoto@ubm.comEXECUTIVE OFFICERSPaul Miller,Chief Executive OfficerDavid Blaza,Vice President, UBM ElectronicsKaren Field,Senior Vice President, ContentFred Gysi,Chief Financial OfficerMike Deering,Chief Technology OfficerStephen Corrick,Vice President/Executive DirectorKevin O’Keefe,Senior Vice President, Events DivisionRoger Burg,Vice President, OperationsJason Brown,Vice President, E-mediaEDN. 33 Hayden Avenue, Lexington, MA 02421. www.edn.com. Subscriptioninquiries: 1-763-746-2792; EDN@kmpsgroup.com. Address changes: Sendnotice promptly to PO Box 47461, Plymouth, MN 55447. Please provide an oldmailing label as well as your new address. Allow two months for the change.UBM Electronics, 11444 W. Olympic Blvd., Los Angeles, CA 90064-1549;1-310-445-4200; fax: 1-310-445-4299.MAXIMUMPRECISIONFrom a full spectrum of pin receptacles

pulseEDITED BY FRAN GRANVILLEINNOVATIONS & INNOVATORSDSP-based analyzers boast85-MHz bandwidthWith the introduction of the RSA5000series signal analyzers, Tektronix hassignifi cantly expanded its spectrumandvector-signal-analysis portfolio. The seriestargets use in design and operations applications,including spectrum management, radar,electronic warfare, radio communications, andEMC (electromagnetic compatibility).As RF signals become more complex andthe use of the ISM (industrial/scientifi c/medical)band becomes more common, designers,engineers, and operators must reliably and effi -ciently discover transient spectral phenomenathat digital-RF circuits can create over widerbandwidths. Traditional signal analyzers cannottrigger on transient problems, and midrangeunits’ maximum acquisition bandwidth until nowhas topped out at 40 MHz. Featuring advancedtime, amplitude, and DPX (digital-phosphortechnology)-triggeringfunctions, as well as aswept-DPX mode, the RSA5000 series enablesyou to discover and capture intermittent andrapidly changing signals at bandwidths as highas 85 MHz. This bandwidth covers the entireISM band, which is home to such commontechnologies as Bluetooth, Zigbee, RF identification, and wireless LANs.The series’ DPX-based live RFspectrumdisplay enables theanalyzers to quickly detect previouslyunseen signal behavior andimproves test confi dence by catchingtransients as brief as 5.8 μsec.The swept-DPX engine can also collectas many as 292,000 spectrumupdates/second over bandwidthsas high as 85 MHz and can sweepthe DPX display across instrumentinputs that range to 6.2 GHz.To capture transients for analysis,the analyzers offer frequency-mask, frequency-edge,density, time-qualifi ed, and runttriggering. You can also use cross-domain triggeringamong multiple instruments to isolatehard-to-fi nd hardware and software anomaliesand capture in deep memory seamless 7-secondtime records of 85-MHz-bandwidth RFsignals. The series lets you analyze captureddata in any domain at any time with correlatedmarkers. Automatic pulse measurement anddetection support multiple measurements onthe same set of captured data, which simplifies testing; reduces test time; and, by providingone instrument that replaces multiple testsets, reduces test cost.The analyzers offer 17-dBm third-orderintercept and −154-dBm/Hz DANL (displayedaverage-noise level) at 1 GHz and carrierreferredphase noise of −131 dBc/Hz at a10-kHz offset from the carrier and −150-dBmDANL at a 10-MHz carrier frequency. Pricesrange from $34,900 for a 1-Hz to 3-GHz unitwith 25-MHz bandwidth to $68,800 for a 1-Hzto 6.2-GHz unit with 85-MHz bandwidth.—by Dan Strassberg▷Tektronix, www.tektronix.com.TALKBACK“While open-loopuses of optocouplersare fun,useful, and profitable—meaningthey work wellfor many applications—iflots ofstability over longperiods of timeis required, thenit’s hard to beat aclosed-loop feedbacksystem, withthe optocouplerinside the loop.”—Engineer Steve Hageman,in EDN’s Talkback section, athttp://bit.ly/egx2kt. Add yourcomments.The RSA5000signal analyzers’dynamic rangerivals that of comparablypricedswept-frequencyspectrum analyzersbut offers doublethe RF bandwidth.12 EDN | FEBRUARY 3, 2011

Analog Devices: enabling the designs thatmake a difference in people’s lives.DSPRF TransceiverMetering ICADI’s family of innovative smart metersolutions offers greater control, communication,and monitoring capabilities.Measurement, processing, and communications solutions for intelligentenergy systems. Optimizing the supply and demand for electricity is a goal weall share. At ADI, we provide data converter, amplifi er, processor, and wired andwireless communications technologies to create leading-edge solutions fortoday’s energy market. Our products give consumers precise, real-time smartmeter information, enabling them to reduce energy consumption; utilities theability to accurately monitor grid conditions, increasing system effi ciency;and energy producers the ability to optimize solar and wind power generation,reducing fossil fuel dependence. For over 40 years, ADI has brought innovationand imagination to critical designs, including those that make our energysensitiveworld a better place. Learn more at www.analog.com/energy.www.analog.com

pulseWireless-power technology,development kit target Chevy VoltGeneral Motors (www.gm.com) hopes tosoon establish theplug-in hybrid Chevy Volt asa technology leader in theeyes of the “green”-car-buyingpublic. Toward that end,GM Ventures (www.gm.com/ventures), the company’s venture-capitalsubsidiary, hasinvested $5 million in Israeli wireless-power-technologystart-upPowermat. The fi rst GM deploymentof the Powermat technologywill be a wireless chargingstation in the 2012 Volt, andthe technology will eventuallyroll out to the Chevy, Buick,GMC, and Cadillac brands. GMannounced the development atlast month’s Consumer ElectronicsShow in Las Vegas, butno demos took place.Wireless power chargingis an apt application for consumerdevices in a car’s cabinbecause, in space-constrainedapplications, you’d like to avoidmyriad chargers and danglingwires. On the other hand, thePowermat technology requiresa receiver for each device, soyou’re basically replacing yourcharger and dangling wire witha $40 receiver, which looks likea rigid cell-phone cover.Powermat currently supportsall iPhone models; the second-and third-generation iPodTouch; all docking iPods; HTCDILBERT By Scott AdamsEvo 4G and HD2; MotorolaDroid X; Nintendo DS Lite andDSi; and Blackberry Tour 9630,Bold 9000/9650/9700 series,Curve 8300/8500/8900/9300series, and Pearl.Powermat is not the onlywireless-charging technologyavailable. The WirelessPower Consortium (www.wirelesspowerconsortium.com), of which Texas Instrumentsis a founding member,last summer released Qi, itsown version of an industry standard.The consortium basedits approach on Fulton Technology’s(www.fultontechnology.The Qi-certifiedbqTesla developmentkit (right) helps developersdesign wirelesspower for plug-inhybrid vehicles, suchas the Chevy Volt(below).com) eCoupled technology.TI last month released what itcalls the industry’s fi rst Qi-certifieddevelopment tools andchip set for wireless power.The $499 bqTesla developmentkit includes the bq500110wireless-power-transmittermanager; the single-input, 5Vbq25046 power-supply IC; andthe MSP430bq1010 wirelesspower-controland -communicationsmicrocontroller.—by Margery Conner▷Powermat,www.powermat.com.▷Texas Instruments,www.ti.com.HIGH-VOLTAGEGATE-DRIVER ICsIMPROVE NOISEIMMUNITYFairchild Semiconductorrecently introduced aseries of high-voltagegate-driver ICs, includingthe two-input, twooutput,high- and low-sideFAN7392 with shutdown;the one-input, two-outputFAN7393 half-bridge withshutdown and controllabledead time; the one-input,two-output FAN73932half-bridge with shutdownand fixed dead-time control;and the two-input,two-output FAN73933half-bridge with controllabledead time. The partssuit use in industrial applications,such as motordriveinverters, distributedpower supplies, andtelecom-system powersupplies.The devices featurea common-mode dV/dtnoise-canceling circuitthat enables stable operationof the high-voltagegate driver under highnoisecircumstances,as well as a level-shiftcircuit that offers highsidegate-driver operationwith negative-floatingsupply-return-voltageswings as high as −9.8Vat a high-side floatingsupplyvoltage of 15V. TheFAN739x series providesstable operation over atemperature range of −40to +125°C. The series alsofeatures floating channelsfor bootstrap operation to600V. The devices sell for$1.52 (1000) each.—by Paul Rako▶Fairchild Semiconductor,www.fairchildsemi.com.02.03.1114 EDN | FEBRUARY 3, 2011

Class D amplifier suppresses EMISilicon Laboratories recentlyintroduced a 5WstereoClass D amplifi erthat uses multilayer technologyto suppress EMI (electromagneticinterference), common inClass D devices, at its source.The Si270x amplifi er fi nds use ina range of price- and noise-sensitiveconsumer-audio products,including smartphone-dockingstations, tabletop radios,TV sound bars and monitors,boom boxes, and battery-poweredradios.Until now, two issues haveimpeded the adoption of ClassD amplifi ers: high EMI emissions,which interfere with AM/FM radio and smartphone operation,and the high cost of addingfi ltering and shielding forEMI-regulatory compliance.The Si270x addresses theseissues by having10 timesless radiated interference inthe EMI-compliance band, 100times less in the FM-radio band,and 1000 times less across theAM band than do other Class Dproducts.The amplifier also features2.5-times more play time thanClass AB-based systems anduses half the number of batteries.A consumer-audio systememploying the Si270x amplifier can provide as much as8.4 hours of play time usingfour AA alkalinebatteries.You can combinethe Si270xamp with theSi473x AM/FMradiotuner. Thelatest Si473xdevices offer astereo analoginput and internalADCs multiplexedwith the radio-tunerThe Si270x-A evaluation board accelerates application development of the Si270x Class D amplifier.Until now,high EMIemissions andcostly filteringhave impededthe adoption ofClass D amps.front end to support auxiliaryanalog-system inputs withoutadditional external ADCs.Samples and preproductionquantities of the Si270x amplifiers are now available in 24-pinQFN packages. Prices begin at$1.17 (10,000).To help accelerate applicationdevelopment, Silicon Labs offersaudio engineers the Si270x-A-EVB (evaluation board). The$325 board comprises a baseboardand a daughtercard. Itincludes a graphical user interfacethat runs on a standardPC and connects to the EVBthrough a USB (Universal SerialBus) interface.—by Paul Rako▷Silicon Laboratories,www.silabs.com.Freescale unveils Cortex-A9-based devicesCortex-A9 is currently the cream of the ARM (www.arm.com)-core crop, so it’s the next logical step inFreescale’s i.MX product progression. Freescale plansto begin making the first few members of its new i.MX6family available for sampling this year, with volumequantities in 2012. The family comprises the four-corei.MX6Quad, the two-core i.MX6Dual, and the singlecorei.MX6Solo. The i.MX6Quad and i.MX6Dual willappear first. Freescale is, at least initially, fabricatingthem on a common sliver of silicon, with two of the fourCPU cores disabled on the i.MX6Dual devices.Freescale isn’t yet releasing details on whose graphicstechnology it’s licensing for i.MX6; likely candidatesinclude Imagination Technologies’ (www.imgtec.com)popular PowerVP cores or ARM’s Mali multimediaengine. The devices include as many as four ARMCortex-A9 cores running at greater than 1 GHz per core;as much as 1 Mbyte of Level 2 system cache; and ARMVersion 7, Neon, VFP Version 3, and Trustzone support.A multistream-capable high-definition video engine delivers1080p60 decode, 1080p30 encode, and 3-D videoplayback in high definition. The devices also include2-D and Vertex acceleration engines and interface tostereoscopic image sensors for 3-D imaging.Interconnect is through HDMI (high-definitionmultimediainterface) Version 1.4 with integrated PHY(physical layer), SD (secure digital) 3.0, multiple USB(Universal Serial Bus) 2.0 ports with integrated PHY,GbE (gigabit Ethernet) with integrated PHY, SATA (serialadvanced-technologyattachment)-II with integratedPHY, PCIe (Peripheral Component Interconnect Express)with integrated PHY, MIPI (mobile-industry-processorinterface)CSI (camera serial interface), MIPI DSI (deviceintellectual-property interface), MIPI HSI (high-speedserial synchronous interface), and FlexCAN (controllerareanetwork) for automotive applications. The devicesalso support the VP8 codec, along with optional integrationof an e-paper display controller for e-readerapplications.—by Brian Dipert▶Freescale Semiconductor, www.freescale.com.02.03.11FEBRUARY 3, 2011 | EDN 15

pulseStep-down dc/dc controller featuresfast transient responseLinear Technology Corp’snew LTC3810H-5 synchronousstep-down dc/dc controller guarantees junctiontemperatures as high as150°C. The device employs aconstant on-time valley-current-mode-controlarchitecture,delivering low duty cyclesand fast transient responsewith accurate cycle-by-cyclecurrent limit without requiringa sense resistor. The controllercan directly step down voltagesfrom 60V to an output voltageranging from 0.8V to 93% of theinput voltage. The 1Ω onboardgate drivers minimize switchinglosses associated with drivingN-channel MOSFETs at highfrequency and high voltage foroutput currents as high as 25A.Applications include 48V powerconversion in telecom and datacom,as well as automotive andindustrial systems susceptibleto high surge voltages.The LTC3810H-5 featuresan adjustable minimum ontimethat you can set as low as100 nsec, enabling a high stepdownratio at high frequency.The operating frequency isselectable from 100 kHz to 1MHz, or you can synchronizeit to an external clock over thesame range. You can confi g-ure pulse-skipping operation tomaintain high effi ciency at lightloads. Other features includeprogrammable soft start ortracking, adjustable cycle-bycyclecurrent limit, output overvoltageprotection, a powergoodoutput signal, and programmableundervoltage lockout.The LTC3810H-5 comesin a 5×5-mm QFN-32 package,and prices start at $3.76 (1000)each.—by Fran Granville▷Linear Technology,www.linear.com/3810.Solid-state drives are one-eighththe size of their siblingsIntel recently announced the310 Series of m-SATA (miniserial-advanced-technologyattachment)solid-state drivesand is currently shipping them in$99 (1000), 40-Gbyte and $179,80-Gbyte versions. The drivesdeliver performance equivalentto that of the company’s comparable34-nm, MLC (multilevelcell)-flash-memory-based X25Intel’s m-SATA drives are one-eighth the sizeof the company’s 34-nm, MLC-flash-memorybasedX25 solid-state drives.solid-state drives but at oneeighththe size. They measure50.8×29.85 mm, are 4.85 mmthick, and weigh less than 10g.Their PCIe (Peripheral ComponentInterconnect Express) MiniCard-based mechanical interfacesare fully compatible with1.5- and 3-Mbps SATA protocols,including NCQ (nativecommandqueuing).Other features include sustainedsequential reads as highas 170 and 200 Mbytes/secand sustained sequential writesas high as 35 and 70 Mbytes/sec for the 40- and 80-Gbyteversions, respectively. Readlatency is 65 μsec for both versions,and write latency is 110and 75 μsec for the 40- and80-Gbyte versions, respectively.Random4-kbytereads are,respectively,as high as25,000 and35,000 IOPS(input/outputoperationsper second);respectiverandom 4-kbyte writesare 2500 and6600 IOPS. The devices havea life expectancy of 1.2 millionMTBF (mean time betweenfailures), a typical power consumptionof 150 mW in activemode and 75 mW in idle mode,and an operating-temperaturerange of 0 to 70°C.—by Brian Dipert▷Intel Corp,www.intel.com.Tool performsFPGA-power designsNational Semiconductor recently introduced theWebench FPGA-power-architect-design tool to modelpower supplies for FPGAs. The tool incorporates thedetailed supply requirements of more than 130 FPGAdevices from Altera (www.altera.com) and Xilinx (www.xilinx.com). Modern power-supply systems, includingadvanced FPGAs, are complex to design, often incorporatingmultiple unique loads to drive precisely specifiedvoltages. In addition to the required voltages and currents,each load may have limitations for ripple, noisefiltering, synchronization, and separation of supplies, aswell as start-up definitions.Webench delivers power-supply designs that incorporatethe comprehensive power requirements that FPGAmanufacturers publish, giving designers the confidencethat their power supplies will meet these constraintsand save time. To begin a design, a designer selects anadvanced FPGA from Altera or Xilinx, and the tool automaticallypopulates the unique power requirements,identifying core and I/O options for every potentialload of the array. Once the designer tunes the design,Webench collects all the loads and creates multiplepower-supply architectures as options for driving theFPGA and the total system.The power supply may involve one or more intermediatevoltage rails between the input supply and thepoint-of-load regulators. The designer can tune therecommended power supply with the turn of a dial, inseconds reducing size, increasing efficiency, or loweringcomplete system cost. The designer can also ordercomponents for prototyping; share the complete systemwith others; or easily print a complete project report,including schematics, BOM (bill of materials), and performancecharacteristics.—by Paul Rako▶National Semiconductor, www.national.com.02.03.1116 EDN | FEBRUARY 3, 2011

SIGNAL INTEGRITYBY HOWARD JOHNSON, PhDTake the fifthFourier analysis represents a perfect square wave as the infinitesum of harmonically related sine waves. The requisite sinewavecomponents comprise all the odd harmonics of the fundamentalsquare-wave frequency with their respective amplitudesset according to this pattern: 1, 1/3, 1/5, 1/7, and so forth. Evenharmonics are not required. The Fourier theory calls out an amplitudepattern that guarantees a sum, which—in the limit—approximatesa square-wave shape with amplitude π/4, or 0.78537316.Figure 1 illustrates the first few termsof the harmonic series. The first waveformis a simple sine wave at the fundamentalfrequency. The next waveformshows the fundamental plus its thirdharmonic, the next takes harmonicsto the fifth, and so on. The more termsyou add, the more “squarish” the waveformbecomes, but it still looks wiggly.Many engineers wonder how manyharmonic terms they must take to adequatelyrepresent a good square wave.In other words: What bandwidth does adigital transmission system need? Let’sexplore that question by first examiningFIRST ONLYFIRST AND THIRDFIRST, THIRD, AND FIFTHFOURIERAMPLITUDESFIRST: 1THIRD: 0.333FIFTH: 0.2SEVENTH: 0.143UP TO SEVENTHUP TO SEVENTHFIRST ONLYFIRST, THIRD, AND FIFTHFIRST AND THIRDFEATHEREDAMPLITUDESFIRST: 0.962THIRD: 0.245FIFTH: 0.1SEVENTH: 0.048Figure 1 Additional harmonics sharpen rising and falling edges but do not improvethe worst-case ripples. In contrast, feathering the harmonic amplitudes improvesthe ripple amplitude, but at the expense of some loss in edge sharpness.the amplitudes that the harmonic-summodel uses. The Fourier theory appliesonly to an infinite sum, which neveroccurs in practice. If you have a finitesum, with only a finite number of sinewavecomponents, why not try otheramplitudes different from the ideal infinite-sumvalues? Instead of choppingoff the infinite sum after the seventhharmonic, with subsequent amplitudesdropping abruptly to zero, the right sideof Figure 1 smoothly tapers the harmonicamplitudes so that, by the timeyou get to the last one, its amplitude issmall. That fact mitigates the discontinuityat the end of the harmonic sequence,making a better waveform.Both the number of harmonics and theprecise schedule of their proportions affectthe quality of the result. Stated differently,when considering the adequacyof bandwidth, not only the −3-dB frequencyof your system but also the preciseshape of its entire transfer function mayaffect your result. Harry Nyquist fleshedout that notion in his theorem of datatransmission, concluding that, given infinitecomplexity in the receiver and perfectcontrol over the exact system-transferfunction, the bandwidth must equalor exceed half the system data rate toachieve reliable binary communication.Suppose that a system transmits datawith a baud frequency of B. The fastestalternating pattern you can send isthe sequence 1, 0, 1, 0, ... . That patternis based on a repeating cell two bitslong, 1 and 0, so the pattern has a fundamentalfrequency equal to only halfof B. According to Nyquist, if you passthat much of the signal, you can in theoryrecover all the data.However, you don’t have access to areceiver of infinite complexity or perfectcontrol over the system-transferfunction. You have a digital system thatneeds a wide-open eye with plenty ofmargin for setup and hold, so you mustdesign your system to accommodate therise and fall time, not just the baud rate.I maintain that the bandwidth requiredto preserve a rise and fall timeof T as it propagates through a systemequals approximately one-half dividedby T. If your rise and fall timesequal 10% of the data-baud interval, sothat T equals 0.1/B, then you can expressthe required system bandwidth of0.5/T, after suitable algebraic manipulation,as five times B.All of these words attempt to explainthe habit among old digital-system designers,when someone asks them howmany harmonics of a data sequence arenecessary for data transmission, to say,“I take the fifth.”EDNHoward Johnson, PhD, frequently conductstechnical workshops for digital engineers.FEBRUARY 3, 2011 | EDN 17

BY MOHIT ARORA AND VARUN JAIN • FREESCALE SEMICONDUCTORBOOT-UP, THE SEQUENCEOF STEPS THAT A SYSTEMPERFORMS BETWEEN WHEN YOUSWITCH ON POWER AND LOADAPPLICATIONS, IS SIMPLEIN THEORY BUT OFTEN COMPLEXIN REALITY. THE MAIN JOBOF A BOOT LOADER IS TO LOADTHE OPERATING SYSTEM,BUT SOFTWARE AND HARDWAREENGINEERS VIEW THIS PROCESSIN DIFFERENT WAYS.UNDERSTANDINGEMBEDDED-SYSTEM-BOOT TECHNIQUESTo load a program into memory, you mustfirst load a program into memory. The bootupprocess, often a complex multistep sequenceinvolving numerous substeps, solvesthis problem. Any boot-up process, includingbooting up Windows, Linux, or an embeddedRTOS (real-time operating system),begins with the application of power to thesystem and the subsequent removal of systemreset. During POR (power-on-reset) assertion, you may have toreconfigure hardware peripherals if operational values differ fromthose of default settings. Embedded microcontrollers, for example,often offer various hardware-reset-configuration schemes.Over the last several decades, bootingup has evolved from a simple DOSbasedstep to more complicated multiple-operating-systemchoices or evenperipheral-based boot-up techniques.A USB (Universal Serial Bus) interface,for example, allows you to boot upa disk image from an external storagedevice; this approach is increasinglypopular in industrial- and embeddedsystemapplications because it providesabundant flexibility. In the case of softwarecorruption, for example, in whicha system requires the reloading of newfirmware, the USB technique allowsa service engineer to simply copy newsoftware onto a flash drive and bootfrom it. The service department thereforesaves the thousands of dollars inexpenses that it would otherwise incurin transporting the equipment to themanufacturer for repair.To enable system-boot flexibilityboth from USB, PCIe (PeripheralComponent Interconnect Express),and SDHC (secure-digital-high-capacity)interfaces and from conventionalmemory devices requires in-depth hardwareand software capabilities. Opensourcefirmware—specifically, the U-Boot (Universal Boot Loader) utility,which finds wide use in embedded-systemplatforms—may also be of value.The Linux-based boot loader can automaticallyboot up the operating system;alternatively, it allows a user to manuallyrun explicit commands to start theoperating system, and it supports bootingfrom a variety of interfaces (seesidebar “The U-Boot”).WINDOWS XP SYSTEM BOOTA simple x86 boot sequence is fairlyself-explanatory (Figure 1). WindowsXP follows comparable steps, albeitPOWER ICON: ZEFFSS1/ISTOCKPHOTO.COM; BUTTON AND BACKGROUND: DSGPRO/ISTOCKPHOTO.COM18 EDN | FEBRUARY 3, 2011

with more sophistication. The boot sequencebegins with the application ofsystem power; the processor remains inreset. When all voltages and currentlevels are acceptable, the power supplysends a power-good signal to the processor.The next step, POR negation,takes place when the availability of thegood-power signal negates the processorreset to allow the CPU to begin operation.The CPU points to the ROMboot address and begins executingthe BIOS (basic-input/output-system)code. The ROM BIOS then performsPOST (power-on self-test), a basic testof core hardware to verify basic performance.The boot-up process then reportsany errors that occur at this point usingbeep codes because the video subsystemhas not yet initialized.The next step is video-card initialization,during which the BIOS searchesfor a video-card adapter by scanningmemory addresses C000:0000 throughC780:0000 to find a video ROM. Thevideo test initializes and tests any discoveredvideo adapter, potentiallyalong with its video memory, and displaysconfiguration information. If a“cold” start is taking place, the ROMBIOS executes a full POST. If it is a“warm” start, the boot-up process skipsthe memory-test portion of the POST.The CMOS now reads from theBIOS. During this step, the BIOS locatesand reads configuration informationstored in the CMOS. A small coinbattery cell on the motherboard typicallymaintains the CMOS, a small—typically, 64-byte area of memory—onthe motherboard. The CMOS memorystores information such as date,time, and peripheral-boot order. If thefirst bootable disk is a fixed disk, theBIOS examines its first sector for anMBR (master boot record). The MBRcomprises a partition table, which describesthe layout of the fixed disk, anda partition-loader code, which includesinstructions for continuing the bootprocess. The boot, or partition, loaderthen examines the partition table foran active partition. The partition loadersearches the first sector of that partitionfor a boot record. The boot processthan checks the active partition’sboot record for a valid boot signature. Ifit finds one, it executes the boot-sectorcode as a program.The NTLDR (New Technology Load -AT A GLANCE↘ The system-boot process,although potentially simple in concept,becomes complex when youconsider various implementationoptions.↘ Windows XP provides a popularcase study of a classic bootsequence.↘ Numerous hardware and softwaretechniques provide differentmeans of getting postreset-configurationdata to the processor.↘ Primary and secondary bootoptions comprehend different startupand kernel-code sizes, read- andwrite-performance expectations,and other variables.↘ U-Boot (Universal Boot Loader)is a powerful open-source tool thatdeserves consideration in Linuxbaseddesigns.er) for Windows, a hidden system filethat resides in the root directory of thesystem partition, controls the loadingof Windows XP. During NTLDR’s initialphase, it moves the processor fromreal mode to protected mode, enabling32-bit memory accesses and turning onmemory paging. It then loads the appropriateminidrivers to allow NTLDRto load files from a partition formattedwith any of the file systems that WindowsXP supports, including FAT (fileallocationtable)-16, FAT-32, and NTFS(New Technology File System).STEPS BEFORE THE BOOTSEQUENCE1 POWER ON2 RESET TO PROCESSOR UNTILGOOD POWER SUPPLY IS AVAILABLE3 RESET NEGATED, PROCESSOR-READY, POINTS TO ROM ADDRESSBOOT SEQUENCE1 POWER ON SELF-TEST2 LOOKS FOR VIDEO CARD,RUNS BUILD IN BIOS PROGRAM3 SYSTEM BIOS LOOKS FOR OTHERPERIPHERAL BIOS, EXECUTESOTHER BIOSIf the boot-initialization file residesin the root directory, NTLDR reads itscontents into memory. If the file containsentries for more than one operatingsystem, NTLDR suspends the bootsequence, displays a menu of choices,and waits for the user to make a selection.NTLDR then continues theboot-up process by locating and loadingthe DOS-based New Technologyexecutable file to perform hardwaredetection. After selecting a hardwareconfiguration, NTLDR begins loadingthe Windows XP New Technologykernel file. During this process, thescreen clears, and a series of white rectanglessubsequently progresses acrossit. NTLDR now loads device driversthat are marked as boot devices. Beforeperforming this load, NTLDR relinquishescontrol of the computer.At this point, the system displays agraphics screen with a status bar indicatingthe load status. During later initializationphases, the system cannotaccept device interrupts. The I/O manageralso begins to load the remainderof the system drivers, picking up whereNTLDR left off. The last task for thisinitialization phase is to launch thesession-manager subsystem, which isresponsible for creating the user-modeenvironment. The session-managersubsystem then loads the Windowsdevice driver, which implements thegraphics subsystem. Windows XP bootis not complete until a user has successfullylogged onto the system. The Windowslog-in file begins the log-in proc-4 BIOS DISPLAYS START-UP SCREEN5 MEMORY-COUNT TEST,HARD-DISK-DRIVE-PARAMETER SETTING6 BIOS LOOKS FOR COM AND LTP PORTS7 BIOS LOOKS FORBOOT-SEQUENCE SETTING8 LOADS OPERATING SYSTEM9 BIOS COPIES FILES TO MEMORY,OPERATING SYSTEM TAKES OVER BOOT-UP10 OPERATING SYSTEM CHECKS SYSTEMMEMORY, LOADS DRIVERS,AND STARTS APPLICATIONFigure 1 Microsoft’s Windows XP moves through a suite of steps as it boots a system.FEBRUARY 3, 2011 | EDN 19

SYSTEM CLOCKPOWERON RESETSYSTEM RESETRESET CONFIGURATIONSETTINGS COPIEDTO SYSTEM REGISTERSCODEEXECUTIONSTARTSFROM HEREFigure 2 Loading default values during boot-up is the most common reset configuration.30 MHz Function/Arb GeneratorThe DS345 is a flexible and powerfulfunction/arb waveform generator for awide range of applications.Using Direct Digital Synthesis (DDS),the DS345 can generate very highresolution waveforms. Its full suite offeatures includes AM, FM, PM and Burstmodulation, linear and log sweeps anda 10 MHz reference input.Dedicated buttons make front paneloperation intuitive, unlike competitivemodels with obscure menu driveninterfaces. This makes the DS345 veryeasy to use.If you’re looking for outstanding valuein a function generator, take a closelook at the DS345.• 1 μHz to 30 MHz• 1 μHz frequency resolution• Sine, square, ramp, triangle, noise• 12-bit, 40 MSamples/sec Arbs• 2.5, 5, or 10 MHz reference input• Log and linear sweeps• AM, FM, PM and burst modulation• GPIB and RS-232 (opt.)$1595 (U.S. List)Stanford Research Systems 408-744-9040www.thinkSRS.comess, which the kernel loads as a service,and displays the log-on dialogue box.EMBEDDED-SYSTEM RESETOlder microcontrollers had a singlefixed configuration state for the entireregister suite after reset deassertion.This situation translated into fixed valuesfor parameters, such as clock-speedconfiguration, start-address location,pad slew, drive strength, external-memory-portsize, and peripheral enable/disablestate. It enforced restrictions onthe way users employed chips after resetdeassertion. For example, if the systemdisabled an on-chip oscillator providinga clock to off-chip peripherals at reset,those peripherals would be able towork only after software re-enabled theoscillator.In simpler systems, such behaviormight be acceptable, but it may notmeet requirements if that same chip isfinding use in more complex applicationsthat require different configurationsduring reset. To provide flexibilityto a configuration of registers after reset,you can implement various microcontroller-baseddesign schemes. Somemicrocontrollers also support multiconfigurationschemes, in which the systemselects any configuration by readingthe state of pins during reset. Herewe discuss four reset configurations: default,fuse programming, external pins,and serial interface.Loading default values during bootupis the most common reset configuration,and it requires no special onboardsetup (Figure 2). Conversely, it providesno flexibility or options to configureany register. All the registers initializeto fixed values; thus, the chip exits resetin only one fixed state. This approach isthe fastest for initializing the system beforethe boot process, but it is the leastpowerful mode in terms of capabilitiesfor controlling the system state. It mightwork for some applications but is lesspreferable if you use the same microcontrollerin varying applications withvarying boot requirements.A system POR asserts internal chipreset, with both signals being active low;when deasserted, it restarts the clockand loads the reset configuration in thesystem registers. Depending on the systemdesign, other tasks might gate theboot-up process. For example, the systemmight wait for a PLL (phase-locked20 EDN | FEBRUARY 3, 2011345_HI4C.indd 112/21/2010 3:04:58 PM

loop) to achieve clock locking beforethe internal reset deasserts and the systembegins executing the boot code.Fuse programming involves a resetconfiguration that is the result of programmingthrough a chip’s special testmode, fuses, or on-chip nonvolatileflashregisters (Figure 3). This modeimplements special bits and registersthrough either on-chip fuses or an arrayof nonvolatile-flash-memory registersfor configuring reset-control-wordinformation. These registers requirewrite-once capability; therefore, youcan program them only once in a chip’slifetime. This mode usually requires ahardware-setup or software sequence toprogram these special registers or fuses.Once fuses are programmed and reset isasserted, reset-control-word informationis read from the fuses and copied tothe desired system registers. The systemthen internally deasserts the reset andbegins executing code.This scheme provides the flexibilityto configure different system-registeroptions but requires the implementationof special fuse registers in the design.Because the fuses are one-timeprogrammable and secure, they caneffectively enable and disable functionswithin the chip, thereby creating“phantom” parts with lower prices.This strategy is common with semiconductorvendors, which sell the samesliver of silicon with different costs andfeatures—achieved by blowing the fusesto enable or disable functions.You can also reset the configurationthrough external pins. This schemeuses a group of microcontroller pins tocontrol the reset configuration. Thesepins are externally pulled high or lowduring reset to define a configurationoption. Once the system reset deasserts,the microcontroller internally latchesthese values and decodes them to configurethe system registers. This schemeprovides limited flexibility in selectingcontrol-word configurations. Theavailable number of configurations isdirectly proportional to the number ofpins for this purpose.Engineers often implement the designby means of an external buffer ora line driver, such as the 74LVC125,which drives either a logic one or a logiczero to the pins for reset configuration(Figure 4). The reset signal usuallyconnects to the enable pin of abuffer appears as an input to the pinsthat eventually handle the reset configuration.This approach provides ad-ONE-TIME-PROGRAMMABLEFUSESRESETRESETCONFIGURATION CONFIGURATIONBITSBITSORtristate buffer—that is, an external linedriver—so any change from one to zeroor vice versa to the input of the tristateON-CHIP FLASHFigure 3 Obtaining reset-configuration bits from on-chip fuses or flash memoryprovides device customization during manufacturing.NONVOLATILEREGISTERSFEBRUARY 3, 2011 | EDN 21

V DDGROUNDGROUNDENABLEENABLEENABLEENABLEV DD74LVC125ditional flexibility and control. Forexample, if one of the buffers controlswhether PLL is enabled or disabled duringboot-up, a buffer output can allowusers to enable the PLL when the bufferconnects to the drain-to-drain voltageand conversely disable it when theRESETCONFIGURATIONSYSTEM ON CHIPFigure 4 Accessing the reset configuration through an external bus provides flexibilitythat only the number of pins for this function limits.buffer connects to the source-to-sourcevoltage. The number of pins availablefor this purpose usually constrains thisapproach. Note, too, that most externalline drivers, such as that of the74LVC125, integrate groups of four oreight buffers. To limit the implementation’scost, define the buffer numbers inmultiples of four.You can also load the reset configurationthrough an external serial interface.For a highly integrated complex microprocessor,it is impractical to either dedicateor share pins for the numerousavailable power-up options. This schemeconversely involves loading the chip-reset-configurationdata from an externalserial memory (Figure 5). In a typicalimplementation flow, when system resetasserts, the chip establishes communicationwith the serial memory, subsequentlytransferring reset-configuration informationto the microcontroller. Upon serial-datareception, the microcontrollerconfigures system registers based on thereceived data and deasserts reset. Thismethod provides the maximum flexibilityin configuring options in system registersbecause serial memories can store alarge number of data bytes.In advanced serial-configurationschemes, the serial memory can evenhold software code. In such cases, thesystem reads both reset-configurationdata and boot code from external seri-THE U-BOOTMany standard boot loaders, such as DINK32, OpenFirmware, and x86 bios, find use in embedded-systemapplications. They facilitate the loading of an operatingsystem and bring the system to a safe operating state.U-Boot (Universal Boot Loader) is an open-source programthat is popular in Linux-based embedded-systemapplications. U-Boot provides an automated interactiveenvironment, which offers substantial flexibility anddiverse options for various boot schemes and interfaces.It provides a platform to the end application-developmentuser who doesn’t want or need to delve into thelow-level specifics of chip hardware.After basic initialization of the system, U-Boot startsan interactive program, which allows the user to provideinput through a serial-communication interface-consoleutility, such as Windows’ HyperTerminal. The user canalso choose to run U-Boot in an automated fashion withoutany intervention. U-Boot can reside in an internalROM or a flash memory. After basic initialization of theCPU, local memories, and buses, U-Boot can relocateitself to a RAM location and then continue executingfrom there. A splash screen appears on the serial consolewhen U-Boot is running.U-Boot supports a powerful set of commands that youcan execute through the interactive command window.Other than loading the operating system, these commandsprovide abundant functions, such as memoryload and dump; serial-interface access; and read, erase,and program functions for external memories, such asNAND- and NOR-flash memory, serial-flash devices, andEEPROMs. U-Boot also allows the system to boot froma variety of interfaces, such as USB (Universal SerialBus), SD (secure digital), PCIe (Peripheral ComponentInterconnect Express), or SATA (serial advanced-technologyattachment).To load an operating system in a typical scheme over anetwork such as Ethernet, for example, U-Boot first initializesthe network environment’s variables, copies theoperating-system kernel’s image to the target board, andthen moves execution to the operating-system kernel.After this point, U-Boot plays no role in the system.U-Boot’s image size generally depends on the bootperipheralsupport you select during source-code compilationor building. You can customize U-Boot’s featuresto appropriately suit the application. U-Boot follows astandardized directory structure, which allows high scalabilityand portability to platforms. When porting U-Bootto a new platform, most of the files, other than CPU-,peripheral-, and board-specific files, remain the same.All of these features make U-Boot a desired choicefor embedded-system developers. Because U-Boot is anopen-source boot loader, embedded-system developersare heavily contributing to the U-Boot environment, addingdevice support and otherwise keeping U-Boot richand up to date. For more details on U-Boot-developmentresources, go to http://bit.ly/eZslLO.22 EDN | FEBRUARY 3, 2011

CLOCKRESETCONFIGURATIONSERIALBOOTLOGICCHIP SELECTDATA OUTDATA INSPI-BASEDSERIAL MEMORYFigure 5 An external memory can house not only reset-configuration data but alsoboot code.al memory during the microprocessorresetsequence, thereby requiring fewI/O pins. By reading data stored in, forexample, external SPI (serial-peripheral-interface)memory, the system wouldalso need to configure the SPI memory’sclock frequency along with the powerupoptions for the microprocessor andoptionally load code into the microprocessor’smemory. The system wouldhave to accomplish all these tasks beforethe negation of device reset, therebyensuring that the chip is properlyconfigured when exiting the reset state.Serial memories’ low cost, simple implementation,high flexibility, and optionalsoftware-boot code often makethis option the preferred one for bootingor loading the reset configuration.SYSTEM-BOOT COMPONENTSBoot components are primary or secondarydevices, depending on their abilityto support boot immediately afterreset deassertion. The primary optionprovides a direct-boot capability andPROGRAM-EXECUTIONFLOWPROGRAMCOUNTERINCLUDES INITIALIZATIONFOR SECONDARYBOOT INTERFACEINITIALBOOT LOADERPRIMARY BOOTfacilitates the processor’s first instructionfetch from the memory location inwhich software-initialization code resides.You should enable the primary interfacesimmediately after reset and cansometimes configure them through thereset option. You can place boot loadersor small initialization-code sequencesin these locations. You initialize andconfigure secondary interfaces in theseprimary boot-code sequences and bootloaders. They mainly hold the largeoperating-system kernel’s code, whichloads after the boot loader performs basicinitialization. You can execute theoperating-system kernel directly fromthese interfaces or copy it in processoraccessiblememory (Figure 6). For example,an on-chip ROM componentcan act as the primary boot device,including initial initialization code,which subsequently switches executionto a secondary option, such as externalDDR SDRAM, which in turn holds thecomplete operating-system kernel.Common hardware-boot componentsPROGRAM-EXECUTIONFLOWPROGRAMCOUNTERFULL BOOT LOADER/COMPLETE OPERATING-SYSTEM KERNELSECONDARY BOOTFigure 6 Primary and secondary boot options combine to comprehend large operating-system-kernelimages.WINone of threesnowboardpackages!Introducingthe easiestway to getyour wirelessproduct offthe ground.Enter the Anaren “BIG AIR” sweepstakes foryour chance to win a snowboard gift cardworth $599!* Go to www.bigairsweeps.comSay hello to the industry’s easiest,most elegant, and cost-effectiveRF implementation: AIR modules!> Surface-mount module makes yourproduct ‘wireless’ in record time –based on TI’s CC1101 or CC2500 chips!> Choose 433MHz, 800MHz, 900MHz,or 2.4GHz models – all with tiny,common footprints> Virtually no RF engineeringexperience required> FCC, Industry Canada/RSS, ETSI compliant(or pending) – no additional “IntentionalRadiator” certifications requiredTo learn more, order samples, or obtain aquote – email AIR@anaren.com or contactone of these distributors.*NO PURCHASE NECESSARY TO ENTER OR WIN. PURCHASE OR PAYMENT OF ANY KIND WILLNOT INCREASE YOUR CHANCES OF WINNING. Sweepstakes will begin at 12:00 AM ET onJanuary 1, 2011 and end at 11:59 PM ET on March 15, 2011. Open to all legal residentsin the United States and Canada (excluding Quebec) that are 18 years of age or olderas of 1/1/11 that are business owners or employees of businesses that design and/or manufacture electronic products/equipment. THIS SWEEPSTAKES IS NOT OPEN TO THEGENERAL PUBLIC. For Official Rules, which include entry methods, prize description, andcomplete details, visitwww.bigairsweeps.comSponsor: Anaren, Inc.Void where prohibited.800-411-6596 > www.anaren.comIn Europe, call 44-2392-232392ISO 9001 certifiedVisa/MasterCard acceptedFEBRUARY 3, 2011 | EDN 23(except in Europe)

Surface Mountand Plug-In400 / 800 HzTransformersNow...up to150Watts• 0.4 Watts to 150 WattsPower Transformers• 115V/26V-400/800 Hz Primary• Secondary Voltages2.5V to 300V• Manufactured to MIL-PRF 27Grade 5, Class S, (Class V,155 0 C available)• Surface Mount or Plug-In• Smallest possible sizeSee Pico’s full Catalog immediatelywww.picoelectronics.comPICOElectronics, Inc143 Sparks Ave., Pelham, NY 10803Call Toll Free: 800-431-1064E Mail: info@picoelectronics.comFAX: 914-738-8225Delivery - Stock to one weekINDUSTRIAL • COTS • MILITARYEXTERNALNAND-FLASHMEMORYINTERNALFLASHPROCESSORNAND-FLASHCONTROLLERSYSTEM ON CHIPallow a system to boot from a variety ofinterfaces (Figure 7). Booting from internalflash memory is among the mostcommon and simplest methods of configuringan embedded microcontrollerthat includes the necessary on-chip resources.This method reduces dependencieson external interfaces becausethe boot loader resides in on-chip memory.After system-reset deassertion, theprocessor points to the flash memory’sstarting address and loads the necessaryinitialization code and operatingsystems. Operating systems with smallfootprints arecompatible withthis approach becausea practicallimit exists on theamount of on-chipflash memory thatcan be available.This approach isalso one of mostsecure ways to boot a processor becausemodifying code residing in on-chipmemory requires fewer changes than dooff-chip boot options.As with Windows XP, some microcontrollersintegrate ROM as the primaryboot option. The boot ROM includesa basic boot loader so that the microcontrollercan subsequently perform amore sophisticated boot sequence onits own, loading programs from varioussources, such as Ethernet, NAND-flashmemory, an SD (secure-digital) card,an MMC (multimedia card), or a USBinterface. Boot-ROM usage enablesmore flexible boot sequences than doesSYSTEMMEMORYEXTERNAL-BUSEXTERNALINTERFACE FLASHCHIP SELECTBOOTROMDRAMCONTROLLERDDRMEMORYFigure 7 Common hardware-boot components allow a system to boot from a variety ofinterfaces.AS WITH WINDOWS XP,SOME MICROCON-TROLLERS INTEGRATEROM AS THE PRIMARYBOOT OPTION.hard-wired logic, and it allows usersthe choice to boot up from various peripherals.Users often employ the boot-ROM feature for system recovery whensomeone inadvertently erases the usualboot software in nonvolatile memoryother than ROM. You cannot reprogramboot ROM, so applications thatrequire a secure boot may include securitychecks so that the boot-up stops ifone or multiple security checks fails.An external bus interface allows thesystem to boot directly from externalNOR flash or other parallel memories.It is one ofthe fastest waysto boot the systembecause theinterface to externalmemorycan be 32 bits ormore with a reasonablefrequencyof operation. Fora full-fledged operating system such asLinux, or Windows, it can take severalmilliseconds to seconds to boot the systemdue to the size of the operating system,which can be annoying to the user.Keeping the boot loader/operating systemin external parallel memory reducesthe boot-up time drastically for systemsin which boot time is critical.NAND-flash memories are gainingpopularity in numerous applications dueto their higher read throughput, althoughthis throughput is lower than that ofNOR-flash memory. NAND flash alsooffers faster erases and lower cost per bytethan does NOR flash. The primary use

EXTERNALNAND MEMORYOPERATING-SYSTEMCODEPROCESSORNAND-FLASHCONTROLLERSYSTEM ON CHIPfor NAND-flash memory—for example,in a USB flash drive—is to store largequantities of data and code. However,in recent years, an increasing number ofembedded-system applications also supportNAND-flash memory as a primaryboot option. In these cases, the microcontrollermust include a NAND-flashmemorycontroller to handle read andwrite accesses. The NAND-flash-memoryinterface also requires more than 20pins, so this option may not be cost-effectiveand otherwise practical unlessyou use a package with more pins.Booting from internal volatile memoryis always a secondary boot techniquebecause a primary boot device mustload the RAM before it can executecode. Commonly, a primary boot loadsthe operating system and drivers, copyingthem to RAM. At this point, theRAM-based code takes control of thesystem. Executing code from RAM isfaster and consumes less power than doother memory technologies, includingeither external or internal flash memory.However, because the internal RAMis volatile, some system designs enablethe RAM to switch to battery power inthe event of the failure of the primarysupply, thereby eliminating any furtherneed to copy the code from externalor internal memory when the primarypower returns and thus reducing subsequentinitialization and boot time.Booting from a DRAM is also a secondaryboot technique. DRAM use iscommon in high-end applications thatmust handle abundant multimedia contentand that require high throughput.You can view DRAM as a bigger andfaster extended RAM buffer for managingcomplicated applications. In one example,ROM includes the boot loader,and NAND-flash memory contains theoperating system and the applicationBOOT ROMBOOT LOADERDRAMCONTROLLERDDRMEMORYOPERATING-SYSTEMEXECUTIONFigure 8 Secondary boot using high-performance DRAM requires a correspondingnonvolatile primary-boot device, such as a ROM or a flash memory.code (Figure 8). The boot process beginswith system initialization through theROM-boot loader, which also includesthe reset vector. The main operatingsystemcode copies from NAND-flashmemory to DDR SDRAM, subsequentlyswitching execution control to the externalvolatile memory. This scheme isefficient because executing from DDRSDRAM is faster than reading directlyfrom NAND-flash memory. You can usesimilar schemes to copy code from otherinterfaces, such as Ethernet.You may want to boot from variousinterfaces, such as SDHC, SPI, I 2 C (inter-integratedcircuit), USB, SATA (serialadvanced-technology attachment),PCIe, and Ethernet. These interfaces allrepresent secondary boots. A primaryboot interface, such as ROM, initializesa secondary boot interface, such asUSB, before code execution switchesto the secondary boot device. Storingboot code in external nonvolatile serialmemories, such as SPI flash memoryor I 2 C EEPROM, can be useful for microcontrollersthat have low pin countsand can afford to have longer boot-ups.Such schemes first copy the boot codefrom the external memory to the onchipRAM and code execution switchesto the RAM after reset, so the schemecan immediately fetch boot code afterreset deassertion.EDNAUTHORS’ BIOGRAPHIESMohit Arora is a systems engineer at FreescaleSemiconductor. He has a bachelor’sdegree in electronics and communicationsengineering from Netaji Subhas Instituteof Technology (New Delhi, India).Varun Jain is a senior design engineer atFreescale Semiconductor. He has a bachelor’sdegree in engineering from Delhi Collegeof Engineering (Delhi, India).NewHighVoltageUp to 500 VDCHi PowerUp to 50 WattsRegulatedDC-DCConvertersMiniaturizedSize Package:2.5" X 1.55" X 0.50"High Voltage, Isolated Outputs100-500 VDCLow Voltage Isolated Outputs5-48VDC also StandardQP SeriesIsolatedOutput Voltages from 500VDCHigh Power: to 50 Watts,Efficiency to 90%5, 12, 24, Wide Input RangesConsult Factory for Special InputVoltagesSafe: Short Circuit, Over/UnderVoltage, and Over Temp. ProtectedOptions Available: ExpandedOperating Temperature, -55 0 C to +85 0 CEnvironmental Screening, Selectedfrom MIL Std.883Ruggedized for Operation inHarsh EnvironmentsExternal Bias Control: For ChargePump ApplicationsCustom Modules: Available tooptimize your designs, SpecialInput or Output Voltages AvailablePICO’s QP Series compliments our 650plus existing standard High VoltageModules. Isolated, Regulated,Programmable, COTS and CustomModules available, to 10,000 VDC andother High Voltage to 150 Watts!www.picoelectronics.comE-Mail: info@picoelectronics.comFAX 914-738-8225PICO Electronics,Inc.143 Sparks Ave, Pelham, NY 10803-1837

ADVANCES INENERGY-STORAGETECHNOLOGYPOWERWIRELESSDEVICESBY MARGERY CONNER • TECHNICAL EDITOR26 EDN | FEBRUARY 3, 2011

ENERGY HARVESTING RELIES ONINTERMITTENT ENERGY SOURCES ANDREQUIRES ENERGY STORAGE, SUCHAS A CAPACITOR OR A BATTERY. NEWIMPROVEMENTS IN THESE COMPONENTSPOINT TOWARD SMALLER, LONGER-LIVEDWIRELESS SENSOR NODES, BUT A LONG-LIVED PRIMARY BATTERY MAY BE YOURSIMPLEST ENERGY-SOURCE CHOICE.COMPOSITE ILLUSTRATION BY TIM BURNS. VAULT: MARK EVANS/ISTOCKPHOTO.COM; CIRCUIT PATTERN: MON5TER/ISTOCKPHOTO.COMGoing wireless in applications such as sensornodes or PC peripherals requires not onlywireless communication but also an alternativeenergy source to an ac wall plug. Itmakes little sense to eliminate wired communicationif your design still requires apower cord. Harvesting the ambient energy,which can be solar or artificial lightingor mechanical energy, such as equipmentvibration, provides an energy source for ultra-low-power devices.Devices such as wireless sensor nodes conform to their ultra-lowpowerbudget by spending most of their time asleep, waking up brieflyto take a sensor reading and transmit it in a burst to their nodenetwork (see sidebar “Characterizing average harvested energy”).However, the burst of power necessaryfor communication, althoughbrief, requires more energy than ambientsources can provide, so some formof energy reservoir must provide thisshort, intense burst, making energystorage the other side of the coin forenergy harvesting. Capacitors, includingsupercapacitors, and batteries arethe main forms of electrical energystorage for energy harvesting. Both devicetypes have constraints that energy-harvestingcircuits for wireless sensornetworks must accommodate tomake wireless nodes practical. Thoseconstraints are high leakage current forsupercapacitors and charge/dischargecyclelifetimes for batteries. New lithium-iontechnologies promise to expandthe scope of both supercapacitorsand batteries.For applications in which your wirelessdesign can operate from the energythat one long-life battery stores, thesimplest source of energy is a large-capacitybattery with minimum self-discharge,or leakage, current (see sidebar“Power constraints of wireless sensornodes”). You may worry about themaintenance costs for replacing spentbatteries. Depending on your device’spower budget, however, some primarynonrechargeable batteries can providepower for 10 to 20 years, a longerlifetime than many electronic devices.Ask yourself how many electronicdevices you have been using for morethan 10 years.Not just any battery chemistry is agood candidate. The common alkalinebattery, for example, has only half theenergy-storage capacity of an equiva-FEBRUARY 3, 2011 | EDN 27

lent-sized lithium battery. On the otherhand, alkaline batteries are inexpensiveand have low leakage current, an importantcharacteristic for use in an applicationthat must go many years withoutservice. An advantage of alkalinebatteries in “mostly asleep” wirelessnetworks is that the keep-alive powerduring the sleep phase can be at the alkalinebattery’s lower nominal voltage(Reference 1). Combining an alkalinebattery with a dynamic-voltage regulatorallows the voltage to drop to an alkalinebattery’s nominal voltage of 1.5Vand supply 600-nA keep-alive current.When it’s time to take a reading ortalk to the network, the regulator canboost the battery voltage to the 3V thatmany electronic circuits require. Usinga primary lithium battery with a nominalvoltage of 3V, the keep-alive poweris twice as large as that of an alkalineENERGY SOURCESFOR HARVESTINGALL TEND TO BEINTERMITTENT.battery because of the two-times-largernominal voltage (Reference 2).Lithium-thionyl battery chemistry,on the other hand, can easily lastfor 10 to 20 years with little leakage.Lithium-thionyl batteries can provide asimple approach to your energy needsif your application has a finite life of amaximum of 20 years and can bear thehigher cost of a lithium-thionyl battery,which is approximately $2 comparedwith 50 cents for an alkaline battery.In addition to low self-discharge rates,lithium thionyl also benefits from a flatdischarge profile over time so that theAT A GLANCE↘ Energy storage is a key componentof any power circuit becauseharvested energy is intermittent.↘ Determining your energy budgettells you how much energy you needto get from either a primary batteryor an energy-harvesting circuit.↘ Supercapacitors have high leakagecurrent but virtually unlimitedlife cycles.↘ Batteries have low leakage currentand long shelf life, butrechargeable batteries have limitedcharge/recharge cycles.terminal voltage stays relatively constantover the battery’s service life. Thisbattery chemistry is more costly thanother lithium chemistries and finds usein applications demanding long batterylife, such as water and gas meters andother industrial- and military-electronicsapplications. The cells have nominalvoltages of 3.6V and discharge voltagesof 2.2V. Industrial- and military-gradelithium-thionyl-chloride primary batteriessupply approximately 80 mAhr/year with a failure rate of only one per 1million cells. As a comparison, a microgeneratorthat generates power frommachine vibrations might cost $100(Reference 3).Consider how many electronic devicesare still in use in industrial or infrastructuresettings after more than 20years. Especially for wireless sensor networks,which are still in their infancy,hardware will likely change over thenext couple of years, let alone decades.A primary battery may be the answerfor your energy-storage needs. However,some applications lend themselveswell to energy harvesting. Take Logitech’sK750 solar-power-harvestingwireless keyboard. Wireless keyboardscan go for a long time on a battery, butthe battery at some point becomes exhausted,and the keyboard dies, mostlikely when it’s least convenient. Logitechhas taken a page from Texas Instruments’solar-powered calculatorsand incorporated solar power in itsK750 batteries (Figure 1 and Reference4). The approach allows a user tostore a keyboard, for example, in a darkcloset for three months before its batterypower leaches away, and the keyboardcan run for two months on thebattery power alone.In the world of mostly asleep sensornetworks, the mostly asleep mode enablesthe long life of an energy-miserlysystem. At some point, however, thesensor node will wake up, sense its surroundings,and transmit its status anddata to some receiving node. This burstof activity requires more power than theambient amount of energy available inthat one instant from the environment.In this case, it comes in handy to be ableto store the trickle of available ambientenergy and spend it all in one burst beforegoing back to trickle-charging thestorage device. In this way, it will beready for the next time the node awakesto sense and communicate.Advanced Linear Devices makes modulesthat harvest energy from solar cellsand transducers (Figure 2). “Energy storageis key to energy harvesting becauseenergy sources for harvesting all tend tobe intermittent,” says John Skurla, directorof marketing and sales at the company.“It’s not steady-state input energy. Amicrogenerator can act as a transducer totransform mechanical energy into electricalenergy, and, from there, the electricalenergy goes into some form of storage.”Advanced Linear Devices’ modulesFigure 2 Advanced Linear Devices’ EM300 energy-harvestingmodules provide power management for a varietyof energy sources, includingpiezoelectric transducers.Figure 1 The LogitechK750 wireless solar keyboard canrun for two months on battery power alone.The solar cells top off the lithium-ion button batteries.28 EDN | FEBRUARY 3, 2011

Have you heard?Oscilloscopes fromthe T&M expertFast and efficient, easy to use, precise results. Our newest product linecomes in three performance classes and a total of five bandwidths. Take a look.¸RTO: high performance up to 2 GHzThe R&S®RTO oscilloscopes detect and analyze faster than conventional scopes.The digital trigger system delivers exceptional accuracy, and the intelligentoperating concept and touchscreen make it fun to use.¸RTM: mid-range scopes with 500 MHz bandwidthThe solid features and outstanding price/performance ratio of the ¸RTMmake it the ideal solution for everyday measurements.HAMEG: basic oscilloscopes up to 350 MHzOur subsidiary HAMEG Instruments develops powerful, cost-effective productsfor smaller budgets, since 2009 including digital instruments up to 350 MHz.For details go to www.scope-of-the-art.com

CHARACTERIZING AVERAGE HARVESTED ENERGYBy Tony Armstrong, Linear Technology CorpAmbient-energy sources includelight, heat differentials, mechanicalvibration, transmitted RF signals, orany other source that can producean electrical charge through a transducer.These energy sources are allaround us, and you can convert theminto electrical energy by using a suitabletransducer, such as a thermoelectricgenerator for temperaturedifferential, a piezoelectric elementfor vibration, a photovoltaic cell forsunlight or indoor lighting, and evengalvanic energy from moisture. Youcan use these so-called free energysources to autonomously power electroniccomponents and systems.The applications in which energyharvestingbenefits are compellingare parts of a long and growing list,and all these applications have twothings in common: They all have alow average-power requirement,and they reside in situations inwhich access to ac-line power iscostly or unavailable, batteries arecostly or difficult to replace, or both.Examples include wireless sensorsin shipping containers, asset-trackingdevices, structural monitors inaircraft and bridges, environmentalmonitors in buildings or factories,and remote outdoor locations.A good starting point for the feasibilityof implementing an energyharvestingsystem or wireless sensornode is to consider how muchenergy is necessary to power it. Thegood news is that only 48 μW of continuouspower is necessary to run asensor node for an indefinite amountof time. Furthermore, lower dutycycles produce even lower averagepowerlevels. Another piece of goodnews is that a transducer’s outputpower is generally more than adequatewhen you properly match itto the application, thereby enablingyou to charge an external storageelement for use when the ambientenergysource is unavailable.Because energy harvesting isgenerally subject to low, variable,and unpredictable levels of availablepower, it is advisable to implementa hybrid structure that can interfaceto both the harvester and a secondarypower reservoir. The harvester,because of its unlimited energy supplyand deficiency in power, is theenergy source of the system. Thesecondary power reservoir, either abattery or a capacitor, yields higheroutput power but stores less energy,supplying power when necessary butotherwise regularly receiving chargefrom the harvester. Thus, applicationslacking ambient energy fromwhich to harvest power must use thesecondary power reservoir to powerthe sensor node.From system designers’ perspectives,this requirement adds a furtherdegree of complexity becausethey must now consider how muchenergy the secondary reservoir muststore to compensate for the lack ofan ambient-energy source. Just howmuch energy is necessary dependson several factors, including thelength of time the ambient-energysource is absent and the duty cycleof the sensor node—that is, howoften a data reading and transmissionmust take place. Other factorsinclude the size and type of a secondaryreservoir—that is, whetherit is a capacitor, a supercapacitor,or a battery—and whether enoughambient energy is available to act asthe primary energy source and havesufficient leftover energy to chargeup a secondary reservoir when it isunavailable for a certain time.The need for these factors variesfrom application to application;however, consideration of these factorswill assist system designers inchoosing an energy-harvesting-systemtopology. Moreover, with analogswitch-mode-power-supply-designexpertise in short supply worldwide,it has been difficult to design aneffective energy-harvesting system.The primary hurdle is the powermanagementaspects of remotewireless sensing.Fortunately, recent product introductionsin this area include devicesthat can extract energy fromalmost any source of light, heat, ormechanical vibration. Furthermore,their comprehensive featuresgreatly simplify the difficult powerconversion-designaspects of anenergy-harvesting chain. As a result,system designers and plannersmust prioritize the needs of theirpower management from the outsetto ensure efficient designs and successfullong-term deployments.Author’s biographyTony Armstrong is director of productmarketing, power products, atLinear Technology Corp.currently use capacitors, but the companyis monitoring advances in rechargeablebatteries and supercapacitors.Rechargeable lithium-ion coin batteries,such as the ML2032 that theLogitech keyboard uses, have beenavailable for only about a year. Becauseof their lack of maturity, Skurla has acautious approach to evaluating them.“[A lithium coin battery] tends to bevery finicky in that you can’t overchargeit, and you can’t undercharge it,either,” he says. “If you don’t manage itjust right, the battery will die.” Chemicalreactions occur during overchargingand undercharging. “If you overcharge,you trigger a different chemical reaction,”says Skurla. “Another chemicalreaction occurs during undercharging,and the cell becomes permanently disabled.”The cells don’t explode fromundercharging, he adds. They just die.“That’s one of the characteristics ofthese postage-stamp-sized rechargeablecoin batteries,” he adds. “You never usedto have to deal with undercharge.”Hitachi Maxell publishes the mostcomplete information on ML-serieslithium-manganese-dioxide rechargeablebatteries. The company sellsthem only to equipment manufacturersas built-in parts and does not supplythem directly to users of equipmentwith these batteries (Figure 3). HitachiMaxell specifies the parts for 1000charge and discharge cycles.Advanced Linear Devices has alsoconsidered supercapacitors for storingharvested energy. “Supercapacitors are30 EDN | FEBRUARY 3, 2011

POWER CONSTRAINTS OF WIRELESS SENSOR NODESBy Dave Freeman and Sriram Narayanan, Texas InstrumentsRecent developments in hardware and software designsfor wireless-sensor-network nodes motivate the need forsystem-level design methods. Figure A shows a network,along with the subsystems of each node. Considerationsfor ease of deployment and cost of installation requireINTERNETOREXTERNALNETWORKULTRA-WIDEBANDZIGBEEWIRELESSLANCONVERGENCEBASE STATIONWIRELESSLANMICRO-CONTROLLERSENSORSNODE 1WIRELESSLANMICRO-CONTROLLERSENSORSNODE Nthat the nodes be able to communicate wirelessly. Toreduce communication overhead and improve responsetime, the nodes should be able to locally process sensordata and control actuators. Performing routine maintenance,such as replacing batteries on a large numberof nodes, may be prohibitively expensive. Sensor nodesshould also last for several years by relying only onstored or harvested energy.The choice of sensors, radio, and microcontrollerdepends on the nature of the application.A sensor network in an office environmentaddresses applications such as energy management,security, or resource planning.TABLE A ACCRUAL RATESSetting Intensity (lux) Power density (μW/cm 2 )Conference 300 7.5roomOffice 500 12.5Hallway 400 10ENERGYSTORAGEENERGYSTORAGEACTUATORENERGYSTORAGEENERGYSTORAGEACTUATORFigure A Harvested energy powers sensor nodes that make autonomous decisionsabout their environment and can communicate using multiple protocols.Light energy is usually the most abundant form of ambientenergy in indoor environments. Modern solar cellscomprising amorphous silicon generate approximately 5μW/cm 2 when illuminated by a 200-lux fluorescent-lightsource. Table A lists estimates of energy-accrual ratesand suggests that a 10-cm 2 solar cell may generate70 to 120 μW.Microthermoelectric generators use a gradientin temperature to produce electricalenergy. To generate a power density of 15 μW/cm 3 , however, thermal harvesters may need athermal gradient of approximately 10°C. Manyapplications, particularly indoor environments,do not experience large swings in temperature.Thus, thermal harvesters have limited applicability.Today’s vibrational-energy harvestersneed acceleration of approximately 1.75 to2g—magnitudes usually absent in indoor environments—toproduce 60 μW of power.With limited onboard capacity to storeenergy and limited opportunities to harvestambient energy, the sensor node must befrugal in energy use. Consider, for example, a100-mAhr battery and a solar cell that accrues70 μW for 50% of a 10-year node lifetime. Thisnode must operate its various subsystems andconsume no more than 39 μW of average power.The microcontroller, radio, sensors, and actuatorsexhibit dramatically different power and performancecharacteristics. Meeting the system’s power budgetrequires the sensor node to optimally manage its subsystems.Modern low-power microcontrollers consumeapproximately 345 μW of peak power to operate at aTABLE B LOW-POWER ARCHITECTURESFrequency(GHz)Datarate(Mbps)Nominalrange(m)Peakpower(mW)Energyper bit(nJ/bit)Bluetooth 2.4 1 1 to 10 92.4 923.1 to 10.6 1 to 27 1 to 30 Less than10IEEE802.15.4aultrawidebandApproximately1Zigbee 2.4 0.25 10 to 100 29.7 119Wireless 2.4 54 100 851 15LANsuper in capacity, but they also leak likesieves,” says Skurla. It’s also difficult toget precise specifications for leakage andshelf-life characteristics. He says thatthe company bought and tested a bunchof supercapacitors that were availableon the market, running the tests overdays and weeks. The leakage current isSUPERCAPACITORSARE SUPER IN CAPAC-ITY, BUT THEY ALSOLEAK LIKE SIEVES.proportional to the capacity: As the capacitanceincreases, the leakage currentalso increases. “What you really want isa 10F supercap with the leakage-currentcharacteristic of a 0.1F supercap,” saysSkurla. “They still have about 11-timesmore leakage than we like to see for anenergy-harvesting application.”32 EDN | FEBRUARY 3, 2011

clock speed of about 1 MHz. Given that the requirementsfor sensor processing are often modest, youcan use a microcontroller with an aggressive dutycycle—less than 1%, for example—to reduce averagepower consumption.Sensor nodes usually communicate informationabout physical phenomena and related control messagesat relatively low rates. Table B summarizesthe salient features of some key low-power wirelesscommunicationtechnologies. The power-consumptionquantities in the table serve only as general guidelinesfor system design. As transceiver designs evolve, theyconsume less power. When choosing a transceiverarchitecture, it is important to consider all aspectsof the design. A wireless-LAN transceiver consumeslower energy per bit than does a Zigbee transceiver,but the LAN transceiver has a higher data rate andconsumes more peak power.Sensors that may find use in indoor applicationsinclude thermometers, humidity sensors, microphones,and passive infrared sensors. Today’s temperatureand humidity sensors and microphonesconsume approximately 70 to 80 μW of peak power.Passive infrared sensors that can detect human activitytypically consume peak power of 100 to 500 μW.Temperature and humidity sensors monitor slowlyvarying phenomena and can operate on low dutycycles, but you cannot power off other sensors thatdetect motion without compromising detection performance.In many applications, the sensor demandsmore energy than the processing of the data or wirelesscommunication. Therefore, meeting the systempower budget requires innovation in the way you managethe sensors.The lack of an adequate power supply and energyremains a formidable challenge to implementing wirelesssensor networks, despite advancements in computation,communication, and sensing. Technologicaladvancements in energy harvesting and storagecontinue to ease power constraints, but the demandsof the end application continue to push their limits.Bridging this persistent power gap requires a systemlevel-designapproach that optimally trades off performancefor energy savings and that guarantees aminimum quality of service. Future wireless sensornodes will autonomously adapt to time-varying-applicationdemands and energy availability.See the Web version of this article at www.edn.com/110203cs for a list of references used in writingthis sidebar.FOR MORE INFORMATIONAdvanced LinearDeviceswww.aldinc.comAnalog Deviceswww.analog.comCap-XXwww.cap-xx.comHitachi Maxellwww.maxell.co.jpIoxuswww.ioxus.comJM Energywww.jmenergy.co.jpLinear Technologywww.linear.comOn Semiconductorwww.onsemi.comTexas Instrumentswww.ti.com

Your Project Acceleration Resource forElectronics ManufacturingFebruary 8–10, 2011Anaheim Convention CenterAnaheim, CAElectronicsWestShow.comApril 6–7, 2011Boston Convention& Exhibition CenterBoston, MAElectronics-NE.comSeptember 20–22, 2011Donald E. StephensConvention CenterRosemont (Chicago), ILElectronicsMidwest.comFind the suppliers, tools, and services you need tomake your product, process, and business moreefficient, cost-effective, and profitable.For information on attending or exhibiting,please call 310/445-4200 or visit CanonTradeShows.comFEBRUARY 3, 2011 | EDN C

Figure 3 Maxell sells rechargeable lithium-manganese-dioxideML-series rechargeable batteries. The company offers thesebatteries only to equipment manufacturers as built-in parts, notdirectly to end users.Another advancement in the technology of supercapacitorsis the emergence of lithium-ion supercapacitors. For example,JM Energy two years ago announced its first lithiumionsupercapacitors, the 1000F/2000F series. The devices remainin limited production, however. In December 2010, supercapacitor-manufacturerIoxus announced its lithium-ionseries with an energy density 115% higher than that of standardelectric double-layer capacitors.Ioxus last year received a federal grant to work with BinghamtonUniversity (www.binghamton.edu) on increasing energydensity in supercapacitors. The company’s resulting workyielded a “hybrid” capacitor, which comes in 220, 300, 800,and 1000F versions. It uses a positive carbon electrode anda negative lithium-ion electrode and has a 20,000-cycle lifetimecompared with approximately 1000 cycles for a battery.Leakage current is 0.6 mA at 72 hours. The devices’ heightand diameter measurements, respectively, range from 45×22mm for the 220F version to 88×35 mm for the 1000F version.Prices for the 300F version range from $4 to $6 each.EDNREFERENCES1Conner, Margery, “Run for your life: ultralow-power systemsdesigned for the long haul,” EDN, May 26, 2005, pg 46,http://bit.ly/f6q6rP.2Odland, Keith, “Selecting the best battery for embeddedsystemapplications,” EDN, Nov 18, 2010, pg 36, http://bit.ly/evzniq.3Conner, Margery, “Energy-harvesting conference coversboth cutting-edge and established technologies,” EDN, June13, 2007, http://bit.ly/gCHicK.4Conner, Margery, “Energy harvesters extract power fromlight, vibrations,” EDN, Oct 27, 2005, pg 45, http://bit.ly/hx8Xqw.You can reachTechnical EditorMargery Connerat 1-805-461-8242 andmargery.conner@ubm.com.

BY ALLAN EVANS • SAMPLIFY SYSTEMS INCLowering the cost ofmedical-imaging R&DMEDICAL-EQUIPMENT COMPANIES ARE SHIFTING FROM A VERTICAL-INTEGRATION MODEL TO A SYSTEM-INTEGRATION MODEL TO DO MOREWITH LESS RESEARCH AND DEVELOPMENT.Alookinside a typical piece of medical-imagingequipment reveals myriad technologies frommany engineering disciplines that must collaborateto form a system. For example, developingX-ray scintillators in CT (computerized-tomography)machines to convert X-rayparticleenergy into photon energy requires materials-scienceresearch into rare-earth materials. Capturing this photon energyover wide dynamic ranges requires custom analog-frontendelectronics, and acquiring the data from tens or evenhundreds of thousands of such X-ray detectors requires customizedstorage subsystems and interface technologies. Complexsoftware algorithms then form images from this substantialvolume of raw data. Radiologists, sonographers, and othermedical-imaging professionals examine these images to maketheir diagnoses. This vertically integrated, multidisciplinaryengineering from materials-science research to custom siliconand complex software design has historically required heftyR&D budgets, which only large companies could afford.The current economic environment has caused an increasedfocus on price and performance and forced medical-equipmentOEMs to re-evaluate their R&D budgets tofocus on technologies that are truly keys to their competitiveadvantage, outsourcing other nondifferentiated imagingfunctions to outside semiconductor and subsystem suppliers.These merchant suppliers of technologies can, in turn, enablecost reductions by amortizing their R&D costs over theentire industry, not just across the market share of one equipmentmanufacturer.The evolution of the ultrasound-equipment market providesan instructive example of how this trend will play out.Figure 1 depicts a 10-year-old ultrasound machine whoseprobe contains 64 to 256 piezoelectric transducers, each individuallyconnected through a microcoaxial cable to theconsole. The probe required a specialized manufacturing processfor matching gains and delays across elements. As manufacturersintroduced imaging techniques, analog front endsprimarily comprised discrete components. Top-tier imagingequipmentmanufacturers held this “black art” close to thevest. Processing the tens or hundreds of gigabits per secondof data could take place only in custom ASICs. When medicalapplications began to use analog CRTs for ultrasound dis-TRANSMITTINGBEAM-FORMERASICADCPROPRIETARYBACKPLANEIMAGE-PROCESSINGCPUADCRECEIVING-BEAM-FORMERASICHOST-INTERFACEFPGAPROBELOW-NOISEAMPLIFIERVOLTAGE-CONTROLLEDAMPLIFIERADC12ANALOG FRONT ENDWORKSTATIONFigure 1 This 10-year-old ultrasound machine required a specialized manufacturing process for matching gains and delays acrosselements.36 EDN | FEBRUARY 3, 2011

Supercapacitor-Based Power Backup Prevents Data Lossin RAID Systems – Design Note 487Jim DrewIntroductionRedundant arrays of independent disks, or RAID, systems,by nature are designed to preserve data in the faceof adverse circumstances. One example is power failure,thereby threatening data that is temporarily stored involatile memory. To protect this data, many systemsincorporate a battery-based power backup that suppliesshort-term power—enough watt-seconds for the RAIDcontroller to write volatile data to nonvolatile memory.However, advances in fl ash memory performance suchas DRAM density, lower power consumption and fasterwrite time, in addition to technology improvements insupercapacitors such as lower ESR and higher capacitanceper unit volume, have made it possible to replacethe batteries in these systems with longer lasting, higherperformance and “greener” supercapacitors. Figure 1shows a supercapacitor-based power backup systemusing the LTC ® 3625 supercapacitor charger, an automaticpower crossover switch using the LTC4412 PowerPathcontroller and an LTM ® 4616 dual output μModule ® DC/DC converter.The LTC3625 is a high effi ciency supercapacitor chargerideal for small profi le backup in RAID applications. Itcomes in a 3mm × 4mm × 0.75mm 12-lead DFN packageand requires few external components. It featuresa programmable average charge current up to 1A,automatic cell voltage balancing of two series-connected5V10μF294k100kV IN V OUTEN SW1LTC3625PFI SW2CTL V MIDV SEL GNDGNDPROGPFOL1 3.3μHL2 3.3μHC TOP360FC BOT360FLTC4412V IN SENSEGND GATECTL STATsupercapacitors and a low current state that draws lessthan 1μA from the supercapacitors.Backup Power ApplicationsAn effective power backup system incorporates a supercapacitorstack that has the capacity to support a completedata transfer. A DC/DC converter takes the output of thesupercapacitor stack and provides a constant voltageto the data recovery electronics. Data transfer must becompleted before the voltage across the supercapacitorstack drops to the minimum input operating voltage (V UV )of the DC/DC converter.To estimate the minimum capacitance of the supercapacitorstack, the effective circuit resistance (R T ) needs to bedetermined. R T is the sum of the ESR of the supercapacitors,distribution losses (R DIST ) and the R DS(ON) of theautomatic crossover’s MOSFETs:R T = ESR + R DIST + R DS(ON)Allowing 10% of the input power to be lost in R T at V UV ,R T(MAX) may be determined:R T(MAX) = 2P INL, LT, LTC, LTM, μModule, Linear Technology and the Linear logo are registeredtrademarks and PowerPath is a trademark of Linear Technology Corporation. All othertrademarks are the property of their respective owners.Q2Si4421DYQ1Si4421DY470kV IN1 V OUT1V IN2 FB122μF LTM4616GND ITHM1V OUT2FB2ITHM2GND4.78k10kDN4JDC OUT1100μFC OUT2100μF21.8V1.2VR PROG78.7kL1, L2: COILCRAFT MSS7341-332NLCTOP, CBOT: NESSCAP ESHSR-0360C0-002R7AFigure 1. Supercapacitor Energy Storage System for Data Backup02/11/487

The voltage required across the supercapacitor stack(V C(UV) ) at V UV :V C(UV) = V UV 2 +P IN TV UVThe minimum capacitance (C MIN ) requirement can nowbe calculated based on the required backup time (t BU ) totransfer data into the flash memory, the initial stack voltage(V C(O) ) and (V C(UV) ):C MIN =Data Sheet Downloadwww.linear.com IN BUV C(O)2 –V C(UV)2C MIN is half the capacitance of one supercapacitor. TheESR used in the expression for calculating R T is twicethe end-of-life ESR. End of life is defi ned as when thecapacitance drops to 70% of its initial value or the ESRdoubles.The Charge Profi le into Matched SuperCaps graph in theLTC3625 data sheet shows the charge profi le for twoconfi gurations of the LTC3625 charging a stack of two10F supercapacitors to 5.3V with R PROG set to 143k. Thisgraph, combined with the following equation, is used todetermine the value of R PROG that would produce thedesired charge time for the actual supercapacitors in thetarget application:10F = 5.3V – V C(UV) t C ACTUAL V OUT –V C(UV)t V C(UV) is the minimum voltage of the supercapacitorsat which the DC/DC converter can produce the requiredoutput. V OUT is the output voltage of the LTC3625 in thetarget application (set by V SEL pin). t ESTIMATE is the timerequired to charge from V C(UV) to the 5.3V, as extrapolatedfrom the charge profi le curves. t RECHARGE is the desiredrecharge time in the target application.Design ExampleFor example, say it takes 45 seconds to store the datain fl ash memory where the input power to the DC/DCconverter is 20W, and the V UV of the DC/DC converteris 2.7V. A t RECHARGE of ten minutes is desired. The fullcharge voltage of the stack is set to 4.8V—a good compromisebetween extending the life of the supercapacitorand utilizing as much of the storage capacity as possible.The components of R T are estimated: R DIST = 10mΩ, ESR= 20mΩ and R DS(ON) = 10mΩ.The resulting estimated values of R T(MAX) = 36mΩ andR T = 40mΩ are close enough for this stage of the design.V C(UV) is estimated at 3V. C MIN is 128F. Two 360F capacitorsprovide an end-of-life capacitance of 126F and ESRof 6.4mΩ. The crossover switch consists of the LTC4412and two P-channel MOSFETs. The R DS(ON) , with a gatevoltage of 2.5V, is 10.75mΩ (max). An R T of 26.15mΩis well within R T(MAX) . The value for R PROG is estimatedat 79.3k. The nearest standard 1% resistor is 78.7k. Thedata sheet suggests a 3.3μH value for both the buck andboost inductors.The LTC3625 contains a power-fail comparator, which isused to monitor the input power to enable the LTC4412.A voltage divider connected to the PFI pin sets the powerfail trigger point (V PF ) to 4.75V.Figure 2 shows the actual backup time of the system with a20W load. The desired backup time is 45 seconds, whereasthis system yields 76.6 seconds. The difference is dueto a lower R T than estimated and an actual V UV of 2.44V.Figure 3 shows the actual recharge time of 685 secondscompared to the 600 seconds used in the calculation, adifference due to the lower actual V UV .1.8V OUT1.2V OUTV SCAPV IN20s/DIVFor applications help,call (978) 656-3768DN4JD F02Figure 2. Supercapacitor Backup Time Supportinga 20W Load1.8V OUT1.2V OUTV SCAPV IN200s/DIVDN4JD F03Figure 3. Recharge Time After BackupConclusionSupercapacitors are replacing batteries to satisfy greeninitiative mandates for data centers. The LTC3625 is aneffi cient 1A supercapacitor charger with automatic cellbalancing that can be combined with the LTC4412 lowloss PowerPath controller to produce a backup powersystem that protects data in storage applications.Linear Technology Corporation1630 McCarthy Blvd., Milpitas, CA 95035-7417(408) 432-1900 FAX: (408) 434-0507 www.linear.comdn487f LT/AP 0211 226k • PRINTED IN THE USA© LINEAR TECHNOLOGY CORPORATION 2011

HIGH-VOLTAGE PULSERTRANSMITTING-BEAM-FORMERFPGAHOST-INTERFACEFPGADISPLAYADCPCIECPU/DSP/OMAPDISPLAYPROCESSINGADC∑PROBELOW-NOISEAMPLIFIERVOLTAGE-CONTROLLEDAMPLIFIEROMAP: OPEN MULTIMEDIA APPLICATIONS PLATFORMPCIE: PERIPHERAL COMPONENT INTERCONNECT EXPRESSADCANALOG-FRONT-END REFERENCE DESIGNBEAM FORMERPC BACK ENDFigure 2 Semiconductor manufacturers have introduced highly integrated analog-front-end devices and modules for ultrasoundapplications and replaced ASICs with lower-cost FPGAs for digital processing.plays, manufacturers began to develop proprietary interfacesand backplanes for image processing.Now fast-forward 10 years (Figure 2). The high-levelpicture looks the same, but the components are different.Manufacturers now typically outsource the labor-intensivemanufacturing of the probe to ODMs (original-design manufacturers).As the imaging modes matured, semiconductorcompanies began to understand the specifications for analogperformance. Consequently, they began to introduce highlyintegrated analog front ends and modules with as many as 32channels for the ultrasound-equipment market. Since mid-2008, companies have introduced more than a dozen analogandmixed-signal semiconductor devices for this market.Digital signal processing for functionssuch as beam forming is also moving tothe merchant semiconductor suppliers.Previously, at the 180-nm node, toptierultrasound OEMs could justify theNRE (nonrecurring-engineering) costsfor custom ASIC designs for digitalprocessing. Many of these designs alsofeatured mixed-signal integration withADCs at 8 or 10 bits/sample. However, with new imagingtechnologies, such as tissue harmonic imaging and pulsemodecolor Doppler, leading-edge ASIC designs have becomecost-prohibitive, and the integration of high-performance12-bit ADCs has become too complex. Today’s maskcosts, even at the 65-nm node, are too expensive even formarket leaders. As a result, ultrasound OEMs have turned toFPGAs for their DSP functions. Manufacturers are buildingFPGAs at the 45- or 40-nm node and are announcing 28-nmparts, and they feature gate densities many times larger thanASIC densities at the 180-nm node. Now ultrasound OEMscan purchase an entire analog-front-end subsystem completewith beam forming, which can generate ultrasound imagesout of the box, rather than taking six months and many hundredsof thousands of dollars to custom-design their own.OEMs ARE ENTERINGTHE ULTRASOUNDMARKET WITHA FOCUS ON NICHES.The range of applications for ultrasound equipment has increasedgreatly by enabling OEMs to focus their human andcapital resources on their core features. Specialized ultrasoundmachines are now available for mammography, surgical guidance,and cardiography, and new imaging form factors areemerging for emergency-room and ambulance applications.The availability of these products has lowered R&D costs,allowing OEMs to enter the ultrasound-equipment marketwith a focus on niches. Top-tier OEMs also have benefitedfrom these subsystems and silicon by allowing theseOEMs to rapidly broaden their product portfolios rather thanrely on a one-size-fits-all machine. The health-care industryhas benefited through an improved price-to-performance ratioacross a range of applications.These approaches can provide thesame benefits to imaging technologiesbesides ultrasound. For example, a CTmachine is the result of vertically integrateddevelopment, much the sameas ultrasound was 10 years ago. Again,OEMs’ research into materials scienceresults in customized X-ray scintillatorsthat turn X-ray-particle energy into photons. These scintillatorsrequire customized analog-front-end silicon. Customizedalgorithms and protocols amplify, digitize, encode, andaggregate these scintillators’ data and then transmit it to aworkstation across a slip-ring device. In the CT-image-reconstructionworkstation, customized network-interface cardsterminate custom protocols, and custom RAID (redundantarray-of-independent-disk)controllers buffer the raw data athigh rates until the workstation can reconstruct the image ata lower rate (Figure 3). These workstations can often reconstructa 30-second CT scan in two to five minutes.The high data rates for real-time CT data acquisition drivemuch of the need for custom hardware in the workstation inwhich the data is stored. Consider a 64-slice CT machinewith 1000 detectors per slice with a rate of 10k samples/FEBRUARY 3, 2011 | EDN 39

SENSORSORARRAYRAYADCMULTI-CHANNELADCADCMULTI-CHANNELADCFPGADATA-ACQUISITION ACQUISITION SUBSYSTEMSLIP RINGMULTI-PLEXERBUFFERSERIALIZER/DESERIALIZERDATA-AGGREGATION CARDCUSTOMNETWORK-INTERFACECARDIMAGERECONSTRUCTIONCPUCUSTOMRAIDCONTROLLERFigure 3 In a CT-image-reconstruction workstation, customized network-interface cards terminate custom protocols, and customRAID controllers buffer the raw data at high rates until the system can reconstruct the image.RAID(a)Figure 4 A cranial image (a) is indistinguishable from the reconstructedimage (b) after 3-to-1 data compression.(b)sec and 16 bits/sample. The aggregate throughput requirementis 1280 Mbytes/sec, a rate that exceeds the throughputof commercially available RAID controllers, which topout at 800 Mbytes/sec. CT-signal compression can reducethe throughput requirement of RAID controllers. For example,GE Healthcare (www.gehealthcare.com), SamplifySystems (www.samplify.com), and Stanford University(www.stanford.edu) have demonstrated the efficacy of signalcompression in both lossless and nearly lossless modesto achieve compression ratios that provide compelling reductionsof 3-to-1 or even 4-to-1 in data-acquisition rates(Reference 1).Figure 4 shows an original image and a sample image of3-to-1 compressed data. For more than 400 images, a Stanfordradiologist could not distinguish the image of compressedsamples from the image of noncompressed samples.With signal compression on the rotor side of the slip ringand decompression in software on the workstation, the CTsystem realizes the benefit of bit-rate reduction across theentire signal chain, including the slip ring, the network-interfacecard, and the RAID controller. Integration of signalcompression into CT slip rings makes the compression transparentto the rest of the CT system so that such compressingslip rings can be drop-in replacements for older slip rings.In CT machines, image reconstruction typically occursthree to 10 times slower than does CT data acquisition. Thisasymmetry creates the need for RAIDs in the system to bufferthe raw data until the CT machine can reconstruct it intoan image. Because the RAID is the last link in the data-acquisitionchain, CT-machine designers must build theworkstation on the CT gantry by considering data-acquisitionrates instead of image-reconstruction rates. A new architecturehelps designers build CT workstations based onthe lower image-reconstruction rates rather than the higherdata-acquisition rates (Figure 5).With compression of the X-ray-detector data in the slipring, you can now replace a custom multilane fiber-optic interfacewith a standard storage interface, such as FibreChannelor InfiniBand. This approach allows you to use off-theshelfRAIDs outside the workstation. In this architecture,an off-the-shelf FibreChannel or InfiniBand switch providesthe connectivity between the gantry or the data-acquisitionsubsystem, storage subsystem, and workstation. The workstationcan now use standard low-cost network-interface cardsrather than the customized interfaces that today’sgantry-to-console connectors use.Furthermore, because image-reconstructionrates are several times lower than data-acquisitionrates, designers can build decompression into theRAID, enabling the integration of compressioninto CT systems in a manner that is transparent toimage-reconstruction software. With the eliminationof the need for a RAID or its transition intothe gantry, the designer can now base the workstationon commodity PC-server hardware.With R&D budgets tightening, medical-imagingOEMs must deliver next-generation technologiesacross a wider range of market segments withlower development costs. This move requires themto rationalize technologies that target the competitiveadvantages of their machines.EDN40 EDN | FEBRUARY 3, 2011

SENSORSORARRAYRAYADCMULTI-CHANNELADCADCMULTI-CHANNELADCFPGADATA-ACQUISITION ACQUISITION SUBSYSTEMSLIP RINGMULTI-PLEXERBUFFERSERIALIZER/DESERIALIZERDATA-AGGREGATION CARDFIBRE-CHANNELSWITCHRAIDCONTROLLERFIBRE-CHANNEL-NETWORK-INTERFACECARDIMAGERECONSTRUCTIONCPUCOMPRESSINGRAIDFigure 5 A new architecture for CT workstations puts compression of the X-ray-detector data into the slip ring, so that you canreplace custom multilane fiber-optic interfaces with a standard storage interface, allowing the use of off-the-shelf RAIDs.REFERENCE1 Wegener, Albert; Naveen Chandra; Yi Ling; Robert Senzig;and Robert Herfkens, “Real-time compression of raw computedtomography data: technology, architecture, and benefits,”Proceedings of the SPIE Medical Imaging Conference,Volume 7258, pg 7258H, March 2009, www.samplify.com/ct.AUTHOR’S BIOGRAPHYAllan Evans is vice president of marketing at Samplify SystemsInc (Santa Clara, CA), a provider of signal-compression andbeam-forming technology for the medical-imaging market. Heholds a master’s degree in electrical engineering from the Universityof California—San Diego and a master’s degree in businessadministration from Santa Clara University (Santa Clara, CA).FEBRUARY 3, 2011 | EDN 41

5A LDO for Digital ICs85mVDropoutVoltageV INLT3070DigitallyProgrammableOutputV OUTV OUT50mV/DIVAC-CoupledTransient Load ResponseDigital orAnalogMarginingEasilyParalleledI OUT2A/DIVI = 500mAto 5AV OUT = 1V20μs/DIVC OUT = 16.9μFI OUT : t RISE /t FALL = 1μsUltrafast Transient Response Minimizes Output CapacitanceOur new generation linear regulator, the LT ® 3070, provides all the necessary performance and features to power high currentand low voltage digital ICs such as FPGAs and DSP cores. It delivers 5A of output current with voltages as low as 0.8V,noise performance of only 25μV RMS and very fast transient response with minimal output capacitance. Its MOS-basedarchitecture enables ultralow dropout of only 85mV at full load with no variation in ground pin current even as the input oroutput voltage changes. The device’s high PSRR makes it excellent for post-regulating a switching regulator. Devices canalso be paralleled for 10A or more output current.FeaturesRipple Rejection vs FrequencyInfo & Free Samples• Ultralow Dropout Voltage: 85mV• Ultralow Noise: 25μV RMS(10Hz to 100kHz)• Fast Transient Response OverLine & Load• Input Voltage Range: 0.95V to 3V• Digitally Programmable Output:0.8V to 1.8V in 50mV Increments• Digital Margining (LT3070)Analog Margining (LT3071)• Stable with only 15μF of CeramicOutput CapacitanceIN Pin Ripple Rejection (dB)807060504030C OUT = 117μFC OUT = 16.9μF20 V OUT = 1VV IN = 1.3V + 50mV P-P Ripple10 V BIAS = 2.5VI OUT = 5A010 100 1k 10k 100k 1M 10MFrequency (Hz)www.linear.com/30701-800-4-LINEAR, LT, LTC, LTM, Linear Technology and the Linear logo areregistered trademarks of Linear Technology Corporation. All othertrademarks are the property of their respective owners.

EDITED BY MARTIN ROWEdesignideasAND FRAN GRANVILLEREADERS SOLVE DESIGN PROBLEMSCompute a histogramin an FPGA with one clockMohit Kumar, Texas Instruments, Bangalore, India↘Histograms are often useful toolsfor analyzing digital data. To getreliable results from a histogram, though,you must collect large amounts of data,often with 100,000 to 1 million points.If you need to collect an ADC’s digitaloutputs for analysis, you can use anFPGA (Figure 1).The figure shows the histogram,RAM, and pulse-generator blocks,which let you capture and display thehistogram computation based on 14-bit data. The RAM block is the FPGA’sbuilt-in RAM, and the histogram blockis the VHDL (very-high-level-designlanguage)code to compute the histogram.You can also download the VHDLcode for this application from the onlineversion of this Design Idea at www.edn.com/110203dia.The 14-bit parallel data, Device_Data[13..0], from an ADC goes tothe histogram block and to the RAMRd_Addr input. The RAM providesthe data at its address location, RAM-DataOut[15..0]. This data loops back tothe histogram block, which incrementsit by one and sends it to output pinDataOut[15..0], a 16-bit data output.When the WREN (write-enable) pin isDIs Inside44 Protect MOSFETs in heavy-dutyinductive switched-mode circuits46 Control an LM317Twith a PWM signal47 High-speed buffercomprises discrete transistors48 Limit inrush currentin high-power applications▶To see all of EDN’s Design Ideas,visit www.edn.com/designideas.at logic level one, the data is written atthe address at pin Wr_Addr[13..0]. Thatapproach is the same as if the data werecoming from Device_Data[13..0].The RAM has a fixed delay from itsDEVICE_DATA[13..0]CNTR_VALUE[14..0]START_CNTR_INPUTRST_RAM_INPUTSEL_DATA_INPUTDEVICE_CLKPULSE_START_INPUTADDRIN[13..0]DATAIN[15..0]CNTR_VALUE[14..0]START_CNTRRST_CNTRRST_RAMSEL_DATACLKPULSE_GENPULSE_STARTCLKPULSE_OUTHISTOGRAMRAM_WR_ADDR[13..0]WRENDATAOUT[15..0]Figure 1 A histogram computational circuit retrieves data from an FPGA’s RAM block.RAMWR_ADDR[13..0]WRENRAMDATAIN[15.0]RD_ADDR[13..0]CLKRAMDATAOUT[15..0]HISTOGRAM_OUT[15..0]input to its output. That is, when theinput is at Rd_Addr, data becomesavailable at its output pin, RAM-DataOut, after a fixed delay. This delaymight vary across the FPGA. To takecare of this delay, allow a two-clockdelay to Device_Data before calculatingthe histogram. The delay in RAMshould be less than two clock periods;otherwise, there could be data loss. Thisconstraint limits the maximum frequencyof Device_Clk.The Cntr_Value gives the number ofinput data for which the histogram blockcomputes the histogram. To reset theFEBRUARY 3, 2011 | EDN 43

Get an edge: laser driversPhoto Courtesy of Microvision, Inc.System designers use Maxim laser drivers to meet critical speed and integration challenges in applications rangingfrom fiber optics to consumer electronics. Maxim’s complementary bipolar/BiCMOS technologies enable fullyintegrated solutions with hyperfast edge rates, precise laser control, and other advanced performance capabilities.Inside Microvision’s PicoP® display engineMicrovision uses Maxim’s RGB laser driver to deliveran HD pico projector module small enough to fitinside portable devices.LASER MODULATION UNITMAX3600INTENSITYMODULATORREDLASERTWO-AXISSCANNER MIRRORAt the heart of the PicoP® display engine is theMAX3600, a triple-channel laser driver that packsthree 10-bit DACs along with ample digital controllogic into a 25mm 2 TQFN. This solution cuts size by75% and offers 2ns edge rates to give Microvisionthe most advanced projector on the market.PicoP is a registered trademark of Microvision, Inc.RINTENSITYMODULATORINTENSITYMODULATORR G BCOLOR TRANSFORMATIONR G BGREENLASERBLUELASERSCANNER CONTROLLERIMAGE PLANEINPUTMODULEGBFRAMEBUFFERGet the laser driver you need at:www.maxim-ic.com/Laser-DriversInnovation Delivered and Maxim are registered trademarks of Maxim Integrated Products, Inc. © 2011 Maxim Integrated Products, Inc. All rights reserved.DIRECTwww.maxim-ic.com/shopwww.em.avnet.com/maximFor a complete list of Maxim’s franchised distributors, visit www.maxim-ic.com/distributor

designideasthe value of V TL:V TLV INL V INH 0.1V1.165V.The IC’s data sheet gives the valuesof 0.8 and 2V as the limits of thelow and high input voltages, respectively.The high-to-low transition ofthe clock has no effect. This transitioncauses a sharp negative exponentialpulse, which an internal Schottkydiode at input IN Asuppresses. Theanode of this internal diode connectsto ground, and its cathode connects toinput IN A. Resistor R Slimits the peakcurrent flowing through the protectivediode to about 10 mA.The IC has an output current of±4A. The typical on-resistance of Q 1is2 mΩ. You interconnect Q 2’s gate andsource pins to create a freewheelingdiode. This diode has a typical reverserecoverytime of 33 nsec at a 25A forwardcurrent. When Q 1turns off, thepeak inductor current flows throughQ 2. Voltage V Roccurs on power resistorR and is superimposed onto the supplyvoltage, V DDMOS. The sum of thesevoltages must be lower than or equal tothe manufacturer’s specified value ofthe drain-to-source voltage of transistorsQ 1and Q 2.When testing the circuit, you shouldmonitor the dc-supply current. You cancalculate the ideal-case supply currentas a function of supply voltage on thepower section and the pulse period asfollows:1 V DDMOS T 2 PONI SID = ×2 L1×f REP = I LPEAK T PON f REP .2You calculate the pulse width of a singleinterval when the channel of Q 1is conductiveas an approximation relating therise, fall, on, and off times of the FET:T PONT P +t DOFF −t DON +t DMOSOFF1−t DMOSON +(t R +t F )× − .2V TV DDThe sum of differences in the propagationdelays of IC 1and Q 1is positiveand totals 32.1 nsec. V Tis the gate-tosourcethreshold voltage of Q 1. The datasheet gives a typical V Tof 1.1V, and thesupply voltage, V DD, has a value of 5V.These values yield 9.8 nsec for the lastterm of the preceding equation. T PONisthus larger by 41.9 nsec. For a good design,an ammeter will indicates a supplycurrent one to 1.5 times the ideal valueof the current.You can check the peak voltage atload resistor R. D 1and D 2and storagecapacitor C Sfunction as a peak detector.The peak-voltage pulses at resistorR cause a dc voltage at C Sof roughlythe same value as the peak voltage. Youcan determine the peak current flowingthrough the inductor from the voltageat the peak detector using the followingequation:VI RPEAKLPEAK .RSet the auxiliary supply voltage at5.078V, the supply voltage at 10V, andthe clock-pulse repetition frequency to11,387 Hz. This approach causes thesupply’s current to be 0.327A and thepeak voltage to be 16.4V. The peak currentof the inductor reaches 30.94A.The experimentally determined turn-ontime is approximately 1.502 μsec.The IC driver contributes to the protectionof the MOSFETs with an undervoltagelockout. If the supply voltage ison, but the auxiliary supply voltage isoff, a voltage could get from the IN Apin through internal protective diodesto the V DDpin. The undervoltage lockoutdisables the control outputs untilthe auxiliary supply’s voltage reaches atypical value of at least 4.2V.The 0.65-mΩ dc resistance of ferritecoreinductor L might seem to be overratedfor the circuit. However, the slopeof current pulses in the inductor representsa megahertz-range equivalent frequency.The effective resistance at theseslopes increases due to the skin effectand the proximity effect. This effectiveresistance can be many times thedc value.EDNControl an LM317T with a PWM signalAruna Prabath Rubasinghe, University of Moratuwa, Moratuwa, Sri LankaPWMR 11MC 1100 nFV 112VV IN V OUTIC 1LM317TR 21410kR 4V ADJR 333072IC 163 5Figure 1 This circuit replaces a potentiometer with an analog voltage that you cancontrol from a PWM signal.V OUT↘The LM317T from NationalSemiconductor (www.national.com) is a popular adjustable-voltageregulator that provides output voltagesof 1.25 to 37V with maximum 1.5Acurrent. You can adjust the output voltagewith a potentiometer. The circuit inFigure 1 replaces the potentiometerwith an analog voltage that you cancontrol from a PWM (pulse-widthmodulation)signal. You control thissignal with a microcontroller or anyother digital circuit. You can use thesame microcontroller to dynamicallymonitor the output and adjust theLM317T.Using an RC lowpass filter and anop amp, you can convert the PWMsignal to a dc level that can adjust the46 EDN | FEBRUARY 3, 2011

YOU CAN IMPROVETHE CIRCUIT BYREPLACING THE RCLOWPASS FILTER WITHAN ACTIVE FILTER.LM317T’s voltage output. Varying thepulse width of the input signal lets yougenerate an analog voltage of 0 to 5V atthe output of the lowpass filter. The opamp multiplies the voltage to achievethe desired voltage range.For scenarios in which you mustmultiply the input voltage by two, theLM317T’s adjustment pin receives 0 to10V. Its output-voltage range is 1.25 to11.25V. The equation V OUT=V ADJ+1.25Vgoverns the LM3175T’s output voltage.You can change the op amp’s gainby choosing proper values for R 4andR 2. You must be able to remove offsetvoltages from the op amp. Use an opamp, such as a National SemiconductorLM741, with null adjustment. The selectionof values for the capacitor and resistorfor the RC lowpass filter depends onthe PWM signal’s frequency. This circuituses values for a 1-kHz PWM signal.You can improve the circuit by replacingthe RC lowpass filter with an activefilter and then feeding a feedback signalfrom the circuit’s output into the microcontrollerfor dynamic adjustments.EDNHigh-speed buffer comprisesdiscrete transistorsLyle Russell Williams, St Charles, MO↘Circuits sometimes need a gainof-onebuffer to lower output impedanceand prevent the load from interferingwith the previous stage. For anapplication involving a 1.5-MHz, lowpowertransmitter and antenna, a BurrBrown (www.ti.com) BUF634 buffer ICwould work, but a discrete transistor buffermay be more convenient and less expensivethan the IC.Figure 1 shows the classic design ofsuch a buffer. This circuit can drive aload as low as 200Ω with a peak outputvoltage of 2V. The maximum collectorcurrent of the transistors limits the output.You can use larger output transistorsif your application requires more outputcurrent. Trimmer resistor R 4across thediodes is, however, a relatively expensivepart, and you must adjust it to producethe correct bias current for ClassAbsolute Hall.AS5510 – Smallest Linear Hall Sensor ICwith I²C InterfaceProgrammableSensitivityI²C Interface10-bit ResolutionPower-Down Modewww.austriamicrosystems.com/AS5510FEBRUARY 3, 2011 | EDN 47

designideas12V12VR 310kR 310kINPUTC 10.1 μFQ 12N3904INPUTC 10.1 μFQ 32N3904D 11N914Q 12N3904R 410kOUTPUTOUTPUTR 110kD 21N914Q 22N3906R 110kQ 22N3906Q 42N3906R 210kR 210k–12VFigure 1 A typical buffer can drive loads as low as 200Ω.–12VFigure 2 Current-mirror transistors replace the diodes in Figure 1.AB operation. The adjustment is likely to drift over time.A simpler circuit, such as the one in Figure 2, uses currentmirrortransistors Q 1and Q 2instead of diodes. Resistors R 2and R 3set the zero-signal bias current in the bias-transistorcircuits. The current-mirror effect causes the current in theoutput transistors to be nearly equal to the current in the biastransistors—approximately 1.2 mA in this case.Because the current-gain-bandwidth product of the 2N3904and 2N3906 transistors is 300 MHz, this circuit should work at100 MHz or higher frequencies. At these frequencies, however,the circuit layout may be critical, and the slew rate, whichis unknown, may limit usefulness. The offset of the circuit isapproximately 0.1V, which is not a problem for this applicationbecause the circuit uses capacitive coupling through C 1.If you use the buffer in the feedback loop of an op amp, the opamp can null the offset.You may want to monitor the current in the output transistors,so the circuit in Figure 3 adds 100Ω resistors R 3andR 4and 10-μF bypass capacitors C 2and C 3in the collectors ofoutput transistors Q 3and Q 4. The voltage across these resistorsreveals the collector currents, which are nearly equal in thetwo output transistors, and is close to the value that the valuesof R 1and R 2predicted.EDNINPUTC 10.1 μFR 510kR 110kR 210kC 210 μFC 310 μF12VQ 12N3904Q 22N3906–12VR 3100Q 32N3904Q 42N3906R 4100Figure 3 Resistors R 3and R 4limit the output current.OUTPUTLimit inrush currentin high-power applicationsJB Castro-Miguens, Cesinel, Madrid, Spain, andC Castro-Miguens, University of Vigo, Vigo, Spain↘A high-power offline supply isnothing more than a half- or fullbridgedc/dc converter. Rectifying the acline yields a dc voltage that feeds theconverter. At power-supply turn-on, thebulk capacitor of the uncontrolled rectifieris completely discharged. It results ina huge charging current for a high instantaneousline voltage because the dischargedbulk capacitor temporarilyshort-circuits the diodes of the rectifierstage. The high inrush current can triggera mains circuit breaker, burn a fuse,48 EDN | FEBRUARY 3, 2011

or even destroy a power supply’s rectifierdiodes unless you take precautions.The circuit in Figure 1 limits the inrushcurrent.At turn-on, if the instantaneous rectifiedac-line voltage, V ACR, is greater thanapproximately 10V, Point A in Figure2, MOSFET Q 2turns on, forcing thyristorQ 1off. In this situation, a little currentflows through R 1and Q 2, injectinga small charge into bulk capacitor C O,Path A to B in Figure 2.When V ACR−V O≤8V or so, where V Ois the output voltage, Q 2is off, lettingQ 1conduct. In this situation, the bulkcapacitor receives the necessary chargethrough Q 1, Path B to C in Figure 2, tomatch V Oto V ACR. After this point, V ACRfalls below V O, and the bulk capacitoralone must support any power the dc/dcconverter demands until V ACR−V O≥5Vor so, Path C to D in Figure 2. At PointD, V ACR−V O≈5V and thyristor Q 1triggers,which conducts the capacitor’scharge current and the current the dc/dcconverter demands until V ACRmatchesthe sinusoidal peak at Point E.When V ACRfalls, thyristor Q 1cuts off,and the bulk capacitor alone feeds thedc/dc converter. The thyristor conductsagain when V ACRmatches V Oto the sinusoidalpeak. This process then repeats.Use a nonsensitive gate thyristor with abreakdown voltage of at least 400V foran ac voltage of 220V rms (root meansquare) and with twice the rms-currentrating of the rectifier diodes.This circuit uses a TYN610 thyristor.You can calculate the value of R 1usingR 1=(6.8−V GT)/I GT−20°, where V GTis theminimum gate-cathode voltage necessaryto produce the gate-trigger currentfor Q 1and I GTis the minimum gate currentto trigger Q 1down to −20°C. TheNTD4815NHG MOSFET is suitable forthis circuit. A MOSFET with a differentthreshold voltage may require differentvalues for R 2and R 3.EDNV ACR TYN612Q 1R 11801W1N4007C O CONVERTERDC/DCQ 2NTD4815NHGV O15VR 23.3kR 33.3kNOTE: 15 nF IS AN OPTIONAL VALUE FOR C 1 .Figure 1 A thyristor and a MOSFET control current to bulk capacitor C O. This circuitlimits the inrush current.EV ACRV OFAB CDFigure 2 if V ACRis greater than approximately 10V, MOSFET Q 2turns on; currentflows through R 1and Q 2, injecting a small charge into bulk capacitor C O.FEBRUARY 3, 2011 | EDN 49

supplychainEDITED BY SUZANNE DEFFREELINKING DESIGN AND RESOURCESA new twist on design chain:EBV collaborates to launch chipsDesign-chain strategyusually involves meetingdesign with suppliersand encouraging productdevelopment with support fromin-house engineers or educationalmaterials. Distributor EBVElektronik (www.ebv.com), anAvnet company, is taking thatstrategy one step further bylaunching chips it developed incollaboration with its line-cardsuppliers.In doing so, EBV stressesthat it will not become a chipsupplier or producer. Instead,the company will accent componentdesign and distributionas parts of its business strategy,aiming to create opportunitiesby matching customerneeds for focused designswith appropriate suppliers foraccess to specially customizedproducts. EBV will combinerequests from customersin a segment and then collectinputs to design a rough specification.Using that input, itwill look to suppliers that providethe desired technology forproduction.“We collect inputs, and wewill make design proposals,but one of our suppliers doesthe production,” says BerndSchlemmer (photo, left),director of communications atthe company. The suppliers’logos are on the parts, and thesuppliers provide the warranty,he adds.The effort, which EBVannounced in January 2009and elaborated on at 2010’sElectronica show, aims to provideaccess to specially customizedproducts to small andmidsized customers, beyondlarger key-account customers.“The major advantage forour customers is access to thelatest technologies and accessto our suppliers,” says KlausSchlund (photo, right), technicaldirector at EBV. “You canThe basicprinciplebehind thisapproach is atailored solutionfor certainapplications.imagine when there is a midsizeenvironment in Italy orFrance; usually, the big playersin the semiconductor marketare not very open to go intonegotiations for an ASIC or forany kind of IC. EBV can bundlethose designs and talk to theright guys at our suppliers.”Because it will not producethe semiconductors, EBVsays that its suppliers haveno concerns about competition.These collaborations willgenerate chips that will notcompete with any products,Schlemmer explains. He notesthat an EBV chip is somethingthat is not available yet and thatcomes as additional businessto suppliers that want to provideaccess to such speciallycustomized products.“Some of our customers arehappy to share their IP [intellectualproperty] and knowledgewith other customers,”Schlund says. “The basic principlebehind this approach is atailored, dedicated solution forcertain applications.”EBV chips will target EBV’svertical automotive, lighting,medical, RFID (radio-frequencyidentification), consumer, andrenewable-energy segments.The company has also partneredwith Texas Instruments(www.ti.com) on industrialinterfacedesign and is workingwith STMicroelectronics (www.st.com) and Vishay (www.vishay.com). Products, whichEBV and Avnet will distribute,should become available in2011.“We are the fi rst distributorto start a program like this,”Schlemmer says.CHINA’S RFIDMARKET TODOUBLE BY 2014China’s RFID (radio-frequency-identification) markethas grown signifi cantly, withrevenue set to more thandouble from 2009 to 2014,according to iSuppli (www.isuppli.com), now part of IHSInc. The market comprisestags, readers, and software/middleware. In early January,iSuppli predicted a 22%increase—from $1.1 billion in2009 to $1.4 billion in 2010.By 2014, iSuppli expectsthe Chinese RFID market toreach $2.4 billion.“The rise of China’s RFIDmarket is the result of risingdemand from applications intransportation, warehouselogistics, electronic payment,medical-equipment tracking,food-security systems, assetmanagement, and more,”says Vincent Gu, senioranalyst for China research atiSuppli. “At the same time,technology integration inthe RFID market has beenadvancing. Many enterpriseshave begun to study technicalinnovations to broadenthe applications for RFIDdevices, including mobilepayment.”The company notes that,although costs are declining,they remain a barrier toRFID’s acceptance. As thetechnology matures, iSupplipredicts, so will manufacturingof the devices, which willallow the devices to reacheconomies of scale.OUTLOOK50 EDN | FEBRUARY 3, 2011

IntroducingEDN’s NewMicrosite on LEDsYour complete source for up-to-the-minuteinformation on LEDs and their applications. Latest News from the Editorsat EDN, Design News andother Engineering Magazines Blogs, including MargeryConner’s Popular “Power-Source” Blog Resource Center, IncludingWhite Papers, eBooks,Tutorials, Webcastsand More Videos, IncludingDesign Tips, ApplicationDemos, and More Updates on FutureIn-Person Workshopsand Seminars, likethis one! Fun Challenges to Test YourKnowledge of LEDsCheck It Out Today atwww.edn.com/leds

productroundupPASSIVESinducing damage to the stops on thepotentiometer. The three-turn model3547 and five-turn model 3548 potentiometerswith the slip-clutch optionsell for $16.99 to $18.31 (1000). The10-turn model 3549 with the optionsells for $14.44 to $15.49 (1000).Bourns Inc, www.bourns.com1.2- and 1.5-mm inductors come in 1212 cases↘The IHLP-1212AB-11 and IHLP-1212AE-11 inductors have 1.2- and 1.5-mm profiles, respectively, and both have 3×3.6-mm footprints. They targetuse in voltage-regulator-module and dc/dc-converter applications, including nextgenerationmobile devices, notebooks, desktop computers, and graphics cards. TheIHLP-1212AB-11 offers an inductance of 0.22 to 0.56 μH, a saturation current of6.7 to 9.3A, and a typical dc resistance of 9.5 to 18.7 mΩ. The IHLP-1212AE-11offers an inductance of 0.22 to 1 μH, a saturation current of 5.3 to 9A, a typical dcresistance of 9.5 to 29.5 mΩ, and a maximum dc resistance of 11.4 to 33 mΩ. Thedevices sell for 35 cents (10,000).Vishay Intertechnology Inc, www.vishay.comPrecision potentiometeroffers slip-clutch option↘The 3547, 3548, and 3549 multiturnpotentiometers now includea slip-clutch option, a safety featurethat helps preventdamageto the potentiometerin applications using fulltravel. The option thus increases reliabilityand reduces potential servicecosts and system downtime. The featuretargets use in mechanized or machineto-machineapplications, such as linearactuators, cable transducers, andmechanical position sensors in whichthe mechanical actuator may occasionallyrotate the shaft beyond themechanical stops or beyond a set numberof turns. The internal design optiondoes not alter the physical dimensionsof the component and allows the deviceto idle at the extreme clockwise orextreme counterclockwise ends withoutHybrid capacitorincreases energy densityover competitors↘This new hybrid capacitor combinesan ultracapacitor and alithium-ion battery, providing a 115%-higher energy density than that of standardelectric double-layer capacitors.The devices provide more than 20,000charge/discharge cycles. The hybridcapacitor finds use in flashlights, LEDs,memory backup, portable hand tools,solar chargers, off-grid lighting, andautomotive power windows and doorlocks. The capacitor operates over arange of −25 to +60°C.Ioxus Inc, www.ioxus.comLow-profile powerinductors target usein mobile devices↘The BRL2515 and BRFL2518wire-wound power inductorsmeasure 2.5×1.5 and 2.5×1.8 mm, respectively,with maximum heights of1.2 and 1 mm, respectively, to offer dc52 EDN | FEBRUARY 3, 2011

ADVERTISER INDEXCompany Page Company PageAgilent Technologies 4Linear Technology 37–38, 42Analog Devices Inc 13Maxim Integrated Products 45Anaren Microwave 23Micrel Semiconductor 8Ansys Inc C-4Mill-Max Manufacturing Corp 11austriamicrosystems AG 47Pico Electronics Inc 24, 25bias characteristics in a thinner, smallersize. The BRL2515 offers a saturationcurrent of 1000 mA and inductance of2.2 μH and has a dc resistance of 0.07to 0.265Ω. The BRFL2518 offers a dcresistance of 0.9 to 0.33Ω. Both ROHScompliantinductors are available forsampling at 20 cents each.Taiyo Yuden, www.t-yuden.comDigi-Key Corp C-1, 5 Renesas Technology Corp 6Global Foundries C-2, C-3 Rohde & Schwarz 28Interconnect Systems Inc 21Signal Consulting Inc 41International Rectifier Corp 9Stanford Research Systems Inc 20Lambda Americas 49Tektronix 31, 33, 35EDN provides this index as an additional service. The publisher assumes no liability for errors or omissions. FEBRUARY 3, 2011 | EDN 53

TALES FROM THE CUBEEARL SCHLENK • ENGINEERTower of babbleWhile working for a railroad, I once had to troubleshoota remote microwave site that had failedwhile communicating with a track-side switchcontroller in southern Missouri. To throw aswitch on one of the tracks, a dispatcher in FortWorth, TX, would send a signal over the microwavelink to a VHF radio that connected to the microwave link.While investigating this situation, I discovered that the VHF radiowas receiving a tone that should not have been there, and this tonewas of a sufficient level that it masked the data pulses that the dispatcherwas sending. On my spectrum analyzer, I saw and heard asteady tone imposed on the signal that I was looking at. Someoneelse must have been transmitting on our assigned VHF channel.This location was a heavy-trafficarea for the railroad, and the pressurewas on to correct the situation as soonas possible. This investigation was costingabout $50,000 an hour. Expensesincluded train delays and paying personnelovertime rates to manually throwthis critical switch. How could I findthe source of this interfering signal? Theantenna on the tower of the microwavehut was 350 feet high.The receiving equipment at thetrack-side location had an antenna thatwas just 8 feet above the tracks, so Itook the spectrum analyzer to the tracksidelocation, where I couldn’t see theinterfering signal. This step confirmedmy theory that the interference was notlocal and that the 350-foot-tall antennawas picking it up.Other staff members were checkingthe Federal Communications Commissionlicense base to see which transmitterswere near this location. I drovearound in the darkness for several hoursuntil I found a radio facility with anattached VHF antenna. We managedto identify the owner of this facility, andthe owner’s maintenance personnel metme there a few hours later. We had topay a premium for calling out the maintenanceworkers after normal workinghours. Nevertheless, when they checkedtheir equipment, they found that itwasn’t the source of the interference.The chief engineer at my companydecided to send a “tower man” to mylocation with a new antenna to replacethe one on the tower. As I waited—forfive hours—for his arrival, I mulledover my options because I knew areplacement antenna would not fix ourproblem.When the tower man arrived toclimb this tower, I told him to waitbecause I had a little test that I wouldlike to make first. He agreed. After all,climbing a 350-foot tower is no fun. Ihad him place the new antenna closeto the ground, supported by the securityfence that surrounded the microwavehut. We then ran coaxial cable intothe hut from this ground-level antennaand connected it to the VHF radio inthe microwave site. Bingo! The radiosonce again began to communicate, andnormal operations resumed. The microwavesite was on a hill overlooking thetrack-side antenna, resulting in a goodpath between the two antennas.Replacing the antenna on the towerwould not have solved our problem. Thesignal remained on the original antennaas I called every licensee I could findin a large radius of the microwave site.No one ever acknowledged having anytrouble with their radio systems, but theinterfering signal vanished three dayslater. My best guess was that a transmittersomewhere had a defective antenna,resulting in a high standing-wave ratio,which in turn caused the radiation ofspurious emissions. One of those emissionsfell into our frequency, and ourantenna on its tall tower picked up thisspur. Although you would normallyplace an antenna as high as possible,lower was better in this case.EDNEarl Schlenk is a retired engineer forBurlington Northern Railroad. He residesin Saint Louis, MO.DANIEL VASCONCELLOS54 EDN | FEBRUARY 3, 2011

www.globalfoundries.com/mwc

13.0Smart engineering simulationis now even smarter.In today’s world of smart electronics, signal integrity, power integrity and EMI analysisare mission critical. Engineers must deliver both reliable and innovative products in evershrinkingdevelopment cycles. For this reason, ANSYS is pleased to introduce ANSYS 13.0,our latest release which includes tools such as HFSS , SIwave and Ansoft Designer ® ,allowing engineers to identify and correct potential design or performance problems beforethey reach the marketplace.ANSYS 13.0 can help your design team achieve an even shorter total time to solution withseveral new and advanced features, including:• Greater accuracy and solution fidelity for very large electromagnetic structuresthrough the use of new solvers• Higher productivity through automated 3D electromagnetic model creation,resulting in more time to design and innovate• Increased computational efficiency, through the use of HPC, resulting ina dramatic speedup of wide-frequency band or parametric simulationsFor speed, reliability and accuracy, more product development leaders worldwide trustANSYS for simulation solutions to help create smarter products in less time.ANSYS 13.0 Advanced Fidelity. Higher Productivity. Leading Performance.For more information, visit www.ansys.com/electronics13or call us at 1.866.267.9724Ansoft is now part of the ANSYS platform of simulation solutions.