13
< Back John Broskie's Guide to Tube Circuit Analysis & Design Next > 16 March 2014 ClassA Wizardry When we desire to label something as being of the highest order, quality, or degree, surpassing and superior to its competition, we seek out superlatives, such as excellent, exemplary, exceptional, firstclass, firstrate, foremost, nonpareil, matchless, outstanding, peerless, pukka, superb, superior, topgrade, topnotch, unmatched, and—of course—classA. And when we describe a classA power amplifier's sonic attributes, we often use those same superlatives. This is something of a coincidence and a small tragedy. In electronics, "classA" is not honor bestowed, but an account of an amplifier's mode of operation. It describes a dynamic process, wherein the output devices continue to conduct current in a meaningful way throughout the amplifier's peak output waveform. By way of a fortunate happenstance, when the two opposing output devices in a pushpull amplifier continually conduct over the entire output waveform, the distortion and output impedance falls. This pleases the ear. Just as our hand prefers to rub across smooth polished marble without encountering scratches or bumps, our ear prefers not to confront sonic discontinuities in a sound wave. ClassB power amplifiers present a scratchlike notch where the waveform switches polarity and neither output device conducts. In contrast, classAB amplifiers display two transitional drops, one above and one below the center of the waveform, the points where either output device cuts off, creating a sonic bump; the area of conduction overlap is called as gmdoubling. In a classA, push pull output stage, the area of gmdoubling covers the entire waveform, saving the ear from the two transition points. But a power amplifier's sonic signature is not entirely dependent on its output stage, as much can go wrong in the input and driver stages. Nonetheless, running a pushpull output stage in rich classA mode is a fairly safe bet for improving the amplifier's sound. When I in was college, I was adjusting the idle current on my solidstate power amplifier and I found the that the higher bias the better the sound, a simple linear relationship. Few things are linear. So, I wondered if instead of a straight line of sonic improvement, maybe the line was a curve, which might climb and then flatten horizontally or, worse, might reach a peak and then decline. I decided that I needed to do some dynamic testing. I bought a long shielded extension cord with three internal wires. I then wired up plug end inside my amplifier to its bias circuit; at the socket end, I attached a potentiometer. I then sat down and listened away to mono LPs, while adjusting the idle current from afar. The higher the idle current the better the sound—seemingly a straight line Kit User Guide PDFs Click image to download

Class-A & Current-Dumping & Class-S.pdf

Embed Size (px)

Citation preview

  • all the way up to smoke and burnt output transistors. Since that day, I haveperformedmanymoreexperimentsandIhave foundthat insomecases,a thinidle current created amore dynamic sound, rough but exciting and a fat idlecurrent, a smoother, more refined but overly damped sound. (The transistorfailingofbetadroopmustalsobefactoredinourreasoning.)

    Thus, strict logic and plain honesty demands thatwe acknowledge thatnot allsuperiorpoweramplifiersoperateinclassAandnotallclassApoweramplifiersare superior. Yes, indeed, I have heard some badsounding classA poweramplifiers. For example, all singleended amplifiers run in classA, there beingbut a single output device after all, but not all singleended amplifiers aresuperior,alas.

    PowerBoostedClassAAmplifiersMylastpostdealtwithgivingthevenerable300Bapowerboost.Wesawhowatransformercoupled, singleended amplifier could attach to an impedancemultipliercircuit(IMC),whichwouldaugmenttheamplifier'soutputwattagebythreefold and still reflect the desired load impedance to the single 300B. Theresultwasthe300Bdelivered12Wofthe36Wthatthe8ohmspeakerwouldseeand12Wofthe72Wthata4ohmspeakerwouldsee.

    Asfarasthe300Bwasconcerned,itwasworkingintoa2400ohmloadasfarasthe output transformer's secondary was concerned, its secondary was workinginto32ohms.Thustheoutputtransformer'swindingrationeededtobe10:1,asthe square root of 2400/24 is 10. The UBT1 singleended output transformer

  • holds a 14.14:1 winding ratio between its primary and the 8ohm tap on itssecondary and, a 10:1 ratio on its 16ohm tap. In addition, this fine outputtransformercaneasilyputout12W.Thatwasaquickrecap.Now,let'slookatthe32ohmtheIMCpresentstothesecondary.

    Headphonesoftensporta32ohmimpedance,butnoloudspeakersaresoldwithsuch ahigh impedance.This is something of a pity, as a 32ohm impedance ismuch more tubefriendly. Tubebased OTL power amplifiers strain to delivertheir current into 8ohm loudspeakers, whereas a 32ohm load would be fourtimes easier. Remember, most tube amplifiers are current limited, while mosttransistorbasedpoweramplifiersarevoltagelimited.Thus,justasatubebasedOTL power amplifier can deliver more power into a higher impedance load,transistorbasedpower amplifier candelivermorepower into lower impedanceloads.Butwhatifwearemoreconcernedwithdistortionthanwithpower,whichimpedancewouldproducelessdistortion?Theansweristhatbothamplifiertypeswoulddeliverlessdistortionwithahigherimpedanceload.

    An additional point to remember is that when it comes to pushpull poweramplifiers, it makes little sense to speak of the amplifier's class of operation,withoutfirstspecifyingtheloadimpedanceandapoweroutput.Forexample,allaudio pushpull power amplifiers run in classA, if the load impedance is 10k.And even the lowly classAB power amplifier within a cheap receiver runs inclassAforasmallportionofitspoweroutput,withan8ohmload.Forexample,if the transistorbased power amplifier runs an output stage idle current of100mA, then it can swing up to +/200mA of current before leaving classAoperationandenteringclassAB.Howmuchpowerdoesthattranslateintowithan8ohmload?Notmuch,only0.16W(oradtheadcopywriterwouldput"abig160 mW"). It would, however, equal 1W into a 50ohm load, but with 10Vpkoutputvoltageswing,notthe1.6Vpkinto8ohms.Incontrast,200mApeakintoa4ohm load causes the wattage to fall to 0.08W. In other words, one way toextendthewindowofclassAoperationistodriveahigherimpedanceload.Thisbringsusbackto32ohms.

    If apoweramplifier can swing+/32Vpk intoan8ohm load, then the speakerwillsee64Wwith4ohmspeakers,128W.Andifthesamepoweramplifierdrovea 32ohm speaker, the speakerwould see 16W of power. The next question is,Howhighwouldthepushpullpoweramplifier'sidlecurrentneedtobetoensureclassAoperationforthefull16Wintothe32ohmload?Theanswer isthattheidlecurrentmustatleastequalhalfofthepeakcurrentflowintotheload.Inthisexample, 1Apeak, as32Vdivedby8ohmsequals 1A, so the amplifier's outputstage must at least idle at 500mA. Now, 500mA may not sound like a lot ofcurrent,comparedtowhataclassApoweramplifierthatworksinto8ohmloadmust idle at, but it is about ten timesmore thanmost transistorbased poweramplifierruntheiroutputstageunder.AsIputitbackinissueNumber9oftheTubeCADJournal:

    Here you can see the painful inefficiency of Class A operation: it is not like twoburlyguysbreaking intoasweat tryingtoworktogether to liftasafeupaflightofstairsno,it ismoreliketwoburlyguysfightingitoutinatugofwarcontest,withthesafetiedtothemiddleoftherope.Eachpullsagainsttheother'sefforts.Iftheypullequally,noonewinsand the rope remainsstationary. Ifone lessenshispull,the ropemoves toward the other.So, too, it iswithClassA, pushpull operation:onedevice,whetheritbetransistor,MOSFET,pentode,ortriode,pullsagainsttheotheranditisthedifferencebetweenthesetwoamountsofcurrentflowthatequalstheamountthatisdeliveredintotheload.

    Inasingleendedamplifier,thetugofwarstillobtains:thesingleoutputdevicefights agains the output transfomer's primary inductance, which opposes achangeincurrentflowbydefinition.

    Still,buildingahighquality,classApoweramplifierthatworksinto32ohmloadis about four times easier thanbuildingone for an8ohm load, asmany feweroutput devices along with a smaller power transformer and heatsink will beneeded.Sure,butwhatgoodwouldahighquality,classApoweramplifierthatworks into 32ohm load be, if you cannot buy a highquality 32ohmloudspeaker? My answer, obviously enough, is that an impedance multiplier

    EmailfromGlassWareCustomers

    HiJohn,

    IreceivedtheAikidoPCBtodaythankyouforthefirstrateshippingspeed.WantedtoletyouknowthatthisissimplythebestPCBIhavehadinmyhands,barnone.Thequalityisfabulous,andyourdocumentationissuperb.Iknowyoudothisbecauseyouloveaudio,butIthinkyourpriceof$39isabitofagiveaway!I'msureyoucouldchargedoubleandstillhavehappycustomers.LookingforwardtobuildingtheAikido,willsendsomecommentswhenI'mdone!Thankyou,regardsGary

    MrBroskie,

    IboughtanAikidostereolinestagekitfromyousomedaysago,andIreceiveditjustthisMonday.Ihaveafewthingstosayaboutit.Firstly,I'mextremelyimpressedatthequalityofwhatI'vebeensent.Infact,thisisthehighestqualitykitI'veseenanywhere,ofanything.IhavenoideahowyoumanagedtofitallthisstuffinunderwhatIpaidforit.Second,yourshippingwaslightningquick.Justmoresatisfactioninthebag,there.Iwisheveryonedidbusinesslikeyou.

    SeanH.

    9Pin&OctalPCBs

  • circuitcanmakeyourexisting8or4ohmloudspeakersappearasa32ohmloadtosuchanamplifier.

    Inotherwords,wecouldbuildastellarsolidstatepoweramplifierthatcouldtaketheplaceofthe300Bbasedpoweramplifier,astheimpedancemultipliercircuitdoesn't care where its input signal comes from. The highquality solidstateamplifierwoulddeliverboththeIMC'sinputsignalandprovideaquarteroftheoutputpowerthattheloudspeakersees.Butwouldn'tdesigningahighquality,solidstate, classA power amplifier that drives 32ohm loads be difficult todesign?Yes, itwould,but thenallhighqualitypoweramplifiersaredifficult todesign. But you don't have to the do design work, as many gifted amplifierdesigners have alreadydone the hardwork for you. For example, JohnLinsleyHooddesignedafunlittleclassApoweramplifierlongago.

    J.L.HoodClassASingleEndedAmplifierJohnL.Hooddescribedinanarticleentitled"'AsimpleclassAamplifier"inthemagazine,WirelessWorld, in April 1969, his simple classA, solidstate poweramplifier.

    Thisamplifier isnotapushpulldesign,beingmoreof a singleendedeffort. (Ithink of it as being a distant relative of Nelson Pass's Zen amplifiers, as thebottom2N3055doesallmostallthepushingandpulling,whilethetop2N3055isanactiveloadthatapproximatesaconstantcurrentsource.)Itputoutabout10WandIactuallyheardaworkingexamplebackinthelate1970s.Ifoundthesoundextremely pleasant, but a bit too tube sounding, in the bad sense of tubesounding, a bit too rounded and too bandwidth limited. With modern, highquality parts, however, that judgmentwould probably not hold true today. Butwerewetobuildittoday,wecouldalsoalterthetopology.Forexample,hereisa

    Highquality, doublesided, extra thick, 2oztraces,platedthroughholes,dualsetsofresistorpads and pads for two coupling capacitors.Stereo and mono, octal and 9pin printed circuitboardsavailable.

    DesignedbyJohnBroskie&MadeinUSA

    AikidoPCBsforaslittleas$24

    http://glassware.stores.yahoo.net/

    SupporttheTubeCADJournal

    &

    getanextremelypowerfulpushpulltubeamplifiersimulatorfor

    Only$19TCJPushPullCalculator

    Version2

    Clickonimagestoseeenlargements

  • simplesingleended,classA(butthenIrepeatmyself)amplifier.

    TheNchannelpowerMOSFETworks intoaconstantcurrentsource(CCS)andthe two transistors define the input stage.What could be simpler. Okay, otherthan a Nelson Pass Zen amplifier? The two transistors create a differentialamplifier,whichprovidessignalgainandallowstheeasyapplicationofnegativefeedback.Thename"Bastode"isReneJaeger'sinspiration,thecircuitbeinglikeacascode, but not really. It could be replaced by a conventional PNPbaseddifferential amplifier whose coupled emitters would connect to a CCS thatterminatedintotheB+voltage.*Butthebastodeissimplerandjustaseffective.

    Speakingofsimple,manytimesIhavebeenouthavingbreakfastorabeerwithabuddyandIdrawanequallysimplecircuitonapapernapkinforhim(ofcourseitisahim,assoveryfewwomenwouldhaveanyinterestinanyelectroniccircuityes,Iknowtheyexist,asIgetemailfromthem,buttheyareasrareasunicorns).Helovesit.Heproclaims,"IwanttobuildthisIcanbuildthis!Whendowestart? " I then proceed to draw the following schematic for him, which showshowsthecompleteamplifier,sanspowersupplyandfusing.

  • Heisdisheartened."Wheredidthesimplecircuitgo?"hebemoans.Itellhimthatit is still there, but that an actually working amplifier will require moresupporting devices, such as the protecting zener for the IRF230's gate and the100pFfeedbackcapacitortolimithighfrequencybandwidthtobelow1MHz.Hedoesn't likemyanswerandIcanunderstandwhy.Perhaps, if Ihadonlyaddedoneessentialdetailatatime,soattheendofthreebeersandtennapkinslater,hecould have babystepped himself to groking the complete, fleshedout circuit.Perhapsnot,asaddingmoretreesjustmakestheforestdisappear.

    Anothergroupofaudiophileswouldhaveimmediatelysniggered,stunnedbymyelectronicinsensitivity,mybadaudiophilemanners.Theywouldwonder,"WhatwasIthinkingwhenIaddedthecouplingcapacitorattheoutput?Don'tIknowthatall capacitorsdegrade theaudio signal?" Iwould thenpoint that allpowersuppliesmustalsodegradetheaudiosignals,astheyareverymuchinthesignalpath that the speakers see. For this group, I would then draw the followingschematic,whichwouldmakethemhappy,astheycannotseethepowersupplycapacitors,sotheyignorethementirely.

    TCJPPCVersion2

  • TheOpAmpbasedDCservolooptakescareoftheDCoffsetattheoutput,whichallowedustolosethebigcapacitorattheendthefeedbackresistorpair.

    Inalltheseexamples,theidlecurrentwassetto1.6A,whichshouldgetuscloseto10W into 8ohm speakers. In spite of the lowpower output, this amplifierwilldissipate a lot of heat, as the 1.6A is against 30V, which creates 48W ofdissipation,solargeheatsinkswillberequired.Butwitha32ohmloadandagoalof32Vpkofvoltageoutputswing,an80VB+(or+/40V)wouldbeneededandtheidlecurrentcouldbesetto1A.

    Now,80V isa lotof voltage.Enough, in fact, to runsome tubes.The followingamplifier is an Aikido design, although itmay not look like it. The totempoleconfigurationofthetwo6DJ8triodesdeliversagainequaltohalfofthemu,orroughly 16.Theyalsodefinea tworesistancevoltagedivider, so50%of theB+ripplewill leakout at thebottom triode'splate.APSRRof 6dBdoesn't soundthatstellaranditisn't,butthesolidstatehalfofthecircuittakescareofthat.

    ImprovementsRebuiltsimulationengineCreatereportsasPDFs*MoreGraphs2D/3D*HelpsystemaddedTargetidlecurrentfeatureRedesignedarraycreationTransformerprimary&secondaryRDCinclusionSaveuserdefinedtransformerdefinitionsEnhancedresultdisplayAddedarrayresultgrid

    *Userdefinable

    TCJ PushPull Calculator has but a singlepurpose: to evaluate tubebased output stagesby simulating eight topologies (five OTL andthree transformercoupled) actual performancewith a specified tube, power supply and biasvoltage, and load impedance. The accuracy ofthe simulation depends on the accuracy of thetubemodels used and the tubemathmodel isthe same True Curves model used inGlassWare's SE Amp CAD and Live Curvesprograms, which is farmore accurate than theusualSPICEtubemodel.

    DownloadorCDROMWindows95/98/Me/NT/2000/XP

    Formoreinformation,pleasevisitourWebsite:www.glassware.com

    Topurchase,pleasevisitourYahooStore:

    http://store.yahoo.com/glassware

  • The two 1k feedback resistorsdonot terminate into ground, but theB+,whichgivethesolidstatehalfagainof2(+6dBandthusthetotalgainoftheamplifierequalsthemuofthetriodesused)andeliminatestheB+ripplefromtheoutput.SeeBlogNumber141formoreexplicationofhowthismiracleoccurs.

    ButifyouarewillingtouseaclassA,pushpullamplifier,youneedonlychooseanexisting classAB,pushpulldesignandbumpup its idle current to500mAyoumightneedtoalsoadd,ortosubtract,outputdevices.

    DesignsfromErnoBorbely,BobCordell, JohnCurl, RodElliott, JeanHiragra,JohnLinsleyHood,NelsonPass,DouglasSelf,Randy Slone, andmany others.(SorryifIleftyouoffthelistandIknowthatIhaveleftmanyoffthelist,asIjustthoughtofA.R.Bailey,JamesBongiorno,DanielMeyer,P.L.Taylor...)

  • BobCordell's book,Designing Audio Power Amplifiers, is a steal at $37.30 atAmazon. Bob covers every aspect of designing, not just building, a solidstatepoweramplifier.RandySlonepassedawayrecently,sadly.Hisbook,HighPowerAudio Amplifier Construction Manual, is also a must read and a bargain at$33.16.

    Slone's perspective is different from Cordell's, being more focused on actualworkingpoweramplifiersinsteadoftheoryordesignphilosophy,sothethetwobookscomplementeachothernicely.

    Ofcourse,thisisnotwhatIwoulddo,asIenjoydesigningpoweramplifiers.Infact,Iwouldprobablyexperimentwithasingleended,classA,solidstatepoweramplifier,but then the500mAidlecurrentwouldhave tobebumpedup to1A.Moreover,IwouldbeinclinedtousepowerMOSFETs,ratherthantransistorsintheoutputstage.Or,Imightdesignahybridoutputstage,whichranbothtriodesandsolidstatedevicesinparallel.Indeed,withthe32ohmload,analltubeOTLpoweramplifierwouldnotbetoodifficulttodesign.

    QuadCurrentDumpingTopologySeveralreadershavecommentedthatthepowerboostingideasurelookslikethe

  • oldQuadcurrentdumpingconcept,whereinasmallclassAamplifierworkedinparallelwithapowerful,butdirtyclassBamplifier.

    QuadCurrentDumpingAmplifier

    Trueenough,bothmyIMCapproachandtheQuaddesignsharebothacommongoal, better sound on the cheap, and use similar components, a small classAamplifierandanotclassAamplifier.Buttheydifferinthatmyapproachdoesnotextend a global negative feedback loop to where the speaker attaches, whichallowsme to treat thesmallclassAamplifierand theIMCasdiscretepiecesofequipment that each reside in its own chassis. In contrast, the Quad currentdumpingamplifierinsinglecircuitthatmustbeenclosedititsownchassis.

    Longago,IwassotakenbytheQuadcurrentdumpingconceptthatItranslatedit into a tubebasedpower amplifier.Unfortunately, I haven't been able to findtheschematic.IfIremembercorrectly,IhadplannedonusingasingleEL84inasingleendedconfigurationalongwithapairofEL34outputtubesinaleanclassAB,pushpulloutputstage.

    A.M.Sandman'sClassSI have written about Dr. Sandman's wonderfully clever distortioneliminationschemesbefore,suchasinBlogNumber20andNumber22.(Ievenreceivedanemail from Dr. Sandman, which was a treat for me.) One of his distortionelimination topologies uses a small classA amplifier and a classB amplifier,whichhecallsclassS.

    The small classA amplifier delivers its clean output signal to the classBamplifier.TheclassBamplifierdoes thebest job it canand theasmallclassA

  • amplifiercleansupthemesscreatedbyclassBamplifier.Thespeakerthenseesacleansignal.Ihaveseenhistopologydescribedasbeinganimpedancemultipliercircuitofsorts,whichitissortof,butnotreally.Letmeexplainifbothamplifierswere perfect, giving rise to zero distortion, the load that the small classAamplifierwouldseewouldapproachinfinity.

    Note how the classA amplifier delivers no current into the load.On the otherhand,iftheclassBamplifierputout100%distortion,itwasturnedoffinotherwords, then the small classA amplifier would see an 8.1ohm load, so noimpedancemultiplying would obtain. In other words, the amount of work thesmallclassAamplifiermustperformdependsonhowdirtytheclassBamplifieris,asitmustdealwiththedelta,thedeparturefromperfectionproducedbytheclassBamplifier. If the classBputsout0.1%THD, the small classAamplifierwill have littlework to do if the classB amplifier puts out 10%distortion, thesmallclassAamplifierwillhave100timesmoreworktodo.

    Hereisaverysimpleexample:imaginethattheclassBamplifiersuffersfroma200mVDCoffset,whichcanbeviewedasaDCdistortion.TheclassAamplifier'snegative feedbackwill force itsoutput to0V,butbydoing so, itmustdeliverasteady2AofcurrentflowtocountertheclassBamplifier'sDCoffset.Effectively,theloadimpedancetheclassAamplifierseesis0.1ohms.Inaddition,theclassAamplifier's output impedance must be extremely low, so you can forget aboutusing a tubebased OTL amplifier in place of the classA amplifier. Oneworkaround is to replace both 0.1ohm resistorswith 1ohm resistors. But this

  • will increase the output impedance that the speaker sees to about 1 ohm. AsRobert Heinlein like to say TANSTAAFL (There ain't no such thing as a freelunch).

    Do Not get me wrong here: I think that Dr. Sandman's classS approach isbrilliant. Just bear in mind that the small classA amplifier must be trulyexemplaryandtheclassBshouldbeascleanaspossible.(Infact,IwoulduseaclassABamplifier,suchastheLM3886,inplaceofatrueclassBamplifier,asmygoalwouldbethepurestamplification,notthelowestdissipation.)

    NextTimeOTLamplifierwillbethetopic.

    *ConventionalDifferentialInputstage

    Ofcourse,wecouldgetmorefancyanduseacurrentmirrorinsteadofthelone510ohmcollectorresistorand/oraddemitterresistors.

    For thoseofyouwhostillhaveoldcomputers runningWindowsXP(32bit)oranyotherWindows32bitOS, Ihavesetupthedownloadableavailabilityofmyoldoldstandards:TubeCAD,SEAmpCAD,andAudioGadgets.ThedownloadsareattheGlassWareYahoostoreandthepriceisonly$9.95foreachprogram.

    http://glassware.stores.yahoo.net/adsoffromgla.html

  • SomanyhaveaskedthatIhadtodoit.

    WARNING:THESETHREEPROGRAMSWILLNOTRUN

    UNDERVISTA64BitorWINDOWS7&8oranyother64

    bitOS.Idoplanonremakingalloftheseprogramsinto64bitversions,butitwillbeahugeordeal,asprogrammingrequiresvastchunksofnoisefreetime,somethingvery rare with children running about. Ideally, I would love to come out withversionsthatrunoniPadsandAndroidOStablets.

    //JRB