穩(wěn)固的控制電路設計外文文獻翻譯譯文3100多字(含：英文原文及中文譯文)文獻出處：DanM,MammanoB.DesigningStableControlLoops[J].ProcUnitrodeDesignSeminars,2002.英文原文DesigningStableControlLoopsDanMTheobjectiveofthistopicistoprovidethedesignerwithapracticalreviewofloopcompensationtechniquesappliedtoswitchingpowersupplyfeedbackcontrol.Atop-downsystemapproachistakenstartingwithbasicfeedbackcontrolconceptsandleadingtostep-by-stepdesignprocedures,initiallyappliedtoasimplebuckregulatorandthenexpandedtoothertopologiesandcontrolalgorithms.SampledesignsaredemonstratedwithMathcadsimulationstoillustrategainandphasemarginsandtheirimpactonperformanceanalysis.I.INTRODUCTIONInsuringstabilityofaproposedpowersupplysolutionisoftenoneofthemorechallengingaspectsofthedesignprocess.Nothingismoredisconcertingthantohaveyourlovinglycraftedbreadboardbreakintowildoscillationsjustasitsbeingdemonstratedtothebossorcustomer,butinsuringagainstthisunfortunateeventtakessomeanalysiswhichmanydesignersviewasformidable.Pathstakenbydesignengineersoftenemphasizeeithercut-and-tryempiricaltestinginthelaboratoryorcomputersimulationslookingfornumericalsolutionsbasedoncomplexmathematicalmodels.Whilebothoftheseapproachabasicunderstandingoffeedbacktheorywillusuallyallowthedefinitionofanacceptablecompensationnetworkwithaminimumofcomputationaleffort.II.STABILITYDEFINEDFig.1givesaquickillustrationofatleastonedefinitionofstability.Initssimplestterms,asystemisstableif,whensubjectedtoaperturbationfromsomesource,itsresponsetothatperturbationeventuallydiesout.Notethatinanypracticalsystem,instabilitycannotresultinacompletelyunboundedresponseasthesystemwilleitherreachasaturationlevel?orfail.Oscillationinaswitchingregulatorcan,atmost,varythedutycyclebetweenzeroand100%andwhilethatmaynotpreventfailure,itwillsultimatelimittheresponseofanunstablesystem.AnotherwayofvisualizingstabilityisshowninFig.2.Whilethisgraphicallyillustratestheconceptofsystemstability,italsopointsoutthatwemustmakeafurtherdistinctionbetweenlarge-signalandsmall-signalstability.Whilesmall-signalstabilityisanimportantandnecessarycriterion,asystemcouldsatisfythisrtquirementandyetstillbecomeunstablewithalarge-signalperturbation.Itisimportantthatdesignersrememberthatallthegainandphasecalculationswemightperformareonlytoinsuresmall-signalstability.Thesecalculationsarebasedupon?andonlyapplicableto?linearsystems,andaswitchingregulatoris?bydefinition?anon-linearsystem.Wesolvethisconundrumbyperformingouranalysisusingsmall-signalperturbationsaroundalarge-signaloperatingpoint,adistinctionwhichwillbefurtherclarifiedinourdesignprocedurediscussion.III.FEEDBACKCONTROLPRINCIPLESThebasicregulatorisshowninFig.3whereanuncontrolledsourceofvoltage(orcurrent,orpower)isappliedtotheinputofoursystemwiththeexpectationthatthevoltage(orcurrent,orpower)attheoutputwillbeverywellcontrolled.Thebasisofourcontrolissomeformofreference,andanydeviationbetweentheoutputandthereferencebecomesanerror.Inafeedback-controlledsystem,negativefeedbackisusedtoreducethiserrortoanacceptablevalue–asclosetozeroaswewanttospendtheefforttoachieve.Typically,however,wealsowanttoreducetheerrorquickly,butinherentwithfeedbackcontrolisthetradeoffbetweensystemresponseandsystemstability.Themoreresponsivethefeedbacknetworkis,thegreaterbecomestheriskofinstability.Atthispointweshouldalsomentionthatthereisanothermethodofcontrol–feedforward.Withfeedforwardcontrol,acontrolsignalisdevelopeddirectlyinresponsetoaninputvariationorperturbation.Feedforwardislessaccuratethanfeedbacksinceoutputsensingisnotinvolved,however,thereisnodelaywaitingforanoutputerrorsignaltobedeveloped,andfeedforwardcontrolcannotcauseinstability.Itshouldbeclearthatfeedforwardcontrolwilltypicallynotbeadequateastheonlycontrolmethodforavoltageregulator,butitisoftenusedtogetherwithfeedbacktoimprovearegulator’sresponsetodynamicinputvariations.ThebasisforfeedbackcontrolisillustratedwiththeflowdiagramofFig.4wherethegoalisfortheoutputtofollowthereferencepredictablyandfortheeffectsofexternalperturbations,suchasinputvoltagevariations,tobereducedtotolerablelevelsattheoutput.NotonlyisthisresultnowindependentofG,itisalsoindependentofalltheparametersofthesystemwhichmightimpactG(supplyvoltage,temperature,componenttolerances,etc.)andisdeterminedinsteadsolelybythefeedbacknetworkH(and,ofcourse,bythereference).NotethattheaccuracyofH(usuallyresistortolerances)andinthesummingcircuit(erroramplifieroffsetvoltage)willstillcontributetoanoutputerror.Inpractice,thefeedbackcontrolsystem,asmodeledinFig.4,isdesignedsothatGHandGH1overaswideafrequencyrangeaspossiblewithoutincurringinstability.WecanmakeafurtherrefinementtoourgeneralizedpowerregulatorwiththeblockdiagramshowninFig.5.Herewehaveseparatedthepowersystemintotwoblocks–thepowersectionandthecontrolcircuitry.Thepowersectionhandlestheloadcurrentandistypicallylarge,heavy,andsubjecttowidetemperaturefluctuations.Itsswitchingfunctionsarebydefinition,large-signalphenomenon,normallysimulatedinmoststabilityanalysesasjustatwostateswitchwithadutycycle.Theoutputfilterisalsoconsideredasapartofthepowersectionbutcanbeconsideredasalinearblock.Thecontrolcircuitrywillnormallybemadeupofagainblock–theerroramplifier–andthepulse-widthmodulator,usedtodefinethedutycycleforthepowerswitches.IV.THEBUCKCONVERTERThesimplestformoftheabovegeneralpowerregulatoristhebuck?orstepdown?topologywhosepowerstageisshowninFig.6.Inthisconfiguration,aDCinputvoltageisswitchedatsomerepetitiverateasitisappliedtoanoutputfilter.ThefilteraveragesthedutycyclemodulationoftheinputvoltagetoestablishanoutputDCvoltagelowerthantheinputvalue.Sinceweassumethattheswitchandthefiltercomponentsarelossless,theidealefficiencyofthisconversionprocessis100%,andregulationoftheoutputvoltagelevelisachievedbycontrollingthedutycycle.ThewaveformsofFig.6assumeacontinuousconductionmode(CCM)meaningthatcurrentisalwaysflowingthroughtheinductor–fromtheswitchwhenitisclosed,andfromthediodewhentheswitchisopen.TheanalysispresentedinthistopicwillemphasizeCCMoperationbecauseitisinthismodethatsmall-signalstabilityisgenerallymoredifficulttoachieve.Inthediscontinuousconductionmode(DCM),thereisathirdswitchconditioninwhichtheinductor,switch,anddiodecurrentsareallzero.Eachswitchingperiodstartsfromthesamestate(withzeroinductorcurrent),thuseffectivelyreducingthesystemorderbyoneandmakingsmall-signalstableperformancemucheasiertoachieve.Althoughbeyondthescopeofthistopic,theremaybespecializedinstanceswherethelarge-signalstabilityofaDCMsystemisofgreaterconcernthansmall-signalstability.ThereareseveralformsofPWMcontrolforthebuckregulatorincluding,?Fixedfrequency(fS)withvariabletONandvariabletOFF?FixedtONwithvariabletOFFandvariablefS?FixedtOFFwithvariabletONandvariablefS?Hysteretic(or“bang-bang”)withtON,tOFF,andfSallvariableEachoftheseformshavetheirownsetofadvantagesandlimitationsandallhavebeensuccessfullyused,butsinceallswitchmoderegulatorsgenerateaswitchingfrequencycomponentanditsassociatedharmonicsaswellastheintendedDCoutput,electromagneticinterferenceandnoiseconsiderationshavemadefixedfrequencyoperationbyfarthemostpopular.Withtheexceptionofhysteretic,allotherformsofPWMcontrolhaveessentiallythesamesmall-signalbehavior.Thus,withoutmuchlossingenerality,fixedfSwillbethebasisforourdiscussionofclassical,small-signalstability.Hystereticcontrolisfundamentallydifferentinthatthedutyfactorisnotcontrolled,perse.Switchturn-offoccurswhentheoutputripplevoltagereachesanuppertrippointandturn-onoccursatalowerthreshold.Bydefinition,thisisalarge-signalcontrollertowhichsmall-signalstabilityconsiderationsdonotapply.Inasmallsignalsense,itisalreadyunstableand,inamathematicalsense,itsfastresponseisduemoretofeedforwardthanfeedback.V.LINEARIZINGTHEBUCKCONVERTERWithidealcomponents,aswitchingregulatorisalinearcircuitforanygivenswitchcondition.Theconceptofaveragingcanbeappliedwhentheswitchingrateisfastwithrespecttotherateofchangeofanyotherparametersofinterest.Toquantifythis,wecansaythattheaccuracyoftheapproximationsisexcellentuptoone-tenththeswitchingfrequency,“prettygood”forone-third,andone-halfisthetheoreticallimitbasedontheNyquistsamplingcriterion.Fig.10showstheprocessofaveragingtheoperationofthecircuitbycombiningtheconditionwhentheactiveswitchisclosedwiththatwhenitisopen.TherelationshipbetweenthesetwoconditionsisthedutycycleoftheswitchinganditseffectisaccountedforthroughtheuseofahypotheticalDCtransformerwithaturnsratioofthedutycycle,d.Withthismodel,theprimarycurrentisnowILdandthesecondaryvoltageisVid.ThisDCtransformerisanartifactfromDaveMiddlebrookatCaltech.ItisnotarealcomponentbutitisvalidforyourSpicelibrary.Wenowhaveacontinuoussystembutitisnonlinearbecausethetransformerturns-ratio,d,isavariable-namelythecontrolvariable-andnotaconstant.VI.COMPENSATINGTHEBOOSTCONVERTERThebasicboostcircuitgainequationsandtheireffectareshowninFig.36.Notethatthegainamplitudecharacteristicappearstobeafirstordersystemwhich,intheory,shouldnotneedcompensationofthevoltageerroramplifiertoachievestability.OnecouldsayeitherthatthecurrentloopcompensatesthevoltagelooporthatC/Mcontrolreducesthesecond-ordersystemtoafirst-order,atleastuptothecutofffrequency,fC.Nevertheless,westilldohaveasecond-orderpowersystemwitharight-half-planezero,sothereisstillthepotentialforatotalof270oofphaseshift.Rememberthatthepoleorzerophaseshiftshowsupanorderofmagnitudelowerinfrequencythanachangeinamplitude.REFERENCES[1]D.M.Mitchell,“DC-DCSwitchingRegulatorAnalysis”,McGraw-Hill,1988,DMMitchellConsultants,CedarRapids,[2]D.M.Mitchell,“Small-SignalMathcadDesignAids”,(Windows95/98version),e/jBLOOMAssociates,Inc.,1999.[3]GeorgeChryssis,“High-FrequencySwitchingPowerSupplies”,McGraw-HillBookCompany,1984.[4]RayRidley,“AMoreAccurateCurrentModeControlModel”,UnitrodeSeminarHandbook,SEM-1300,AppendixA2.[5]LloydDixon,“ControlLoopDesign”,UnitrodeSeminarHandbook,SEM-800.[6]LloydDixon,“ControlLoopDesign–SEPICPreregulatorDesign”,UnitrodeSeminarHandbook,SEM-900,Topic7.[7]LloydDixon,“ClosingtheFeedbackLoop”,UnitrodeSeminarHandbook,SEM-300,Topic2.中文譯文設計穩(wěn)固的控制電路DanM本文研究的主要目標是為設計人員提供一個應用于開關電源反饋控制的環(huán)路補償技術的實用評論。從大體的反饋控制概念開始采用自頂向下的系統(tǒng)方式，并致使慢慢的設計進程，最初應用于簡單的降壓穩(wěn)壓器，然后擴展到其他拓撲和控制算法。利用MathCAD模擬演示樣本設計，以說明增益和相位裕度及其對性能分析的影響1引言確保提出的電源解決方案的穩(wěn)固性往往是設計進程中更具挑戰(zhàn)性的方面之一。沒有什么比讓您的精心制作的面包板突然進入瘋狂的振蕩，就像它向老板或客戶展示時那樣令人不安，可是為了避免這種不幸的事件需要一些分析，許多設計師以為這些分析是恐怖的。設計工程師所采用的路徑通常強調(diào)在實驗室中進行切割和實驗的實驗測試，或在運算機模擬中尋覓基于復雜數(shù)學模型的數(shù)值解。雖然這兩種方式都對反饋理論有了大體的了解，但通常能夠用最少的計算量來概念可同意的補償網(wǎng)絡。2穩(wěn)固性概念

圖1給出了至少一個穩(wěn)固性概念的快速說明。用最簡單的話來講，一個系統(tǒng)是穩(wěn)固的，若是受到某些源的擾動時，它對那個擾動的反映最終會消失。請注意，在任何實際系統(tǒng)中，由于系統(tǒng)將達到飽和水平，不穩(wěn)固性不會致使完全無窮的響應？或失敗。開關穩(wěn)壓器中的振蕩最多能夠在零和100％之間改變占空比，而且可能無法避免故障，但它最終會限制不穩(wěn)固系統(tǒng)的響應。

另一種可視化穩(wěn)固性的方式如圖2所示。雖然這用圖形方式說明了系統(tǒng)穩(wěn)固性的概念，但它也指出咱們必需進一步區(qū)分大信號和小信號穩(wěn)固性。雖然小信號的穩(wěn)固性是一個重要而必要的標準，但系統(tǒng)能夠知足這種要求，但在大信號擾動下仍然不穩(wěn)固。設計師記住，咱們可能執(zhí)行的所有增益和相位計算僅用于確保小信號穩(wěn)固性，這一點超級重要。這些計算是基于？只適用于？線性系統(tǒng)，開關穩(wěn)壓器是？按照概念？一個非線性系統(tǒng)。咱們通過在大信號工作點周圍利用小信號干擾進行分析來解決那個難題，這一點將在咱們的設計流程討論中進一步闡明。3反饋控制原則大體的調(diào)節(jié)器如圖3所示，其中一個不受控制的電壓源（或電流或功率）被施加到咱們系統(tǒng)的輸入端，預期輸出端的電壓（或電流或功率）將超級好受控。咱們控制的基礎是某種形式的參考，輸出和參考之間的任何誤差都會致使錯誤。在一個反饋控制系統(tǒng)中，負反饋被用來將那個誤差降低到一個能夠同意的值-接近于零，因為咱們想花費精力去實現(xiàn)。但是，通常情形下，咱們也想盡快減少錯誤，但反饋控制所固有的是系統(tǒng)響應和系統(tǒng)穩(wěn)固性之間的折衷。反饋網(wǎng)絡越響應，不穩(wěn)固風險就越大。在這一點上，咱們還應該提到還有另一種控制方式-前饋。利用前饋控制，控制信號直接響應于輸入轉變或擾動而形成。由于不涉及輸出感測，所以前饋比反饋精準度低，可是沒有延遲等待輸出誤差信號的產(chǎn)生，而且前饋控制不會致使不穩(wěn)固。應該清楚，前饋控制通常不足以作為電壓調(diào)節(jié)器的唯一控制方式，但它通常與反饋一路利用以改善調(diào)節(jié)器對動態(tài)輸入轉變的響應。用圖4的流程圖說明了反饋控制的基礎，其中目標是使輸出可預測地跟從參考，并將外部擾動（例如輸入電壓轉變）的影響降低到輸出端的可同意水平。這不僅與G無關，而且與系統(tǒng)中可能影響G（電源電壓，溫度，元件容差等）的所有參數(shù)無關，而僅由反饋網(wǎng)絡H決定（而且，固然，由參考）。請注意，H（一般是電阻容差）和求和電路（誤差放大器失調(diào)電壓）的精度仍然會致使輸出誤差。實際上，如圖4所示，反饋控制系統(tǒng)被設計成使得GH和GH1在盡可能寬的頻率范圍內(nèi)不會產(chǎn)生不穩(wěn)固性。咱們能夠通過圖5所示的框圖對咱們的廣義功率調(diào)節(jié)器進行進一步改良。在這里咱們已經(jīng)將電源系統(tǒng)分成兩個模塊-功率部份和控制電路。功率部份負責處置負載電流，而且通常較大，很重，而且會受到大范圍溫度波動的影響。按照概念，其開關功能是大信號現(xiàn)象，通常在大多數(shù)穩(wěn)固性分析中被模擬為僅具有占空比的雙向開關。輸出濾波器也被以為是功率部份的一部份，但能夠視為線性模塊?？刂齐娐吠ǔＳ稍鲆婺K-誤差放大器-和脈寬調(diào)制器組成，用于概念電源開關的占空比。4降壓轉換器上述通用電力調(diào)節(jié)器最簡單的形式是降壓？或下臺？拓撲結構的功率級如圖6所示。在此配置中，DC輸入電壓在施加到輸出濾波器時以某種重復速度切換。濾波器對輸入電壓的占空比調(diào)制進行平均，以成立低于輸入值的輸出DC電壓。由于咱們假設開關和濾波器組件是無損的，因此此轉換進程的理想效率為100％，而且通過控制占空比來實現(xiàn)對輸出電壓電平的調(diào)節(jié)。圖6的波形假設了持續(xù)導通模式（CCM），這意味著