p1 Ex Theory

38
 Introduction to Tube Amplifier Theory: 10.02.15 Featuring the AX84 P1-eXtreme Amplifier  by David Sorlien, revised and updated by Stephen Keller The P1-eX amplifier is a simple three-stage vacuum tube electric guitar amplifier. As outlined in Fig. 1, it consists of two preamp stages driving a power amp stage. Depending on the choice of output tube, this amp is can deliver between 7 and 15 watts into a 4-, 8-, or 16-ohm load. Fig. 1: P1-eX block diagram Placed between the two preamp stages is a preamp volume control. In a similar manner, the bass, middle, and treble tone controls and a master volume control are placed between the preamp and the final power amp stage. The guitar amplifier performs two primary functions: One, it amplifies the small voltages and currents produced by the guitar pickup into a signal powerful enough to drive a speaker. Two, it shapes the frequency response, tonality, and distortion characteristics of the raw guitar signal into a form pleasing to the musician. To better grasp how a guitar amplifier accomplishes these functions, let's take a walk through the inner workings of the AX84 P1-eXtreme amp (P1-eX). For reference, there is copy of revision 06.03.16 of the schematic provided in Appendix 1 at the end of this document. To fully understand how a guitar amp works, even a simple one like the P1-eX, you must know how to read schematic diagrams, and understand what things like resistors and capacitors are. You also need some knowledge of basic algebra and electronic theory. HOW THE HECK DO TUBES WORK ANYWAY? So you run to the bookstore and buy a book on basic electronics. Hnm..., no tube chapter. Even some good tube amp books fail to explain how tubes work. Many different types of tubes have been developed over the years, but guitar amps generally use only three types: diodes, triodes and pentodes. P1-eX Theory 1

description

tube amp p1

Transcript of p1 Ex Theory

  • Introduction to Tube Amplifier Theory: 10.02.15Featuring the AX84 P1-eXtreme Amplifier

    byDavidSorlien,revisedandupdatedbyStephenKeller

    TheP1eXamplifierisasimplethreestagevacuumtubeelectricguitaramplifier.AsoutlinedinFig.1,itconsistsoftwopreampstagesdrivingapowerampstage.Dependingonthechoiceofoutputtube,thisampiscandeliverbetween7and15wattsintoa4,8,or16ohmload.

    Fig.1:P1eXblockdiagram

    Placedbetweenthetwopreampstagesisapreampvolumecontrol.Inasimilarmanner,thebass,middle,andtrebletonecontrolsandamastervolumecontrolareplacedbetweenthepreampandthefinalpowerampstage.Theguitaramplifierperformstwoprimaryfunctions:One,itamplifiesthesmallvoltagesandcurrentsproducedbytheguitarpickupintoasignalpowerfulenoughtodriveaspeaker.Two,itshapesthefrequencyresponse,tonality,anddistortioncharacteristicsoftherawguitarsignalintoaformpleasingtothemusician.

    Tobettergrasphowaguitaramplifieraccomplishesthesefunctions,let'stakeawalkthroughtheinnerworkingsoftheAX84P1eXtremeamp(P1eX).Forreference,thereiscopyofrevision06.03.16oftheschematicprovidedinAppendix1attheendofthisdocument.Tofullyunderstandhowaguitarampworks,evenasimpleoneliketheP1eX,youmustknowhowtoreadschematicdiagrams,andunderstandwhatthingslikeresistorsandcapacitorsare.Youalsoneedsomeknowledgeofbasicalgebraandelectronictheory.

    HOWTHEHECKDOTUBESWORKANYWAY?

    Soyouruntothebookstoreandbuyabookonbasicelectronics.Hnm...,notubechapter.Evensomegoodtubeampbooksfailtoexplainhowtubeswork.Manydifferenttypesoftubeshavebeendevelopedovertheyears,butguitarampsgenerallyuseonlythreetypes:diodes,triodesandpentodes.

    P1eXTheory 1

  • VACUUMDIODES

    Let'sstartwiththevacuumdiode,becauseitisthesimplesttypeofvacuumtube.Insidetheglassbottle,thereareafewmetalparts:thefilament,thecathode,andtheplate.Thefilamentissometimescalledtheheater,becausethatisexactlywhatitdoesitheatsthecathode.Likethefilamentinalightbulb,thetubefilamentisathinlengthofwirethatgetshotwhenelectricityflowsthroughit.

    Asthefilamentheatsthecathode,itemitselectronsfromitssurfaceinaprocesscalledthermionicemission.Withthecathodewarmedup,acloudofelectronsgathersaroundthecathode,asshowninFig.2(b).Theelectronsrepresentedinthisdrawinghavenoplacetogobecausetheplateisnegativelychargedwithrespecttothecathode.Inmattersofelectricity,likechargedobjectsrepeleachother,sothenegativelychargedelectronsjusthangaroundthecathodelikefarmboysdreamingofbigcitylights!ButconsiderFig.2(c).Thingschangewhentheplateismadepositivewithrespecttothecathode.Nowtheelectronsstreamofffromthecathodetotheplate,causingcurrenttoflow1intheplatecathodecircuit.Ifyouthinkaboutthisbehavior,youseethatthevacuumdiodefunctionsasaonewayvalvethatallowscurrenttoflowinonlyonedirection.Itisthisfunctionthatmakesitusefulforconvertingalternatingcurrent(AC)intodirectcurrent(DC).

    Invacuumtubeamplifiers,vacuumdiodesareoftenusedtorectifytheAClinevoltageintoaDCvoltagesuitableforpoweringtheplatesofothertubesinthecircuit.TypicalexamplesincludetheEZ80andthe5U4types.Solidstatesilicondiodesbehaveinasimilarmannerandareoftenusedinplaceofvacuumdiodesintubeampsbecauseoftheirlowercostandimprovedreliability.Yet,therearesomesonicgainstobehadfromusingavacuumrectifierinpushpullpoweramplifiersduetothegreateroverallloaddependentvoltagedropintroducedbytherectifier.Thisissometimescalledsag.Singleendedamplifiers,suchastheP1eX,presentroughlythesameloadtothepowersupplywhethertheyareatidleorrunningflatout,sothevoltagesagfromthetuberectifierdoesnotoccur.Consequently,avacuumtuberectifieraddscostandcomplexity,butlittlesonicvalue,toampsliketheP1eX.WhilesomesomeearlyversionsoftheP1ampusedtuberectifiersintheirpowersupplies,theP1eXreliesonsolidstatediodesformainsrectification.Besidestheirfunctionasrectifiersinpowersupplies,diodesalsoareusedinmanyothertypesofcircuits,includingautomaticvolumecontrols,compressor/expandercircuits,andradiosignaldetectors.Wewon'tspendanymoretimeonvacuumtubediodes,thoughwewilldiscusstheP1eXpowersupplylateron.

    1 Anoteonconventionalelectriccurrentvs.electronflow:Conventionalelectriccurrentisdefinedastheflowofpositivechargeinacircuit.Thisisbecausetheconventionaldefinitionofelectriccurrentwasdevelopedbeforethediscoveryoftheelectronanditsnegativecharge.Backintheearlydays,electronicsexperimentersknewthatinvisiblebitsofchargeflowedthroughthewireinacircuit,buthadnowayofdeterminingthedirectionthebitstraveled.Theytooktheirbestguess.Bzzzzt!Wrong!Bythetimethenotsoancientsrealizedthemistake,itwastoolate.Byconvention,mostelectronicstextsrefertoelectriccurrentandmeanthenegativeflowofelectroncurrent.Thereareexceptionsinelectronicstextsaswellasintextsfromotherfields.Sometreatcurrentasflowingfrompositivetonegativeandsomeasflowingfromnegativetopositive.Additionally,somematerialsallowcurrentflowsinbothdirectionsbutcarriedbydifferentparticles.Donotletthissituationconfuseyou.Fromtheperspectiveofvacuumtubedesign,thedirectionofthecurrentflowrarelymatters,aslongasyoutreatitconsistently.Theimportantthingtorememberisthatcurrentisameasureofthevolumeofelectronsflowingpastaparticularpointinacircuit.Thegreaterthevolume,themoreusefulwork(ordamage)youcandowithit.

    P1eXTheory 2

    Fig.2:Thediode,forwardand

    reversedbiased

  • VACUUMTRIODES

    Thingsgetmuchmoreinterestingwhenyouaddanelectrodebetweenthecathodeandplateofavacuumdiode,creatingatriode(Fig.3).The12AX7commonlyfoundinguitaramplifierscontainstwotriodesinsideitsglassenvelope.Thisnewelectrodeiscalledthecontrolgrid(orgrid,forshort)andisusuallyconstructedofameshofthinwirespositionedbetweentheplateandcathodeveryclosetothecathode.Ifyouweretoconnectthegridtothecathode,thetubewouldbehavemuchlikeadiode.Mostoftheelectronswouldflowrightpastthegridontheirwaytotheplate.Donottrythis.Mosttriodesarenotdesignedtobeoperatedinthismanner.

    Whenchargeonthegridismademorenegativewithrespecttothecathode,theelectronflowfromcathodetoplatestartstogetpinchedoff.Fig.4illustratesthisproperty.Thedeviceinthisdrawingisconceptual,sodon'tgohuntingforarealtriodethatproducestheseresults.InFig.4(a)thegridvoltageis1V,soitisnotveryrepulsivetoelectronsandmanyofthemsuccessfullytraveltotheplate,whichallowstheplatecircuittodraw10mA.Asthegridvoltagebecomesmorenegativewithrespecttothecathode,asshowninFig.4(b),moreoftheelectronsarerepelledbythegrowingnegativechargeonthegrid,andthenumberofelectronsthatpassthroughthegridisreduced.Inthisexample,thecurrentintheplatecircuitfallsto5mA.Asthegridvoltageisdrivenevenmorenegativethanthecathode,theelectronflowisreducedfurtheruntilatsomenegativevoltagethecurrentcutsoffcompletely,asillustratedinFig.4(c).Whenthetriodeiscutofflikethis,nocurrentcanflowintheplatecircuit,andincreasinghownegativethegridiswithrespecttothecathodebeyondthispointhaslittleeffectonthebehaviorofthetriode.

    EveryACsignalcyclesupanddownwithrespecttosomelevelofDCbiasvoltage.Ifyourhouseislikemine,themainACvoltageisabout125VRMS(rootmeansquare),meaningthevoltagecyclesbackandforthinasinewavefromapositivepeakof+177Vto0Vtoanegativepeakof177V.ThatzeroinmiddleindicatesithasaDCbiasvoltageof0V(oritshouldhaveifyourwiringiscorrect).Inasimilarmanner,thesignalonthegridhasaDCbiasvoltage.Itcouldbe0V,likeyourhousemains,oritcouldbesomepositiveornegativeDCvoltage.ThatDCbiasvoltage,whenappliedtothegrid,iscalledgridbias.ItisalsosometimesreferredtoasthezerosignalgridvoltagebecauseitisthevoltagethegridsitsatwhenthesignalisreducedtozeroACvolts.AnegativeDCbiasvoltagesufficienttoreduceplatecurrenttozeroiscalledthecutoffbias.

    Whenthegridvoltagebecomesthesameasthecathodevoltage,thegridbiasissaidtobeat0V,andtheonsetofgridconductionoccurs.Gridconductionmeansthatcurrentflowsinthegridcircuitaswellasintheplatecircuit.Althoughmostsmallsignaltriodesdonottoleratesustainedoperationinthe

    P1eXTheory 3

    Fig.4:Currentfallsasgridvoltagegoesmorenegativewith

    respecttocathode

    Fig.3:VacuumTriode

  • gridconductionregion,manypowertriodesandothersortsoftubesaredesignedforit.Continuedincreasesingridbiasvoltageallowevenmorecurrenttoflowintheplateandgridcircuits,uptoalimit.Thepointwhereplateandgridcurrentsnolongerincrease,regardlessofincreasesingridbiasvoltage,iscalledthesaturationpoint.Thesaturationpointisdeterminedbythecathode'sabilitytoemitelectronsatagivenoperatingtemperature,whichisgovernedbythecathode'smaterialandconstruction.

    Betweenthecutoffpointandtheonsetofgridconduction,atriodebehavesasaroughlylineardevice.Smallsignaltriodesaretypicallydesignedtooperateinthislinearregion.Betweengridconductionandsaturation,atriode'sbehaviorbecomesverynonlinearbecausethegridisdrawingoffprogressivelymoreofthecurrentflow.Foragiventriodestage,weensureoperationinthislinearrangebysettingthezerosignalgridbiasabouthalfwaybetweenthecutoffpointandtheonsetofgridconduction.Sowhatdoesitmeantosetthebiasvoltageofastageandhowdowedothat?Theanswertothatquestionisinthenextsection.

    BIASINGA12AX7GAINSTAGE

    Beforewebeginthediscussionofsettingthezerosignalbiasvoltage,let'sestablishthreecommonabbreviationsthatwillappearintheequationsbelow:

    E, meaningelectromotiveforcemeasuredinvolts.

    I, meaningelectriccurrentmeasuredinamperes.R, meaningelectricalresistancemeasuredinohms.

    TherelationshipamongthesethreeelectricalpropertiesisdefinedbyOhm'slaw:

    E= IR whichcanalsobeexpressedas I=ERorR=

    EI

    Don'tletthemaththrowyou.Ohm'slawbasicallysaysthatthecurrentthroughagivencircuitelementtimestheresistancepresentedbythatelementgivesthevoltagedropacrossthatelement.ThecoolthingaboutOhm'slawisthatifyouhavetwoofthevalues,youcanfigureouttheremainingvalue.Supposeyoumeasuredthevoltagewithrespecttothecircuitgroundjustbeforeandjustaftera1Kohmresistorandobserveda12voltdifferencebetweenthetwo.Theresistorissaidtodrop12volts.YoucanplugthevoltageandresistancevaluesthatyouknowintoOhm'slaw,asfollows:

    I=ER=

    121000

    =.012 A=12mA

    Youcanseethat12mAflowsthroughtheresistor.Thatmeansthatthecircuitfollowingtheresistanceisdrawing12mA.Ohm'slawisapowerfultoolthatwillhelpusanalyzehowtheP1eXamplifierfunctions.

    P1eXTheory 4

  • TakealookatFig.5.ItshowsthefirstpreampstagefromtheP1eXschematic.Thisisanexampleofaresistancecapacitancecoupled,groundedcathodevoltageamplifier.Itoperatessuchthatthevoltageacrossplateloadresistor(R22)isanamplifiedandinvertedcopyofthegridsignalvoltageatpin7.ThesignalenterstheamplifierviaJ2,whereyouplugintheguitar.R25providesaloadacrosswhichthesignalfromthepickupscanform.ThegroundedcirclearoundtheleadfromtheinputjacktoR24indicatesshieldedwireintendedtopreventelectromagneticfield(EMF)interferencefrommixingwiththesignal.R24isagridstopperresistoranditspurposeistorolloffhighfrequenciestofurtherlimitEMFinterference.Typically,itismounteddirectlyonthetubesockettofurtherreducethechanceofEMFnoise.SomedesignersprefertoreducethevalueofR24becauseofitseffectonthehighfrequencyresponseofthepreamp.Othersremoveitalltogether.Insuchcases,itisimportantthatthegroundattheinputjackservesasthesignalgroundtochassisgroundtiepoint.Otherwise,theleadfromtheinputjacktothefirstpreampgridcanactasaradioantenna.

    Let'sthinkaboutwhatishappeninginthatcircuit.FirsttakealookattheIpEpgraphfromthe12AX7datasheet,showninFig.6.Itlooksconfusingatfirst,butthisgraphcantellyouallsortsofthingsaboutthetube'sbehaviorinacircuitonceyoulearnhowtounpackit.Thecurvedlinesthatslopeupwardtotherightarecalledplatecurves.Eachonerepresentsthebehaviorofthetriodewhenthegridisataspecificbiasvoltage(0V,0.5V,1Vandsoforth).RecallinFig.4,thatasthegridvoltagebecamemorenegative,theplatecurrentdropped.Youcanseethatbehaviorinthe12AX7triode.Ataconstant150Vontheplate,theplatecurrentisabout3.2mAwhenthegridisat0V,about1.2mAwhenthegridisat1V,andabout0.2mAwhenthegridisat2V.Ataconstant150Vplatevolts,thetriodecutsoffwithagridbiasvoltagesomewherebetween2Vand2.5V.

    P1eXTheory 5

    Fig.6:12AX7IpEpgraph

    Fig.5:StageoneoftheP1eXpreampisa12AX7gainstage

  • ThesignalcomingintothegridofthetriodeisanACsignal.ItsvoltageisswingingupanddownaroundtheDCbiasvoltageatthefrequencyofthenotesbeingplayed.It'sgoingpositivewithrespecttothebiasvoltageandthennegativewithperhapsa1Vto2Vpeaktopeakswing.Ifthisgridsignalswingssufficientlynegativetocausethecurrentintheplatecircuittocutoff,thentherewouldbenovoltagedropinR22(rememberOhm'slawE=R*I),andthevoltageattheplatewouldequalthepowersupplyvoltage,whichis274VforstageoneoftheP1eXpreamp.Youcanseethatagridvoltageofapproximately3.6Visnecessarytocausethetriodetocutoffwhentheplateisat275V.Gridconductionoccurswhenthegridvoltageis0V,sowewantabiaspointaboutmidwaybetweenthesetwopoints;somewherebetween1Vand2Vwouldbefine.

    Whennosignalispresentattheinput,thegridofV4BisconnectedtogroundthroughresistorsR24andR25,sotheDCvoltage(Eg)atV4B'sgridis0V.Theproblemishowtomakethegridvoltagenegative.WecoulduseafixednegativeDCsourcelikeabatteryoraseparatepowersupply.Butwedon'thavetoresorttothosemeans.Remember,thegridonlyhastobenegativewithrespecttothecathode.IfwecanraiseV4B'scathodevoltagetoabout1.4V,wecanaccomplishthesamethingasloweringthetriode'sgridfrom0Vdownto1.4V.

    Oneconvenientwaytodothisistoplacearesistorbetweenthecathodeandground.Thisraisesthecathodeabovegroundbythevoltagedroppedacrossthecathoderesistor.Thistechniqueforsettingthebiasvoltageiscalledselfbiasorcathodebias.LookbackattheschematicinFig.5.ThevoltageacrossR22is:

    ER22

    =274V181V=93V

    UsingOhm'slaw,thecurrentthroughR22is:

    I R22=E R22R22

    =93V

    100K=0.93mA

    So,0.93mAflowsintheplatecircuitofV4B.Thecurrentinthecathodecircuitofatriodeisthesumofthecurrentsintheplateandgridcircuits.Becauseourtriodeisoperatingbelowtheonsetofgridconduction,nearlyzerocurrentflowsinthegridcircuit.Forpracticalpurposes,thecurrentflowinginthecathodecircuitisequivalenttothecurrentintheplatecircuit.Therefore,usingOhm'slawagain,wecanmultiplyIR22timesthecathoderesistance(Rk)todetermineV4B'scathodevoltage.RkintheschematicisidentifiedasR23andhasavalueof1.5Kohms:

    Ek V4B= I R22RkV4B =0.93mA1.5K=1.40V

    Rememberthatthebiasisthegridvoltagereferencedtothecathodevoltage.WecaneasilycalculatethebiasvoltageofV4BbysubtractingcathodevoltageofV4Bfromitsgridvoltage.Recallabovethatatzerosignalthegridvoltageis0V,sothatgives:

    EgridbiasV4B=E gV4BEk V4B=01.4=1.40V

    That'sgood;1.40Visrightwherewewantittobe.

    P1eXTheory 6

  • Wasn'titconvenientthatweknewwhattheplatevoltagewas?Ifyouweredesigningyourownpreampstagefromscratch,youwouldhavetocalculatethezerosignalplatevoltageandcorrespondingbiasvoltage.Thereareafewwaystodothis.SomeusecomputerstomodelthecircuitinPSPICE(modelsforpopulartubesareavailableonDuncanMunro'swebsite).Otherspreferthechallengeofdoingthemathbyhand(ouch!).I'mavisualsortofengineerandlikedrawing,soI'llshowyouhowtodrawaloadline.

    Aloadlineissimplyagraphicalrepresentationofthecircuitthatshowstherelationshipbetweentheplateloadresistor(R22)andthetriode(V4B)inthecontextoftheIpEpgraphfromthedatasheet.SupposeyouhadalreadydecidedthatyourB+wastobe275VandyourloadresistorR22wastobe100K.BecauseR22isafixedresistance,itwillplotasastraightline.Oneendofthat100KloadlinewouldbeatthecutoffpointwhereplatecurrentiszeroandplatevoltageequalsB+(275V,0.0mA).Tofindtheotherendofthe100Kloadline,useOhm'sLawtodivideB+(275V)bythevalueoftheplateloadresistor(100K).Thisgivestheplatecurrentof2.75mA,suchasitwouldbewhenthetheloadresistorisdroppingtheentireB+voltageandtheplateisat0V(0V,2.75mA).Fig.7showsthislineinredonthe12AX7IpEpgraph.Itsupperleftterminalpointisat(0V,2.75mA)onthegraph,anditslowerrightterminalpointisat(275V,0mA)onthegraph.YoucandothiswithanyvalueofloadresistorandB+voltage.

    Nowlet'sassumewewantagridbiasvoltageof1.4V.Howdowefigurethatout?ThereisnoplatecurveontheIpEpgraphforagridvoltageof1.4V.It'seasy:Weextrapolatethevalue.Imaginethattherewereplatecurvesfor1.1V,1.2V,1.3V,and1.4Vonthegraph.Becausethedeviceislinearinthisregion,thosecurveswouldbemoreorlessevenlyspacedbetweenthe1Vand1.5Vplatecurves.Thatmeansthatthe1.4VplatecurvewouldintersecttheR22loadlineupfromthe1.5Vintersectionatabout1/5thedistancebetweenthe1Vand1.5Vplatecurves.I'vedrawnadotthereandlabeleditBiasPoint.Thebiaspointisatapproximately(180V,0.93mA)onthegraph.Thismeansthatthezerosignalplatevoltagewillbeabout180V,withaidlecurrentofabout0.93mA.Weknowwewantthecathodetobeabout1.4Vabovegroundandweknowtheidlecurrent;therefore,wecancalculatethedesiredcathoderesistanceusingOhm'slaw:

    R=EI=

    1.4V0.00093

    =1505

    P1eXTheory 7

    Fig.7:A100Kloadlinewithanoperatingpointat1.4Vgridbias

  • Afteranampisbuilt,ofcourse,itiseasytomeasuretheactualplateandcathodevoltagesandcalculatethecurrentfromthosevalues.Butifyouaredesigningastagefromscratch,thenyoucangettherefromaloadline.Supposeyouwantedtousea75KohmloadresistorwithaB+of250Vandthegridbiassetat1.2V.Tryworkingthatoutonthe12AX7IpEpgraph.Istheloadlinegoingtobesteeperorflatterthanthe100Kloadline?Ifyousaidsteeper,youareontherighttrack.Seeifyoucandeterminewhatcathoderesistancewouldbenecessarytoachievethatbiasvoltage.Whatwouldthequiescentplatevoltagebeatthatbiaspoint?Don'tlookbeforeyoutrytoworkthisout,butthereisagraphofthatloadlineandoperatingpointinAppendix2atthebackofthisdocument.

    Thechoiceofbiaspointaffectsthetonalanddistortioncharacteristicsofagivenpreampstage.Youmaywanttotuneyourampfordifferenttonalqualitiesbymodifyingthebiaspointofeitherthefirstorsecondpreampstages.Ingeneral,makingthebiaslessnegativewithrespecttothecathode(increasingthecurrentthroughthetriode)makesforawarmertonebutaffordslesscleanheadroom.Alternatively,makingthebiasmorenegativewithrespecttothecathodemakesforacoolertone.Tastesvary,soasyouadvancefrombuildinganexistingdesignliketheP1eXtocreatingyourowndesigns,youmaywishtoexperimentwithdifferentoperatingpoints.Youcaneventweakanexistingdesignwithloadchangesandoperatingpointchanges.Note,however,thatyoucanonlypushtheoperatingpointsofarineitherdirectionbeforeyouhavetostartrethinkingtheoveralldesignofthestageandperhapstheentireamp.

    AMPLIFYINGTHEINPUTSIGNAL

    Wesetthebiaspointtokeepthetubeoperatinginthelinearpartofthecurve.InplainEnglish,thismeansthetubecircuitisdesignedsothataninputsignalthatdoesn'ttrytopushthegridintoconductionorpastthecutoffvoltagewillgetamplifiedwithminimaldistortion.Thereisalwaysaslightamountofdistortionbecauseavacuumtubeisnotaperfectlylineardevice.Wewon'tworryaboutthishere;itwouldbetoocomplicatedtoaddthisnonlinearityintothecalculationsbelow.

    ThetwopreampstagesintheP1eXcircuitamplifytheinputsignalinseriesbeforeitisdeliveredtotheoutputstageorpoweramp(V1).Sothequestionis,howmuchamplificationoccursinthepreamp?Nowthatweknowtheoperatingpointofthefirsttriodestage,wecandetermineitsvoltagegain.Todothis,weneedtounderstandthreeoperationalparametersthatapplytotriodes:theamplificationfactor,transconductance,andplateresistance.

    Theamplificationfactorisameasureofhowwellthetriodeamplifiesthegridvoltage;itistheratioofthechangeinplatevoltagetothechangeingridvoltage.Ontubedatasheetsitisusuallydenotedbythesymbol

    =changeinplatevoltagechangeingridvoltage

    (platecurrentheldconstant),ormoreformally=E p Eg

    , I p const

    Ofthethreetriodeparameters,itisthemostconstantacrossvariousoperatingpoints.Somuchsothatwegenerallytreatitasaconstantandtakethevaluegiveninthedatasheet.Becauseitisaratiooftwovoltages,ithasnounitofmeasurement.

    Transconductance(ormutualconductance)istheratioofchangeinplatecurrenttothechangeingridvoltage.Inthelinearoperatingregion,itisroughlyconstant,butfluctuatesconsiderablyastheoperating

    P1eXTheory 8

  • pointmovesoutofthisregion.Thisratioislooselyameasureofhowefficientlythegridcontrolstheplatecurrent.Itisusuallydenotedintubedatasheetsbythegm:

    gm=changeinplatecurrentchangeingridvoltage

    plate voltageheld constant ormoreformally gm= I pEg

    ,E p const

    TheunitoftransconductanceiscalledaSiemens,orintheoldertextsthemho.WhileSiemensisnowtheinternationalstandardnamefortheunitofconductance,itwasnotwidelyusedattheheightoftubedesign.BecausemanyofthetubedatasheetsavailableontheInternetusethetermmho,sowillwe.

    Thethirdprimaryoperatingparameterofinterestisplateresistance(rp,sometextsrefertothisasra.)ortheratioofthechangeplatevoltagetothechangeplatecurrent.Ithelpstothinkofatriodeasavariableresistor.Whenthegridvoltageismademorepositive,morecurrentflowsthroughthetube.Thismeanstheeffectiveresistancebetweentheplateandcathodedecreases.Makethegridvoltagemorenegative,andthisresistanceincreases.Plateresistance,likeanyotherresistance,ismeasuredinohms:

    r p=changeinplatevoltagechangeinplatecurrent

    grid voltage held constant ormoreformally r p=E p I p

    ,E gconst

    Lookingatthesimilarityoftheseparameterswithrespecttoeachother,youmightwonderiftheyaresomehowrelated.Theyare,infact,mathematicallyrelated,andwewilltakeadvantageofthatlater.First,however,let'sexaminethetriodeparametersinabitmoredetail.

    The,gm,andrpvaluesforaparticulartubeareusuallyspecifiedatacoupleofdifferentplatevoltagesinthetube'sdatasheet.Forinstance,a12AX7datasheetIhaveindicates:

    PlateVoltage 100 250 VOLTSAmplificationFactor()alsocalledmu 100 100Transconductance(gm) 1250 1600 MHOsPlateResistance(rp) 80K 62.5K OHMS

    Whiletheamplificationfactorofmosttriodescanusuallybeconsideredaconstantregardlessoftheoperatingpoint,thesameisnottruefortransconductanceandplateresistance.Becausetheplatevoltageatouroperatingpoint(181V)isneither100Vor250V,wehavetodeterminethetransconductanceandplateresistanceforourquiescentoperatingpoint.Thereareavarietyofwaystoestimatethesevaluesforagivenoperatingpoint,buttoreallyseewhattheymeanforthetube,it'sagoodexercisetoplotthemontheplatecurves.

    Considertheenlargedviewofthe12AX7IpEpplatecurvesaroundouroperatingpointthatisshowninFig.8.Thefirstthingwedoisdrawaverticallinethroughtheoperatingpoint,makingsureitintersectsboththe1Vand2Vgridvoltagecurves.Thisrepresentstheconstantplatevoltageneededforcalculatingthetransconductancevalue.Next,weplothorizontallinesfromtheintersectionsatthe1Vand2Vgridcurvestoobtaintheplatecurrentsatthosegridvoltages.ThesearemarkedIaandIdonthedrawing.

    P1eXTheory 9

  • AtEg1(1V),theplatecurrentIais1.73mAandatEg2(2V),theplatecurrentIdis0.30mA.Pluggingthosevaluesintothetransconductanceformulagives:

    gm= I pE g

    =I a I d

    Eg1Eg2=1.73mA.31mA1V2V

    =1420mho

    Calculatingplateresistancetakesasimilarapproach.First,plotagridcurvethroughtheoperatingpointthatisroughlyparalleltothenearestspecifiedgridcurve.Thisrepresentsthe1.4Vgridvoltagethatmuststayconstantfortheplateresistancecalculation.Nextchoosetwovoltagesaroundtheoperatingpoint,say175Vand185V.ThosearemarkedasVaandVbonthegraph.Nextdrawverticallinesfromthosevoltagestointersectour1.4VgridcurveandobtaintheassociatedplatecurrentsatIbandIc.

    AtVa(175V)onthe1.4Vgridcurve,theplatecurrentIcis0.88mAandatVb(185V)onthe1.4Vgridcurve,theplatecurrentIbis1.03mA.Pluggingthosevaluesintotheplateresistanceformulagives:

    r p=E p I p

    =V bV aI bI c

    =187V173V

    1.05mA0.86mA=73.6K

    Asyoucansee,itisrelativelyeasytoplotthetransconductanceofatubeontheplatecurves.Allittakesisdrawingaverticallinethroughyourquiescentoperatingpoint.Accuratelyplottingaplatecurvethroughanarbitraryoperatingpointismuchmoredifficultandgenerallynotworththetroublebecause,asmentionedearlier,thethreeparametersareinterrelated.Specifically,theamplificationfactorofatriodeistheproductofitstransconductanceanditsplateresistance.Mathematically,thisisexpressedas:

    =gmr p whichisalsoequivalenttogm=r por r p=

    gm

    Recallthattheamplificationfactor()ofatriodeisrelativelyconstantoveralargerangeofoperatingconditions.Therefore,ifyouknoweitheroneoftheotheroperatingparameters,itissimpletocalculatethemissingparameter.Sincetransconductanceiseasiertoplotaccurately,wetypicallyplotitforthedesiredoperatingpointandthenplugitandthetriode'samplificationfactorintotheaboveformulaandsolvefortheplateresistance.

    r p=gm=

    100.001420

    =70422 (Wewillroundthisto70Kohms.)

    P1eXTheory 10

    Fig.8:CalculatingtransconductanceandplateresistanceontheIpEpgraph

  • Oncewehavetheplateresistanceandthetransconductance,wecalculatethevoltagegainforthissortofcircuitusingthisformula:

    Gain=r pRlRg1gmRk

    where:

    gm isthetransconductanceofthetrioderp istheplateresistanceofthetriodeRl istheplateloadresistance(note:somedocumentsrefertothisasRaorRp)Rg isfollowinggridresistanceRk istheunbypassedresistanceinthecathodecircuit

    Thesymbol||isshorthandtoindicateresistancesinparallel.Theplateloadresistanceissimplythevalueoftheplateresistor(R22ofthepreampstageone).Thefollowinggridresistanceisthetotalresistanceofthecircuitconnectedtotheplateofthetube,notincludingtheplateloadresistor.Thecathoderesistanceistheunbypassedpartofthecathodecircuit.

    SincewearetalkingaboutACsignals,let'sjustimaginethatC15hasnoresistanceatalltotheACoutputsignal(inactualpractice,itsresistancevarieswithfrequency).TheonlyreasonC15isinthecircuitistoblockthe181VDCpresentattheplateofV4B,andpassonlytheACsignal.Thatsaid,thefollowinggridresistanceseenbyV4Bisequalto1.2Mohms(R21+VR6).

    Thecathodebypasscapacitor,C16,allowsapathforACsignalstobypassR23.ConsiderwhathappensifC16isnotpresent.WhenanACvoltageisappliedtothegrid,itcausesthecurrentflowingthroughthetubetochange.WithoutC16,thevaryingcurrentthroughthetubewouldcausethevoltagedropacrossR23tovary,whichwouldcausethegridbiasvoltagewouldvarywiththesignal.Suchdegenerativefeedbackfromcathodetogridreducesthegainofthestage.ThepresenceofC16preventsthatfeedbackforsignalsaboveacertainfrequency.Thoughwewon'tdiscusshowtocalculateitjustyet,the1FvalueofC16causesthefirststagetohaveaneffectivecathoderesistanceof0ohmstoallACvoltagesabove200Hz.Pluggingthesevaluesintothegainformulagives:

    Gain=70K100K1.2M

    11400MHOS

    0=55.7

    SoanACsignalof100mVwithafrequencyabove200HzonthegridofV4Bproducesa5.57VACsignalontheplate.AtthejunctionofR21andthetopofVR6themaximumsignalstrengthisabout4.3VAC.Signalsbelow200Hzaresomewhatattenuatedbecauseofthechangesingridbiaswejustmentioned.

    ThevalueofC16hasamarkedeffectonthestage'slowfrequencyresponse.Forexample,doublingC16to2Fcausesthestagetobefullybypassedatallfrequenciesabove100Hz.DoublingC16againto4F

    P1eXTheory 11

  • lowersthebypassfrequencydownto50Hz.GoingtheotherwayandhalvingthevalueofC16to0.5Fbypassesthecathodeonlyforfrequenciesabove400Hz.ThisisanotherplaceintheP1eXwhereyoumaywanttoexperimentwithdifferentcomponentvalues.Generallywithaguitaramplifier,youwantsomebasscutinordertokeepthelowfrequenciesfrombecomingmuddywhentheampisdrivenhard.

    COUPLINGCAPACITORS

    Inthelastsection,weassumedthatC15presentsinfiniteresistancetoDC,andzeroresistancetoAC.HighDCresistanceisveryimportantwhenacapacitorcouplesampstages.

    ConsiderwhathappensifsomeoftheDCvoltagemakesitswaythroughC15intothenextstage.Stagetwoofthepreampisdesignedsothegridisat0.7Vpotential(withrespecttothecathode)whennoinputsignalisapplied.IfapositiveDCvoltagewereappliedtothegrid,itwouldalterthebiasandwouldrequireadjustingthecathoderesistortomaintainthedesiredbias.So,ifsomeoftheDCvoltagepresentattheplateofV4BweretogetpastC15,itwouldcomplicatethedesignofstagetwo.Whilesomedesignsoperatewithdirectcouplingbetweenthestages,wewon'texplorethosehere.

    Thus,C15needstoblockDCvoltagesbutlettheaudiosignalthrough.Usedinthisway,acapacitorissometimescalledaDCblockingcapacitorandmorefrequentlycalledacouplingcapacitorbecauseitcouplestheoutputfromoneamplifierstagetotheinputofanotherstage.

    ThecombinationofC15andVR6createsahighpassfilter.Lowerfrequencysignalsencountermoreresistancethanhigherfrequencysignals.IntheP1eX,C15isselectedsuchthatthecutofffrequencyofthishighpassfilterisbelowthefrequencyrangeofaguitar.IfyousubstituteasmallervaluecapacitorforC15,youdecreasethebassresponseoftheamp.IncreasingthevalueofC15givesincreasedbassresponsethatis,lowerfrequenciesarelessattenuatedbythecouplingcapacitor.

    VOLUMECONTROL

    R21andthepotentiometerVR6formavariablevoltagedividerinthecircuit,theoutputofwhichfeedstheinputgridofthesecondstage.AtthelowestsettingofVR6,thegridofV4Aisconnecteddirectlytogroundandallofthesignalfromthefirststageofthepreampisshuntedtoground.Atthehighestsetting,theACvoltageappliedtothegridofV4Aisabout78percentoftheACvoltagepresentatV4B'splate.

    R21limitsthestageonegainbeforefeedingittostagetwo.Butwaitaminute!Isn'tgaingood?Uptoapoint,yes.Buttoomuchgaincandrivethenextstageintoaparticularlynastysoundingformofdistortioncalledblockingdistortionaverynonmusicalsoundthatissometimescalledfartingout.

    P1eXTheory 12

    Fig.9:Couplingcapacitors

  • I'llleaveittoyoutoguesswhatthissoundslike.R21attenuatesthesignalalittletopreventthisfromhappening.Thegreatertheattenuation,themorenoticeableitappearsathigherfrequencies,C14isaddedtoallowagreaterproportionofthesignal'shighfrequencyenergyintostagetwo.Thishelpsbrightenthestageandkeepitfromsoundingmuddy.

    GAINMANGEMENT

    BesidestheapproachusedintheP1eX,therearevariousotherwaystomanagegainbetweenstages.Youcaninsertavoltagedividerafterthevolumepotentiometer,asshowninFig.10.Withsuchanarrangement,thebrightcapbypassesthetophalfofthedivider.Thisexampleshowsthe50percentdividerusedintheP1andHOprojects.

    Asecondapproachtogainmanagementistouseasplitloadplateresistor.Thistechnique,showninFig.11,placesthedividerintheplateload.Themainadvantageofasplitloadisthatitattenuatesthesignalwithoutchangingitstoneasmuchasotherdividermethods,thusreducingtheneedforbrightcaps.RefertoSteveAhola'sarticle,SplitLoadPlateResistorsformoreinformation.

    Athirdapproachistoeliminatetheinterstageattenuationandemployalowergaintube,suchasa12AU7.Thedisadvantageofapproachisthatitreducesthegainofbothstages,whichmaynotbedesirable.Therearealsodissimilartriodesavailable,suchasthe12DW7,thatcontainalowgainandahighgaintriodeinthesameenvelope.Thesepresentadditionalopportunitiesforexperimentation.

    P1eXTheory 13

    Fig.11:Splitloadplateresistors

    Fig.10:Voltagedivider

  • PREAMPSTAGETWO

    Usingthesametechniquesdescribedearlier,wecaneasilydeterminethebiascurrentandvoltageforV4A.Sincetheschematicshowsthevoltages,Iwillnotbotherwiththemath.Feelfreetotryyourhandatthecalculationsandseeifyoucomeupwiththesamevoltagesasshownintheschematic.Fig.12showsaloadlineforstagetwowithaloadresistanceof120Kohmsandaquiescentoperatingpointat136Vontheplateand0.7Vonthegrid.

    Therearetwosignificantdifferencesbetweenthisloadlineandthatofthefirststage.Thefirstisthatitislesssteepthanthe100Kohmloadinthefirststage.Thisresultsinslightlygreatervoltagegainforthestage,whichovercomessomeofthelossesintroducedbythetonestackthatfollowsthisstage.

    Theseconddifferenceisthatthequiescentoperatingpointisclosertotheonsetofgridconduction.Ratherthansettingthequiescentplatevoltageabouthalfwaybetweentheplatevoltageatthecutoffpointandtheplatevoltageatgridconduction,thebiasonthesecondstagesetsthequiescentplatevoltageaboutonethirdthewaybetweentheplatevoltageatthecutoffpointandtheplatevoltageattheonsetofgridconduction.Thismeansthatlargesignalstendtocompressduetogridclampingbeforetheybegintoclipduetothesignaltryingtopushbeyondthecutoffvoltage.Thisproducesasmootherformofdistortionwhencomparedtoclippingthatmanyconsiderpleasingtotheear.Tounderstandwhythisisso,let'stakealookatwhathappensinthestageasthegridsignalswinggrowslargerthan0.7Vpeaktopeak.Inotherwords,itspeakvoltageisgreaterthanthegridbiasvoltage,asshownininFig.13:

    ThecurvesinFig.13representtheoriginalinputsignalvoltage(ein),thelevelinvoltsthatthecathodeisliftedabovethecircuitground(ek)andtheresultantclippedsignalvoltagethatappearsatthegridofV4A(eg).1Asyoucansee,thenegativegoingswingofegtrackscloselywiththeinputsignal.Onthepositivegoingportionofeg,however,thingsgetinteresting.Asthegridvoltageapproachesandpassesthecathodevoltage(thatis,asthegridbecomespositivewithrespecttothecathode),thegridmustdrawdirect

    1 WaveformsinFig.13andFig.14courtesyofMerlinBlencowe.

    P1eXTheory 14

    Fig.12:120Kloadlinewiththeoperatingpointat136V/1.158mA

    Fig.13:Gridclippingduetoanoverdrivensignal.

  • currenttotrackwiththepositivegoingswingoftheinputsignal.Whilethereareplentyofelectronsflowingoutofthecathodethatcouldgointothegrid,recallthattheDCcircuitthroughwhichthoseelectronsmustflowincludesa1Mohmvolumepotentiometerattheinputofthestagestage.Thisresistancelimitsthetotalcurrentthatcanflowthroughthegridtoafractionofamilliampere.Becauseofthisgridcurrentlimitation,thepositivegoingswingoftheegisclipped.Suchgridclippingoftheinputsignalisinvertedandamplifiedontheplateload,asshowninFig.14b:

    Whilethenegativeswingoftheinputsignaltoanresistivelyloadedstageisnotlimitedbythegrid,theinvertedpositiveswingoftheoutputsignalislimitedbythemaximumpowersupplyvoltageavailabletothetube.Ifthegridvoltageswingsfarenoughnegativewithrespecttothecathode,thenplatecurrentcutsoffandtheoutputsignalclipsasshowninFig.14b.

    NoticeinFig.14a,andsimilarlyinFig.14b,howthedownwardswingoftheplatesignalfrom136Vatthebiaspointto75Visshorterinamplitudethantheupwardswingfrom136Vtoabout210V(or275VinFig.14b?Thisiscalledcompressionandistheresultofthecombinationofthenonlinearbehaviorofthetubeandchoiceofoperatingpoint.

    Allthreeformsofdistortiongridclipping,plateclippingandcompressionhavetheirplaceinguitaramplificationandeachcontributesdifferentharmoniccharacteristicstothesignal.Thus,theselectionofoperatingpointsandsupplyvoltagesdeterminesanamplifier'scharacteristictonalqualities.Whendesigningapreamplifieritmaybedesirable,asisdoneintheP1eXdesign,tobiassomestageswarmlybyreducingthecathoderesistance.Inothercases,adesignermightchoosetobiasastagecoldlywithmuchlargercathoderesistancestopushtheoperatingpointmuchclosertothecutoffpoint.Asyouprogressbeyondasimpletwostagepreamplifier,youwillfindplentyofroomforexperimentation.

    P1eXTheory 15

    Fig.14:Relationshipbetweengridinputandplateoutputsignals

  • Let'smoveontocalculatingthegainofthesecondstage.Aswiththefirststage,wedeterminethetransconductanceandplateresistanceattheoperatingpoint.Recallthevaluesfromthedatasheet:

    PlateVoltage 100 250 VOLTS

    Transconductance(gm) 1250 1600 MHOS

    PlateResistance 80K 62.5K OHMS

    Aswithstageone,theplatevoltageofV4Afallsbetween100Vand250V,sowemustestimateagain.Todothatweusethesamegraphingtechniqueusedforthefirststage,butwitha120Kohmloadlineandanoperatingpointat136Vontheplateand0.7Vonthegrid.Thisgivesatransconductancevalueof1600andplateresistanceof62.5K.

    StagetwooftheP1eXpreamp(showninFig.15)differsfromstageoneinthatithasatonestackinparallelwiththeplateload.Itseemslikeitwouldbedifficulttodeterminetheeffectiveresistanceofthiscircuit.Well,itis,andthatmakescalculatingthegainofthisstageabitmorecumbersome,especiallyifwewanttoconsiderhowthestagegainchangesatdifferentfrequencies.However,wecanapproximateitseffectonthegainofthestageifwemakesomeassumptions.

    Remembertheformulaforgain:

    Gain=r p [RlRg]

    1gmRk

    Hey!Thereareacoupleofparenthesestherethatweren'ttherebefore.That'sdeliberate.Wearegoingtopullthattermoutandcalculateitseparately.Hereistheterm:

    RlRg

    ThistermrepresentstheeffectiveresistanceofthecircuitattachedtotheplateofV4A.Oncewehavetheeffectiveloadresistance,weplugitintotheformulaabove.Hereishowtocalculatetheeffectiveload:

    Reffective=R17Rg

    P1eXTheory 16

    Fig.15:P1eXstagetwoofthepreampisfollowedbythetonestack.

  • PluginthevalueforR17:

    Reffective=120KRg

    Nowtakealeapoffaith.Justignorethegridresistanceofthefollowingstage,asifthetonestackdoesnotexistandtheoutputimpedanceisinfinite.Then,theequationabovereducesto:

    Reffective=120K

    Plugthenumbersintotheformulaforgain:

    Gain=65.2 K120K

    11600MHOS

    Rk

    The1Fofbypasscapacitance(C12)setsthehalfboostfrequencyatabout283Hz(seethenextsectionfordetailsabouthowwedeterminethehalfboostfrequency),soforaudiofrequenciessignificantlyabovethehalfboostfrequencywegetagainof:

    Gain=65.2K120 K

    11600MHOS

    =67

    Asthefrequencyfallssignificantlybelowthehalfboostfrequencythegaingraduallydropstoaminimumof34,whichyoucalculatebyassumingthecathodeisnotbypassed:

    Gain=65.2K120K

    11600MHOS

    604=34

    Atthehalfboostfrequency,thestagegainisabout50;andaroundthehalfboostfrequencythegainshiftsfromtheminimumtothemaximum.Afullybypassedgainof67isconsiderablygreaterthantheactualgainofstage.Toseewhy,we'llhavetoaddintheeffectofthetonestackonthestage.Beforewelookatthetonestacklet'sseehowtocalculatethehalfboostfrequency.Themathinthenextsectionmaybemoredifficultthanyouarewillingtotakeonrightnow.Ifyouwanttoskipitandmovedirectlytothetonestackdiscussion,youcandosowithoutworry.Thereisnothinginthetonestacksectionthatdependsonknowinghowtocalculatethehalfboostfrequency.

    P1eXTheory 17

    Fig.16:Stagetwotreatedasifthe

    outputloadwasaninfiniteimpedance.

  • CALCULATINGTHEHALFBOOSTFREQUENCY

    Asmentionedearlier,itisoftendesirabletoapplysomebasscuttoguitaramplifiers.Byusingarelativelysmallcathodebypasscapacitorweonlypartiallybypassthecathode,leavingthelowerfrequenciesunbypassedandthereforeattenuatedorshelvedtoalowerlevel.Thetechnicalnameforthisisalowshelffilter.Itissonamedbecausethefrequencyresponsegraph(Fig.17)looksalittlelikeashelf.

    MerlinBlencowe(MerlinontheAX84bulletinboard)andDavidIvanJameshavepublishedahandyformulafordeterminingthehalfboostfrequencyofaplateloadedtriodegainstage.Givenvaluesfortheplateresistance,theloadresistance,thecathoderesistance,cathodebypasscapacitance,andtheamplificationfactorofthetube,youcancalculatethefrequencyatwhichthegainishalfwaybetweentheunbypassedgainandthefullybypassedgain.Thisformulais:

    f halfboost=1 Rk 12 RLr p12 Rk 1

    2RkC k

    where:

    istheamplificationfactorofthetrioderp isplateresistanceofthetriodeRp istheplateloadresistanceortheeffectiveloadresistance(Rp||Rg)Rk isthecathoderesistanceCk isthecathodebypasscapacitance

    Donotlettheapparentcomplexityofthisformulathrowyou.Givenwhatwealreadyknowabouttheseamplifierstages,itisasimplesubstitutionproblemtosolveforthehalfboostfrequency.Let'sworkthroughitusingthestagetwodatafromthepreviouspages.Fromthedatasheet,weknowthe

    P1eXTheory 18

    Fig.17:Eachstageformsalowshelffilterwiththehalfboostfrequencyaround200300Hz

  • amplificationfactorofthe12AX7triodeis100.Wecalculatedthe12AX7plateresistanceatthestagetwooperatingpointtobe65.2K.Theplateloadresistoris120Kandthecathoderesistoris604ohms.Thecathodebypasscapacitanceis1F.Pluggingthesevaluesintotheformulagives:

    f halfboost=1 Rk 12 RLr p12 Rk 1

    2RkC k=1

    6041001

    2120K62.5K126041001

    2604.000001

    Simplifyingthesevalues,weget:

    f halfboost=1 6100436500030502

    0.0038=1.1542

    0.0038=282.72Hz

    Forthosereaderswhoaremathematicallyinclinedandwishtoknowhowtoderivethisformula,pleaserefertoBlencowe,M.andJames,D.I.ChoosingCathodeBypassCapacitors,AudioXpress,August2008,pp.1920.

    YoumightwanttoexperimentwithdifferentvaluesofCkinthisformulaandseehowtheyaffectagivengainstage'shalfboostfrequency.Itisalsoaninterestingexercisetostudythecathodebypassvaluesofvariousproductionamplifiersandseehowtheycompare.Modifyingsuccessivestagecathodebypasscapacitancesisoftenusedtoshapetheoverallfrequencyresponseofanamplifier.

    P1eXTheory 19

  • INCLUDINGTHETONESTACKRESISTANCEINGAINCALCULATIONS

    Whathappenstothegainofthisstageifwedonotignorethetonestackcircuit?

    ThesimplestapproximationofthetonestackistoimaginethatC8,C9,andC10behaveasshortcircuitstoACvoltages,effectivelyremovingthebasscontrol(VR3)andtophalfofthemiddlecontrol(VR4)fromthecircuit.Thecapacitorsinthetonestackdonotactuallybehavethatway.Instead,thefrequencyofthesignalplaysalargeroleinthetotalimpedanceofthestack,butthissimplificationgivesanacceptableapproximation.Supposealsothatthetonecontrolsaresetattheirmidpoints.ApplyingtheseassumptionsgivestheequivalentcircuitshownatthebottomofFig.18.

    ThetotalACresistanceofthisapproximatecircuitis:

    Rtonestack=R11VR2VR42

    Now,pluginthevalues:

    Rtonestack=100K250K 252=100K250K 12.5=83.9K

    Theeffectiveresistanceontheplateturnsouttobe:

    Reffective=120K83.9K=49.4K

    Plugthenumbersintothebypassedandunbypassedgainformulastoobtain:

    Gainbypassed=65.2K49.4K

    11600MHOS

    =28.1K

    625=44.9

    andGainunbypassed=

    65.2K49.4K1

    1600MHOS604

    =28.1K1229

    =22.9

    Asyoucansee,thisisasignificantdifferencefromthegainnumbersweobtainedwithoutconsideringthetonestack.Additionally,thechangeineffectiveloadmovesthehalfboostfrequencyupalittleto297Hz.Recall,however,thatthisisonlyanapproximation.CapacitorsdonotactasdeadshortstoACvoltages,butratherasfrequencydependentimpedances.Sotheattenuationfromthetonestackisnotquitesopronounced.

    Therearesomequestionsyoumightwanttothinkabout:Whatistheeffectongainifthecontrolsaresetattheirlowestpoints?Orthehighestpoints?WhatistheeffectofraisingorloweringthevalueoftheR11?

    P1eXTheory 20

    Fig.18:AmodelofthetonestackthatassumesC9andC10behaveasACshortsandeach

    controliscentered

  • TONESTACK(THELONGCOMPLICATEDEXPLANATION)

    Ifyouhavereadmanyschematicsoftubeguitaramps,thetonecontrolcircuitryshouldlookfamiliarbecausemanyampcompanieshaveusedthisarrangementofcomponentsfortheirtonecontrols.ThetubeampguruscallthiscircuitaFMVtonestack,whichstandsforFender/Marshall/Vox.TheFMVtonestackhasbeenduplicatedinhundredsofguitarampsinthelasthalfcentury.

    Technicallyspeaking,thetonestackisasetofpassivefilters,eachwithitsown3dbpoint.Anotherwaytothinkofthetonestackistocallitapassivenetworkwithfrequencydependentinsertionloss.Whenasignalisinsertedintothiscircuit,somefrequenciesareattenuatedmorethanothers.

    Let'sfirstdiscusstwotypesofpassiveRCfiltershighpassandlowpass.

    Firstthehighpassfilter:Imagineyouhaveinputandoutputconnectionswhosesignalssharethesameground.Wireacapacitorbetweentheinputandoutputconnections,andputaresistorbetweentheoutputconnectionandground.ThiscreatesanRChighpassfilterthathasa3dbcutofffrequencyof:

    F cutoff=1

    2RC

    The3dbpointreferstothefrequencyatwhichtheoutputsignalis3dblowerthantheinputsignal.Inahighpassfilter,lowerfrequenciesareattenuatedmore.

    Alowpassfilterissimplytheoppositeofahighpass.Again,imagineyouhaveinputandoutputconnectionswithacommonground.Wirearesistorbetweentheinputandoutputandplaceacapacitancebetweentheoutputandground.ThiscreatesanRClowpassfilterthathasa3dbcutofffrequencyof:

    F cutoff=1

    2RC

    Yes,thisisexactlythesameformulaasabove.Convenient,huh?

    Ifyoufeedtheoutputofahighpassfilterintotheinputofalowpassfilter,yougetabandpassfilter.

    P1eXTheory 21

    Fig.19:Filters

  • Now,let'sdissecttheP1eXtonestack,startingwiththetreblecontrol.

    ImaginethatR11,C9,andC10areremovedfromthecircuit.Youareleftwithonecapacitorbetweentheinputandoutput(C8),andasetofpotentiometersacrosstheoutputandground(VR2,VR3,andVR4).Theresistancesofthepotentiometersaddupandcanbethoughtofasasingleresistor.Soweareleftwiththesimple,highpassRCfiltershowninFig.20.

    NowconsiderwhatisgoingonatVR2.Signalswithafrequencygreaterthanthe3dbcutofffrequencypassrightthroughthecapacitortothetopterminalofVR2.Frequencieslowerthanthe3dbcutofffrequencyarenoticeablyattenuated.Rememberhowthevolumecontrolworks?Itisaresistivevoltagedivider.ThesamethinghappenswithVR2(thetreblecontrol).WhenthewiperisclosertothetopofVR2,thesignalthatjustmadeitswaythroughthehighpassfilterisattenuatedless.MovingthewipertowardthebottomlegofVR2causesmoreattenuationofthehighpassfilteroutput.

    TakeacloselookatVR3(thebasscontrol).Asthewipermovestowardthebottomlegofthepotentiometer,theresistanceofVR3isreduced.Thiscausesthe3dbcutofffrequencytoincreaseasVR3rotatescounterclockwise(movingthewiperclosertothebottomleg).

    ImaginethatC8,C10,andVR2areremovedfromthetonestackcircuit,asshowninFig.21.Whatremainsisactuallytwofiltersinseries.R11andC9formalowpassfilter,andC9andthecombinationofVR3andVR4formahighpassfilter.Hereagain,the3dbcutofffrequencyofthehighpassfilterdependsonthesettingofVR3.Atthehighestbasscontrolsettings,the3dbcutofffrequencyofthehighpassfiltercreatedbyC9,VR3,andVR4iswellbelowthefrequencyoftheguitar'slowEstring.Lowersettingsofthebasscontrolraisethe3dbcutofffrequency,attenuatingthebass.

    MovingontoVR4(themiddlecontrol),wehaveacircuitalmostidenticaltothebasscontrol.ImaginethatC8,C9,VR2,andVR3areremovedfromthecircuit,asshowninFig.22.R11andC10formalowpassfilter,justlikethebasscontrolcircuit.C10andVR4createahighpassfilter,butwithamuchhigher3dbcutofffrequencythanthehighpasselementinthebasscontrolcircuitry.Thedifferencehereiswhathappenstotheoutputofthiscombinationoffilters.Thinkofitthisway:TheoutputofthefiltersthatmakeupthemidcontrolispresentatthewiperofVR4.MovethewiperofVR4towardsthetopleg(midcontrolup),andlessofthisoutputisshuntedtogroundthroughVR4.Turningdownthemid

    P1eXTheory 22

    Fig.20:Thetreblecontrolinisolation

    Fig.21:Thebasscontrolinisolation

  • controlshuntsmorefilteredsignaltoground,attenuatingthemidfrequencies.

    Notonlydoesthemidcontrolattenuatemidfrequencies,italsoattenuatestheoverallleveloftheoutputsignal.ConsiderwhathappenswhenVR4isturnedallthewaydown,essentiallyshortingonesideofC10toground.Inthissetting,C10effectivelyshuntsmostofthesignalpresentatthelowerlegofR11toground.Andforpracticalpurposes,thelowerlegofVR2isalsoshuntedtogroundthroughthecombinationofC9andC10,atleastforACsignals.Thus,themidcontrolaffectstheoveralloutputofthetonestack.

    Wemadeahugeassumptionherewhenweignoredsomeofthetonestackcomponentstoanalyzethehigh,mid,andbasscontrolcircuitsseparately.Doingthismadeiteasiertoseehoweachsectionworked,butweignoredthefactthatthefiltersinteractwitheachother.Thisinteractionisfartoocomplextoexplainhere.Ifyouarecuriousaboutit,youcanfindamorethoroughanalysisoftheFenderstyletonestackinRichardKuehnel'sexcellentbookCircuitAnalysisofaLegendaryTubeAmplifier:TheFenderBassman5F6A.ExcerptsfromthisbookareavailableatthePentodePresswebsite:

    www.pentodepress.com.

    TONESTACK(THESHORTSUMMARY)

    TrebleControl(VR2):ThispotentiometeractsasabalancecontrolbetweentheoutputofahighpassfilterformedbyC8andthethreepotentiometers,andtheoutputofthecomplexfiltercreatedbyR11,C9,C10,VR3,andVR4.

    BassControl(VR3):Thispotentiometersetsthelower3dbcutofffrequencyofabandpassfilterformedbyR11,C9,VR3,andVR4.Italsoaffectsthe3dbcutofffrequencyofthetreblecontrolcircuit.

    MidControl(VR4):ThispotentiometercontrolstheattenuationofthebandpassfilterformedbyR11,C10,andVR4.Italsoactsasavariableattenuatorforthetonestackoutput.

    P1eXTheory 23

    Fig.22:Themiddlecontrolinisolation.

  • VACUUMPENTODES

    Adiscussionaboutpentodetubeshastostartwithanexaminationofthelimitationsoftriodetubes.Inthetalkabouttriodes,welearnedthatinsideatriodethereareafewmetalpieces,separatedfromeachotherbyavacuum.Now,considerthatacapacitorismadebyseparatingtwoconductivesurfacesbyaninsulatingdielectric.Avacuumisasortofdielectric.Soinsidethetriode,thereareacoupleofvirtualcapacitors,onebetweentheplateandgrid,andonebetweenthegridandcathode.Theseinternalcapacitancesarecalledinterelectrodecapacitances.

    Lookingatthe12AX7datasheet,wecanfindtheinterelectrodecapacitancespecifications:

    TriodeUnit1 TriodeUnit2

    GridtoPlate 1.7 1.7 pf

    GridtoCathode 1.6 1.6 pf

    PlatetoCathode 0.46 0.34 pf

    Theabilityofatriodetoaccuratelyamplifyhighfrequencysignalsislimitedbythiscapacitance.Wedonothavetoworrymuchaboutthis,sinceaguitaramplifierisnotdesignedtoamplifyradiofrequencies,butthedesiretobuildradioamplifiersintheVHF,UHFandhigherfrequencybandsledresearcherstolookforwaystoreduceinterelectrodecapacitances.

    Anotherlimitationoftriodesisthatplatecurrentdependsnotonlyongridvoltage,butalsoonplatevoltage.Forexample,supposeyoulowerthegridvoltageinatriodecircuit.Aswelearnedabove,loweringthegridvoltageresultsinanincreaseinplatecurrent.AndOhm'slawdictatesthatthevoltageacrosstheloadresistormustincreaseascurrentthroughitincreases.Now,thesupplyvoltageisaconstant,sotheplatevoltagemustdecreaseasthevoltageacrosstheloadresistorincreases(thetwovoltagesmustadduptothesupplyvoltage).Loweringtheplatevoltageresultsinreducedplatecurrent;gainislimitedinatriodebecauseofthis.Inapreampstage,reducedgainpresentsnoproblembecausewehavetodiscardsomeofthegainbetweenstagestopreventblockingdistortion.However,inanoutputstage,interactionbetweenplatevoltageandplatecurrentisnotgooditonlyreducesthemaximumoutputpoweroftheamp.Whatwasneededwasawaytodecoupletheoutputpowerfromtheplatevoltage.

    Theeffortstoreduceinterelectrodecapacitanceandtoincreaseoutputpowerledtothedevelopmentofthetetrode.Atetrodehasanextraelementcalledthescreengridfixedbetweenthegridandplate.Thescreengridsolvesbothproblemswiththetriode.Addinganelementbetweenthegridandplatereducesthedistancebetweenelectrodesinsidethetubewhich,inturn,reducetheinterelectrodecapacitance.Thisallowsthetetrodetosuccessfullyamplifymuchhigherfrequencies.

    Furthermore,thescreengridallowstheplatecurrenttooperateindependentlyfromtheplatevoltage.Howdoesitdothis?Rememberourdiscussionofdiodesandtriodes.Theelectronsstreamingoffthecathodeareattractedtotheplate,becausetheplatehasapositivevoltagepotentialcomparedtothe

    P1eXTheory 24

  • cathode.Youcouldsaythattheplateexertsapullingforceontheelectrons.Inatetrode,thescreengridisoperatedatavoltagethatisslightlylessthantheplatevoltage.Now,becausethescreengridisphysicallyclosertothecathodethantheplate,thescreengridexertsmorepullontheelectronsthantheplate.Someoftheelectronsactuallyhitthescreengrid,butmostpassrightthroughonthewaytotheplate.Thus,platevoltagehasverylittleeffectonelectronflowinatetrode.Thesteadyvoltageofthescreengridprovidesanearlyconstantpullingforceontheelectrons.Asaresult,thegridhasalmostcompletecontroloverplatecurrent,regardlessofchangesinplatevoltage.

    Whilewearetalkingaboutelectronswhizzingaroundinsideatube,nowisaperfecttimetodiscussthemainlimitationofatetrode.Imaginewhathappenstoanelectroninsideatetrodetube.Itgetsknockedoutofthecathodebecauseoftheheatingeffectofthefilament.Then,thepullofthescreengridmakesitspeedawayfromthecathode.Itacceleratespastthegridandthescreengrid.Theshorttripisoverwhenitsmashesintotheplate.Thisimpactactuallycausessomeelectronstobeknockedoutoftheatomicstructureoftheplate.Foreveryelectronthatsmashesintotheplate,twoorthreeelectronsareknockedoutoftheplate.

    We'retalkingaboutparticlephysicshere!Impressyourfriends:Tellthemthatsmallparticleacceleratorspoweryouramp.

    Thisphenomenoniscalledsecondaryemission.Triodesdonotsuffermuchfromsecondaryemission.Electronvelocitiesintriodesarelowerandthereistypicallylittleelsethatispositivelychargedtoattractthosestrayelectrons,sotheyquicklyfallbacktotheplateelectrode.Inatetrode,however,thescreenispositivelycharged,soittendstoattractthoseelectronsaswell.Thisisparticularlytruewhenalargegridsignalcausestheplatevoltagetotemporarilydropbelowthescreenvoltage.Undersuchconditions,theslowmovingelectronsfromsecondaryemissionareattractedtothescreengridanddonotmaketheirwaybacktotheplate.Thisreducesplatecurrentandincreasesscreengridcurrent,reducinggainandclippingthepeakoftheamplifiedsignal.

    Toaddressthisproblem,researchersatPhilipsplacedanadditionalelementcalledasuppressorgridbetweenthescreenandplateandlocatedveryclosetotheplate,creatingthepentode.Thesuppressorgridisusuallyconnectedtothecathode,ofteninternally,whichputsitatavoltagepotentialmuchlowerthantheplate.Itdoesnotinterferewiththeflowofelectronsfromthecathode,becausethoseelectronsaretravelingatsuchahighvelocitythattheypassrightthrough.Theslowermovingsecondaryemissionelectrons,however,areeasilyrepelledbythesuppressorgridstraightbacktotheplate.Theeffecthandilyeliminatestheproblemsassociatedwithsecondaryemission.

    OUTPUTSTAGE

    TheoutputstageoftheP1eXemploysandEL34powerpentodeasacathodebiased,singleended,transformerloadedpoweramplifier.Wehavealreadydiscussedcathodebiasing.Singleendeddescribesanoutputconfigurationwhereatubes(orasetoftubesinparallel)isconnectedtotheoneterminalofaloadandtheotherterminaloftheloadconnectstoground(oftenviathepowersupply).

    P1eXTheory 25

    Fig.23:Thepentode

  • InthecaseoftheP1eX,theloadconsistsoftheoutputtransformer(T2)andwhateverspeakeryouhaveconnectedtotheoutput.

    R9ontheschematicisagridstopperresistorthatreducestheblockingdistortionthatoccurswhenoverdrivingtheEL34grid.

    AttachedtothecathodeofV1isacircuitalmostidenticaltothecathodecircuitofthepreampstages.R10isthecathoderesistor,chosentosettheproperbiascurrent.C7isthecathodebypasscapacitor.

    Calculatingthegainoftheoutputstageisnotreallynecessary.Thedesigngoalisnottoprovideacertainamountofgain,buttodeliverthemaximumpossiblepowerintotheloadwithoutburningupthetube.Muchlikethepreampstages,thisgoalisaccomplishedbysettingthebiascorrectly.BecausetheP1eXisasingleendedamp,theoutputtubemustbebiasedapproximatelymidwaybetweenthecutoffpointandtheonsetofgridconduction.Justaswiththetriode,apentodeiscutoffbymakingthegridvoltagesufficientlynegativewithrespecttothecathodesuchthattheplatenolongerconductscurrent.Gridcurrentflowstartstoflowasthegridvoltagebecomesequaltothecathodevoltage.

    Keepingthetubefromburningoutisthemostimportantissuetoaddresswhendesigninganoutputstage.Setthebiastoofarpositive,andthetubewillfailasitattemptstodissipatemorepowerthanitcanhandle.Rememberyourbasicelectronictheory:P=I*E(powerequalscurrenttimesvoltage).ItisverysimpletodeterminethepowerdissipatedinacathodebiasedtubestagesuchastheP1eXoutput

    P1eXTheory 26

    Fig.24:TheP1eXpoweramplifiertakesthehighvoltage,lowcurrentsignalfromthepreampandconvertsittoalowvoltage,highcurrentsignalsuitablefordrivingaspeaker.

  • stage.Becauseallthecurrentinthetubemustflowthroughthecathoderesistor,dividingthecathodevoltagebythecathoderesistorvaluedeterminesthecathodecurrent.CathodecurrentintubeV1equalsplatecurrentplusscreengridcurrent.Lookingatthevoltagesontheschematic,youcanseethatourscreencurrentisabout10mA(a10Vdropacrossthe1KresistorR8).

    Theplatecurrentisthecathodecurrentminusthescreengridcurrent,asfollows:

    I p=I k I g2

    Ohm'slawtellsusthatthecathodecurrentisequivalenttothevoltagedropacrossthecathoderesistordividedbythevalueofthecathoderesistor:

    I k=EkRk

    Substitutingthislastformulaintotheplatecurrentformulashowninthepreviousformulagives:

    I p=E kRk I g2

    Todeterminethescreencurrent,youmeasurethevoltagedropacrossthescreenresistorR8anddividethatvoltagebythevalueofR8.Usingthemeasurementsshownontheschematic,weseethereisa10VdropacrossR8.Thatmeans10mAofcurrentareflowingthroughthescreengrid.Pluggingthevoltageacrossthecathoderesistor,thevalueofthatresistorandthescreencurrentintothepreviousformulagivestheplatecurrent:

    I p=27.2380

    .010=.062=62.0mA

    Todeterminetheplatedissipationinwatts,youthensubtractthecathodevoltage(Ek)fromtheplatevoltage(Ep)andmultiplytheresultbytheplatecurrent:

    Plate Dissipation=E pE kI p=38627.2 .062=22.2W

    ThedatasheetindicatesthatthemaximumplatedissipationforanEL34is25W.Assumingthevoltagesontheschematicarethevoltagesmeasuredintheactualcircuit,theplatedissipationisabout22W.Atthesevoltages,theampisrunningatapproximately88percentofthemaximumplatedissipation.Inprinciple,youcanrunasingleendedoutputstageat100percentofitsmaximumplatedissipationbecausethisoutputstructurehasarelativelyconstantcurrentdrawfromzerosignaltomaximumsignal.Inpractice,you'llwanttobackitdownabitbecausetubesaren'tperfectlylinear.ThissettingisfinefortheEL34.

    Donotassumethatyourampwillbehaveexactlylikethis.Itisveryimportantthatyoumeasuretheactualvoltagesanddothemath.Productiontubeswanderconsiderablyfromthepublishedspecifications,soitisimportanttocheckatubeincircuitandmeasureitsdissipation.Forexample,anEL34withahighergainthanthepublishedspecificationswilldrawmoreplatecurrentforagivengridvoltage.Itcouldbedamagedifplatedissipationexceeds25W.Tocorrectforsuchasituation,you

    P1eXTheory 27

  • increasethevalueofthecathoderesistortodecreasetheplatecurrent.Differentpowerdissipationinanoutputstageproducesdifferenttonalresponses.Ultimately,youmaywishtobiasbyeartoobtainthetoneyouprefer.Thisisfineaslongasyoudon'tviolatethetubespecifications.Bearinmindthathighercurrentsgenerallyresultinshorterlifespansforoutputtubes.

    OUTPUTTRANSFORMER

    Theoutputtransformerisoneoftheleastunderstoodcomponentsinatubeamp.Theconstructionisfairlysimple:twocoilsofwirewoundaroundamagneticcore.Thecoilontheinputsideiscalledtheprimary.Thecoilontheoutputsideiscalledthesecondary.

    Thenametransformerisdescriptiveofitsfunctionitperformsatransformationofthevoltageandcurrentfromtheinputtotheoutput.Forinstance,intheP1eX,theoutputtransformerconvertsthehighvoltage,lowcurrentoutputfromtheoutputtubeintoalowvoltage,highcurrentsignalsuitabletodrivealoudspeaker.Somebasictransformerformulasfollow.Asyoucansee,theturnsratioisequivalenttothevoltageandcurrentratios.

    EACPrimaryEACSeconary

    =N TurnsPrimaryN TurnsSecondary

    andIACSecondaryIACPrimary

    =N TurnsPrimaryN TurnsSecondary

    Theformulasaboveassumeanidealtransformerwithnolosses,buttheyarecloseenoughforourpurposeshere.

    Inatubeamp,themostimportantjoboftheoutputtransformeristoreflectthelowspeakerimpedanceintothehighimpedanceloadneededbytheoutputtube.Atransformeralonehasnoprimaryorsecondaryimpedance;ithasanimpedanceratio.Ifyoulookinatubeamppartscatalog,youwillseetheprimaryandsecondaryimpedancesspecifiedasratiosaccordingtothisformula:

    Z PrimaryZ Secondary

    = NTurnsPrimaryNTurnsSecondary 2

    orN TurnsPrimaryN TurnsSecondary

    = Z PrimaryZ SecondarySupposeyoubuyatransformerwitha5Kohmsto8ohmsimpedanceratio.Theturnsratiois:

    N TurnsPrimaryN TurnsSecondary

    = 50008 =625=25Ifyouweretouseaspeakeranimpedanceof4ohms,theimpedancereflectedbackontheprimarywouldbe2.5Kohms.

    N TurnsPrimaryN TurnsSecondary

    = 25004 =625=25P1eXTheory 28

    Fig.25:Atransformer

    Fig.26:Theoutputtransformerwiring.

  • TheoutputtransformershownintheP1eXschematichasa16ohmssecondarywithadditionaltapsat8and4ohms.Assumingyouconnectyourspeakertotheappropriatetap,thereflectedprimaryimpedanceis5Kohms.Thisprimaryimpedancewasnotchosenbyaccident.RatherthevalueistakenfromtypicaldesignexamplesinEL34datasheetforoperatingpointssimilartothatusedintheP1eXoutputstage.

    POWERSUPPLY

    ThepowersupplyintheP1eXisanexampleofafullwaverectifiersupply.Thepowertransformerhasthreesecondarywindings.TheredwindingiscentertappedandprovidesahighvoltagesourcefortheB+powertothepreampandoutputstage.Thegreen6.3VACwindingprovidesthefilamentsupply.Theyellow5VACwindingisintendedtoheatavacuumtuberectifier.ThestockP1eXdesignhasnotuberectifieranddoesnotusetheyellowwinding.

    Fig.27:TheP1eXpowersupplycircuitfrommainstoplateincludingfilamentsupplies

    First,let'slookattheB+supply.Thecentertapofthepowertransformer'shighvoltagesecondarywindingisconnectedtoground.Thisgroundconnectionreferencesthesecondarysignaltozerovolts,soitisalternatingcurrentaroundthecircuit'szerovoltagepoint.InthecaseoftheP1eX,thevoltageswingsfrom0Vto+400V,backdownto0V,thento400V,andbackto0V.IntheUS,eachcycletakes1/60ofasecond.Perhapsyouarequestioningthevalueof400V,becausetheHammondtransformerspecificationssaythesecondaryvoltageis275VAC.TransformersecondariesaregenerallyspecifiedbytheirRMSACoutputvoltage.RMSisameasureofvoltagedeliveredintoaloadovertimeratherthanameasurementofaninstantaneouspeakvoltage.TogetthepeakvoltageyoumustmultiplytheRMSvoltagetimesthesquarerootof2(approximately,1.414):

    V peak=V RMS2=2752=389V

    Hmm...389V.Sowhataccountsfortheextra11V?Fortheirclassicseriesoftransformers,Hammondassumesalinevoltageof115VAC,whichwasaUSstandardbackintheday.Sincethen,USwall

    P1eXTheory 29

  • voltagehasclimbedtoabout120VAC.That5VRMSdifferenceontheprimarysideaccountsforahigherpeakvoltageofabout405Vonthesecondaryside.SubtractacoupleofvoltsforthedropacrossthediodesandafewmorevoltsforDClossesinthesecondarywinding,andyouhaveaB+valueofabout400VDC.You'llalsofindthesameproportionalvoltageincreaseintheothersecondarywindings.

    Eachsideofthehighvoltagesecondarywindingconnectstotheanodeofaseriesstringoftwo1N4007siliconrectifierdiodes.Thecapacitorsacrosseachdiodesetsnuboutsomeofthehighfrequencyswitchingnoisethatthediodesintroduceintothepowersignal.Thecathodesofthesetwodiodestringsaretiedtogethertoformafullwaverectifier.Arrangedthisway,thediodesformasortofonewayvalvesoonlythepositivevoltageswingsareallowedthrough.ThetwooutofphaseACvoltagesareturnedintoonebouncingDCvoltage.Whatdoesthatmeanexactly?Considerthefollowingdiagram:

    Fig.28:RectifyingACintopulsatingDC

    Thelinevoltageisconnectedtotheprimary.Thetransformerstepsituptoabout287VAC.Noticehoweachsideofthesecondarywindingisoutofphasewiththeothersuchthatwhenonehalfofthewindingisswingingpositive,theotherisswingingnegative.Eachdiodeallowsonlythepositiveswinginghalfthrough,sowhentheycombineafterthediodes,thesignalformsapulsatingDCvoltage.

    ThestandbyswitchimmediatelyfollowingthediodesallowsyoutodelayturningontheB+totheplatesuntilafterthefilamentshavewarmedup.Considerwhathappenswhenatubeampisturnedonwiththe

    P1eXTheory 30

  • standbyswitchclosed.Thefilamentsinsidethepreampandpowertubesarecold,sonocurrentflowsfromtheplatetocathode.Withsilicondiodesastherectifier,theB+voltagejumpstofullvaluealmostinstantlywhenpowerisapplied.Rememberwhenwedeterminedthevoltagespresentontheplatesofthepreamptubesbymultiplyingcurrentbythevalueoftheplateloadresistor.Well,whennocurrentisflowingthroughthetube,thereisnovoltagedropacrosstheplateloadresistor.So,shortlyafterthepowerswitchisthrown,thevoltageattheplatesrisetofullpowersupplyvalue.ThecouplingcapacitorsaresubjectedtothefullB+value,whichiswhyallcouplingcapsinatubeampshouldbeabletowithstandthefullB+voltage.Additionally,biasvoltagesareat0V,becausenocurrentthroughthetubesmeansnocurrentthroughthecathoderesistor.

    Oncethecathodeheatsupenough,currentwillbegintoflowthroughthetube.Whentheampisfirstturnedon,however,thereisfullB+ontheplate,andnearlyzerobiasvoltage.Thisquicklysettlesasthetubeswarmup,butforashortperiodoftimetheyarestressedbeyondtheirspecifiedlimits.Thestandbyswitchisintendedtoaddresstheseissues,butitbringsitsownsetofissuestothedesign.OneoftheproblemsisthattoggleswitchesarerarelyratedforDCoperationattypicalB+values,whichmeanstheswitchitselfwillbeoperatedbeyonditsspecifications.Anotherissueisthatrunningatubeamplifierforlongperiodswithoutanyvoltageontheplatesisalsoharmfultothetubes.TheRadiotronDesigners Handbook,4thEd.recommendsnomorethan15secondsonstandby.Forthisreasonthe330KresistorR1wasadded.Thereismuchdebateaboutwhetherthestandbyswitchaddsanyvaluetothedesign.

    Thebouncingoutputvoltageatthecathodesofthosetwodiodesaddsatremendousamountof120HzhumtotheB+.Thishummustbefilteredout.C3servesasthefirststepinthatfilteringprocess.Capacitorsusedforthispurposearecalledfiltercapacitors.

    R4,whichiswiredtoparallelC3,isthebleederresistor.Itisasafetydevicethatquicklydrainsthecurrentstoredinthefiltercapacitorswhentheampisswitchedoff.Someamplifiersalsouseableederresistortoprovideaslightvoltageregulationeffectbyensuringthereisalwayssomeloadonthepowersupply.R4doesnotprovidemuchregulatoryeffectbecauseitscurrentdrawisverysmall.

    Afterthefirstfiltercap,thereisanRClowpassfilterbuiltfromR5andC4.ThesetwocomponentssmooththeB+voltageevenmore,anddropthevoltageslightly.TheoutputofthisfiltersuppliesthemainB+totheoutputstage.DownstreamfromthisfirstlowpassfilterisanotherRCfilterbuiltfromR6andC5.R6lowersthevoltageatad,andC5smoothsthevoltage.TheoutputofhisfiltersuppliespowertothescreengridoftheEL34.Finally,anotherRCfilterisbuiltfromR7andC6.Theoutputofthisfiltersuppliespowertothepreampsection.

    Therearetwoprimaryreasonswhythepowersupplyisdesignedwithsuccessivestagesoffiltering.Thefirstisthatyouwantthepreamppowersupplytobeveryquiet.Inordertofunctionproperly,thefirststageofanelectricguitarpreampmustbesensitivetoverysmallsignals.Itdoesn'tcarewhetherthesourceofthosesignalsisaguitarpickuporthepowersupply.Anynoiseorhumonthepowersupplyendsupbeinggreatlyamplified.Thesecondreasonformultiplestagesoffilteringistodecouplethepowerstagesfromeachother.Withoutthisdecoupling,alowfrequencyfeedbackpathexistsfromthepowerstagetothepreampstages.Lowfrequencyoscillationcanstartviathispathandcausetheamplifiertomakeaputtputtputtingsoundknownasmotorboating.

    P1eXTheory 31

  • FILAMENTSUPPLY

    The6.3VACwindingofpowertransformersuppliespowertothetubefilaments.Whendesigninganamp,itisimportanttoselectapowertransformerthatcanprovidethecurrentrequiredbyallthefilaments.TheHammond270DXcansupplymorethanenoughcurrentforthislittleamp.

    Ifyouhaveexaminedschematicsofvarioustubeamps,youmayhavenoticed100ohmresistorsconnectedbetweeneachsideofthefilamentsupplyandground.Theseresistorsprovideagroundreferenceforthefilamentvoltage.Itisimportanttoprovideareferencepointforthefilamentsupply,butthereisaproblemwhenthisreferenceisthesameastheamplifier'sgroundreference.

    Tounderstandthis,wehavetoreviewsometubetheory.Rememberthataheatedcathodeemitselectrons.Infact,thecathodehasacoatingofmaterialthatisverygoodatspewingoffelectronswhenheated.Eventhoughthefilamentdoesnothavethissamecoating,itdoesemitafieldofelectronsduringnormaloperation.Becausethefilamentisconnectedto6.3VAC,thisfieldhasa60Hzsignature.Foratleastpartofeach60Hzcycle,ithasapotentialthatislessthanthevoltagepresentonthecathodeinour12AX7stages.Duringthistime,thecathodeattractssomeelectronsfromthefieldaroundtheheater,whichinturnaremixedintothesignalthegridandpassedontotheplate.Theendresultisthatpowersupplyhumisintroducedintothepreampstages.

    TheP1eXdesignusesaclevertechniquetoavoidthisproblem.Insteadofhavingthefilamentsupplyreferencedtoground,itisreferencedtoapositivevoltagetakenfromthetopofthecathoderesistorontheEL34,whichisapproximately25V.Thiselevates(orbiases)theheatervoltagesufficientlyabovethepreamptube'scathodevoltagetopreventheaternoisefromenteringthepreamp.AnalternatewaytoelevatetheheatersistousetwoseriesresistorsacrossC4toformavoltagedividerwithataptoobtain1050VDCheaterreference.

    Thatconcludesourdiscussionofthepowersupply.

    Fig.29:TheP1eXpowersupplyrevisited

    P1eXTheory 32

  • AFEWFINALWORDS

    Buildingyourownamplifierisanexitingandrewardingproject.Nothingquitecomparestothesatisfactionfeltthefirsttimeyoupowerupyourampanditworks.ThisdocumenthasofferedabasicoverviewoftheinsandoutsoftheP1eXdesigninthehopesthatitwilldemystifywhatgoesoninsideourfavoritehouseholdparticleaccelerator.

    Anoteaboutsafety:Thevoltagesinsideanelectricguitaramplifierarepotentiallydeadly.Pleasetaketimetoreaduponhowtoworksafelybeforeattemptingtobuildanyelectronicsprojects.Manygoodresourcesonthistopiccanbefoundforfreeonline.TwosourcesofnotearetheDrifterAmpsguide,foundathttp://www.drifteramps.com/safety.htmlandAikenAmp'ssafetyguidelinesathttp://www.aikenamps.com/Safety/Tips.html.Thisinformationcouldsaveyourlife.Readit!

    Thisoverviewhasprovidedjustatasteofthevacuumtubeamplifierdesignprocess.Itisnottheendofthestory.Therearemanyotherdesignaspectstobeexplored:pentodestages,directcoupledstages,pushpulloutputstages,transformercoupledstages,cathodefollowers,alternatepassivetonecontrols,activetonecontrols,feedbackamplifiers,multichannelamplifiers,switchingtechniques,effectsloops,variousonboardeffectssuchasreverb,tremoloandvibrato,andhostofothertopics.Wearen'tgoingtotrytocoverthoseatall.Wehopethatthisbriefdocumenthasgivenyouasolidbaselineknowledgeabouttubeamplifiertheory.Youwillfindthatthemoreyoulearnaboutthetopic,themorethereistolearn.Thatisagoodthing.Thereisawideuniverseofcircuitvariationstotryandthetonalpossibilitiesarenearlylimitless.

    Asyoustudyotheramplifierdesigns,remembertheconceptsyoulearnedhere:howtoconstructloadlines,howtoanalyzethegainandfrequencyresponseofeachstageinanamplifier,andhowthecouplingbetweenstagescausesthemtointeract.Thosearepowerfultoolsforunderstandinganyvacuumtubeamplifier.Youareembarkingonathoroughlyenjoyablequest.Neverstoplearning,besafe,havefun.

    P1eXTheory 33

  • Appendix1:P1eXREV06.03.16CompleteSchematic

    P1eXTheory 34

  • Appendix2:12AX7WITHA75KOHMLOADAT250V

    Aspromised,hereistheloadlinefora12AX7witha75KohmloadandaB+of250V.

    The75Kohmloadissteeperthana100Kohmload.WithaB+of250Vandabiaspointat1.2Vonthegrid,thezerosignalplatevoltageisabout170Vat1.1mA.Gainissomewhatreducedbythischangebecausethelowerplateresistancedoesnotallowforaslargeanoutputsignalswing.Toverifythis,simplyreplacetheplateloadandcathoderesistancesinthegaincalculations:

    Gainfullybypassed=70K75K1.2M

    11400MHOS

    0=49.2 andGainunbypassed=

    70K75 K1.2M1

    1400MHOS1.1K

    =19.4

    Additionally,thesteeperloadlinewithagridbiasalittlecloserto0Vlowersthehalfboostfrequencyfrom199.4Hzforthestockstageonepreamp,downtoabout192.4Hz:

    f halfboost=1 Rk 12 RLr p 12 Rk 1

    2RkC k = 1

    11001001

    270.6K70K 1211001001

    21100.000001 = 192.4Hz

    P1eXTheory 35

    Fig.30:Asteeploadlineresultsinreducedgainatgenerallyhighercurrents

  • Appendix3:AbbreviationsandSymbols

    Symbol Meaning

    E Electromotiveforcemeasuredinvolts(V)

    I Electriccurrentmeasuredinamperes(A)

    R Electricalresistancemeasuredinohms()

    mA Milliampere,aunitofelectricalcurrentequivalentto1/1000ofanampere

    Theamplificationfactorofanvacuumtubeelectrode

    gm Mutualconductancebetweenthegridandplateofatriode

    rp(ra) Dynamicplate(anode)resistance

    RL(Ra) DCplate(anode)loadresistance

    Rk DCcathodeloadresistance

    Rg DCgridresistance

    G Gainofanamplifier;thatis,theratioofoutputvoltagetoinputvoltage

    V Volt,theunitofelectromotiveforce

    Ohm,theunitofresistance

    P1eXTheory 36

  • CREDITS(DaveSorlien)

    Thisdocumentcontainsmaterialfromvariouspostsfromrec.audio.tubesandalt.guitar.ampsbythefollowingauthors:DuncanMunro,NedCarlson,Anonymous.

    Thankstothefollowingindividualsfortheirassistance:RandallAiken,Carl,Nuke,Stephen,andespeciallyChrisHurley,withoutwhomtherewouldnotbeanAX84project.

    CREDITS(StephenKeller)

    InadditiontothefineworkthatDaveSorliendidwiththefirstversionofthisdocument,IwouldliketoacknowledgeandthankMerlinBlencowe,CliffChappell,ChrisHurley,Kyle,EvanLudeman,DannyNoordzy,andMilesO'Neil(akaHarrisonFordPrefect)andJeffSpenserwhoread,encouraged,andcommentedonearlydrafts.Theirmanyinsightshavemadethisamuchbetterdocument.AlsothankstoHalReichartandJoeVallinawhosevalianteffortsascopyeditorswerehelpfulbeyondcalculation.

    Whilemanypeoplehavecontributedtothedevelopmentofthisamplifier,itisfittingtomentionandthankPhilRowley(akaZaphod),whodesignedtheoriginalAX84P1eXtreme.

    REFERENCES

    Bench,S.SimpleTubeGainCalculationshttp://www.triodeel.com/gaincalc.htm(lastupdated02/23/2008;downloaded9/13/2009).

    Blencowe,M.DesigningTubePreampsforGuitarandBass,Leeds:MerlinBlencowePublisher,2009.

    Kuehnel,R.CircuitAnalysisofaLegendaryTubeAmplifier:TheFenderBassman5F6A,2ndEd.,Seattle:PentodePress,2005.

    LangfordSmith,F.ed.,RadiotronDesigner'sHandbook,4thEd.,Sydney:TheWirelessPressforAmalgamatedWirelessValveCo.,1953.

    M.I.T.Staff,PrinciplesofElectricalEngineeringSeries,AppliedElectronics,NY:JohnWiley&Sons,Inc.,1943.

    Schwartz,M.,AmecoAmateurRadioTheoryCourse,Mineola,NY:AmecoPublishingGroup,1963.

    P1eXTheory 37

  • LEGALSTUFF

    Copyright1998DavidSorlien,newmaterialcopyright2009byStephenKeller

    THISDOCUMENTATIONISPROVIDEDFREEOFCHARGEFORNONPROFITRELATEDPURPOSESSUCHASEDUCATIONALORHOBBYUSE.REPRODUCTIONOFTHISDOCUMENTFORCOMMERCIALUSEISSTRICTLYPROHIBITEDUNLESSWRITTENPERMISSIONHASBEENPROVIDEDBYTHECOPYRIGHTHOLDERS.

    NOLIABILITYCANBEACCEPTEDFORERRORSINTHISDOCUMENTATION.FURTHERMORE,NOLIABILITYCANBEACCEPTEDFORLOSSESINCURREDDIRECTLYORINDIRECTLYARISINGOUTOFTHEUSEOFINFORMATIONINTHISDOCUMENTATION.

    ALLTRADEMARKSACKNOWLEDGED.

    P1eXTheory 38

    Fig. 1: P1-eX block diagramFig. 2: The diode, forward and reversed biasedFig. 3: Vacuum TriodeFig. 4: Current falls as grid voltage goes more negative with respect to cathodeFig. 5: Stage one of the P1-eX preamp is a 12AX7 gain stageFig. 6: 12AX7 Ip-Ep graphFig. 7: A 100K load line with an operating point at -1.4V grid biasFig. 8: Calculating transconductance andplate resistance on the Ip-Ep graphFig. 9: Coupling capacitorsFig. 10: Voltage dividerFig. 11: Split-load plate resistorsFig. 12: 120K load line with the operating point at 136V/1.158mAFig. 13: Grid clipping due to an overdriven signal.Fig. 14: Relationship between grid input and plate output signalsFig. 15: P1-eX stage two of the preamp is followed by the tone stack.Fig. 16: Stage two treated as if the output load was an infinite impedance.Fig. 17: Each stage forms a low-shelf filter with the half-boost frequency around 200-300 HzFig. 18: A model of the tone stack that assumes C9 and C10 behave as AC shorts and each control is centeredFig. 19: FiltersFig. 20: The treble control in isolationFig. 21: The bass control in isolationFig. 22: The middle control in isolation.Fig. 23: The pentodeFig. 24: The P1-eX power amplifier takes the high-voltage, low-current signal from the preamp and converts it to a low-voltage, high-current signal suitable for driving a speaker.Fig. 25: A transformerFig. 26: The output transformer wiring.Fig. 27: The P1-eX power supply circuit from mains to plate including filament suppliesFig. 28: Rectifying AC into pulsating DCFig. 29: The P1-eX power supply revisitedFig. 30: A steep load line results in reduced gain at generally higher currentsIntroduction to Tube Amplifier Theory: 10.02.15Featuring the AX84 P1-eXtreme Amplifier