ACL 1 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn INTRODUCTION We aL lracLal Audlo SysLems are proud Lo announce a slgnlflcanL advancemenL ln Lhe fleld of ampllfler slmulaLlon Lechnology: MulLl-olnL lLeraLlve MaLchlng and lmpedance CorrecLlon (MlMlC 1M ). MlMlC offers a level or reallsm unmaLched by any oLher producL aL any prlce-polnL. ln Lhls whlLe-paper we presenL Lhe lnLeresLed user wlLh a slmpllfled explanaLlon of Lhe Lechnology and lLs lnherenL beneflLs. Guitar Amplifier Evolution and the Paradigm of Electric Guitar Tone 1he orlgln of Lhe elecLrlc gulLar and Lhe elecLrlc gulLar ampllfler sLems from Lhe classlc necesslLy ls Lhe moLher of lnvenLlon" moLlvaLlon. As muslcal genres evolved, gulLar players found Lhemselves unable Lo compeLe wlLh Lhe volume of brass lnsLrumenLs and acousLlc drum seLs. 1he elecLrlc gulLar and ampllfler were devlsed as a means for gulLarlsLs Lo achleve enough volume so as Lo be sufflclenLly audlble ln a llve conLexL. 1he earllesL gulLar ampllflers were noLhlng more Lhan, well, ampllflers. 1hey were deslgned Lo slmply make Lhe slgnal louder. 1he devlces avallable aL Lhe Llme mosL sulLable Lo Lhls Lask were Lhe vacuum Lube and dlrecL radlaLlon loudspeaker. 1hese early ampllflers were crude and dlsLorLed easlly when operaLed beyond Lhelr llnear range, whlch was qulLe small. unlnLenLlonally, and serendlplLously, however, Lhe sound of a dlsLorLed gulLar amp lnLo a llmlLed frequency response speaker became deslrable and evenLually ushered an enLlre new genre of muslcal sLyles. ueslgners soon began Lo add feaLures Lo gulLar amps Lo allow Lone shaplng and oLher forms of slgnal manlpulaLlon. 1hese very flrsL ampllflers serve as a foundaLlon for Lhe fundamenLal paradlgm of elecLrlc gulLar Lone" whlch ls a nonllnearlLy (aka dlsLorLlon) preceded and followed by fllLerlng. 1he flgure below lllusLraLes Lhe fundamenLal paradlgm of elecLrlc gulLar Lone:
ACL 2 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn We can furLher elaboraLe on Lhe paradlgm by breaklng lL lnLo lLs consLlLuenL componenLs, as deslred. lor example, Lhe ouLpuL fllLerlng ls comprlsed of Lhe ouLpuL fllLerlng of Lhe gulLar amp and Lhe fllLerlng of Lhe speaker. So a more deLalled model mlghL be as follows:
As gulLar amps evolved deslgners added more galn, more Lone shaplng and more feaLures. A modern gulLar amp can have mulLlple sLages of galn and varlous fllLerlng beLween Lhe sLages. 1herefore an equlvalenL block dlagram of a modern Lube amp mlghL be:
Amplifier Simulation Theory and Practice TUBE AMPS 8efore we can geL lnLo modellng we need Lo Lalk more abouL Lube ampllflers. 1ube ampllflers (amps) are Lhe mosL popular form of elecLrlc gulLar ampllflcaLlon. 1hey are cherlshed for Lhelr muslcal dlsLorLlon characLerlsLlcs and dynamlc response. Whlle Lube amps were never orlglnally deslgned for Lhese speclflc aLLrlbuLes (Lhey were orlglnally deslgned Lo be as neuLral as posslble glven Lhe Lechnology of Lhe day) muslclans were qulck Lo recognlze Lhe deslrable sonlc quallLles of an amp belng operaLed beyond lLs lnLended operaLlonal boundarles. Cver Lhe years Lube amps have evolved wlLh Lhe express lnLenL of belng dlsLorLlon generaLors, explolLlng Lhe parLlcular nonllnear characLerlsLlcs of vacuum Lubes. 1he sound of a Lube amp ls prlmarlly aLLrlbuLable Lo Lhe equallzaLlon applled Lo Lhe lnpuL slgnal and Lhe unlque way ln whlch vacuum Lubes dlsLorL when overdrlven. uesplLe Lhelr slmple clrculLs, Lhe sound generaLed by a Lube amp ls a complex relaLlonshlp beLween Lhe varlous parLs of Lhe ampllfler.
ACL 3 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn SIMULATION BASICS 1he meLhod used by all ampllfler slmulaLors, desplLe whaL monlker Lhey elecL Lo use, be lL modeler", slmulaLor", profller", dlglLal ampllfler", eLc., ls based on Lhe fundamenLal paradlgm we seL forLh ln llgure 1 above. 1he mosL popular ampllfler slmulaLors use dlglLal slgnal processlng (uS) Lo provlde Lhe varlous slgnal processlng necessary buL lL ls enLlrely posslble Lo use analog slgnal processlng, and lndeed Lhere are several producLs LhaL do. 1here are Lwo baslc bulldlng blocks used ln ampllfler slmulaLlon: fllLerlng and waveshaplng. lllLerlng, or equallzaLlon (LC), manlpulaLes a slgnal Lo change lLs frequency response. lllLers can be lmplemenLed ln a varleLy of ways. ln dlglLal processlng fllLers are Lyplcally one of Lwo Lypes: flnlLe lmpulse response (ll8) and lnflnlLe lmpulse response (ll8). ll8 fllLers are complex fllLers sulLable for a wlde range of fllLerlng Lasks and mosL producLs use ll8 fllLers (commonly called l8s) for Lhelr speaker slmulaLlons, someLlmes ln con[uncLlon wlLh ll8 fllLers. ll8 fllLers are slmple buL efflclenL fllLers LhaL are Lyplcally used Lo lmplemenL baslc fllLers such as lowpass, hlghpass, shelvlng, eLc. fllLers as well as graphlc and parameLrlc LC. Waveshaper" ls a fancy name for a nonllnear Lransfer funcLlon. Waveshapers are used Lo lnLroduce dlsLorLlon. All ampllfler slmulaLors use a speclflc class of waveshaper known as a slgmold funcLlon. A slgmold funcLlon ls a parLlcular class of nonllnearlLy LhaL ls mosLly llnear near Lhe orlgln and geLs progresslvely more nonllnear. 1he name ls derlved from Lhe shape slnce Lhe shape of a slgmold ls a lazy 'S'". Slgmold waveshapers lnLroduce dlsLorLlon by cllpplng a slgnal and are used ln ampllfler slmulaLors as Lhey mosL closely repllcaLe Lhe Lype of dlsLorLlon LhaL occurs ln a Lube ampllfler when overdrlven. 1he earllesL ampllfler slmulaLors used llLLle more Lhan ll8 lnpuL fllLers, sLaLlc waveshapers and ll8 ouLpuL fllLers. Sound deslgners crafLed Lhe lnpuL and ouLpuL fllLers ln an aLLempL Lo repllcaLe Lhe volclngs of popular ampllflers and speakers and Lhe waveshapers aLLempLed Lo dupllcaLe Lhe dlsLorLlon. needless Lo say Lhese early slmulaLors were crude soundlng by Loday's sLandards. Much of Lhls was due Lo Lhe poor speaker slmulaLlon. 1he frequency response of a speaker ls exLremely complex. 1here are mulLlple dlps and peaks ln Lhe response LhaL are lmposslble Lo accuraLely reproduce uslng ll8 fllLers. IMPULSE RESPONSES (IRs) Cne of Lhe blggesL Lechnologlcal advances ln ampllfler slmulaLlon was Lhe use of ll8 fllLers Lo repllcaLe Lhe fllLerlng of Lhe loudspeaker. Whlle early producLs relled on whaL was baslcally a graphlc equallzer Lo repllcaLe Lhe complex response of a loudspeaker, ll8 fllLers (or l8s) allowed deslgners Lo capLure Lhe acLual response of a loudspeaker (plus mlcrophone). 1hls afforded a huge lmprovemenL ln overall reallsm.
ACL 4 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn l8s are capLured by essenLlally sampllng Lhe loudspeaker. A sample of Lhe speaker's lmpulse response (hence Lhe name) ls recorded and dlglLlzed and Lhe daLa sLored for subsequenL use. 1he sample Lhen becomes Lhe coefflclenLs of Lhe ll8 fllLer dlrecLly. All modern dlglLal slmulaLors use l8s ln some fashlon for Lhelr speaker slmulaLlon. Many producLs use l8s and Lhen augmenL Lhe response wlLh ll8 fllLers Lo compensaLe for Lhe lack of low-frequency resoluLlon. 1o adequaLely repllcaLe Lhe response of a loudspeaker requlres aL leasL 20 mllllseconds of response Llme (Lyplcally around one Lhousand samples). Powever Lhls requlres coplous slgnal processlng power and many producLs are lncapable of processlng l8s of Lhls lengLh. 1runcaLlng an l8 Lo 236 samples (abouL 3 ms) ls common buL Lhls severely reduces Lhe low-frequency resoluLlon. 1herefore Lhe l8 ls ofLen augmenLed wlLh several, or more, bands of low-frequency equallzaLlon Lo correcL for Lhe lack of resoluLlon aL Lhose frequencles. (noLe LhaL Lhe Axe-lx ll can process l8s of up Lo 43 ms ln lengLh and does noL suffer from Lhls llmlLaLlon). ll8 fllLerlng ls achleved Lhrough a process known as convoluLlon. ConvoluLlon can be performed vla frequency-domaln meLhods uslng Lhe lasL lourler 1ransform (ll1) or Llme-domaln meLhods (referred Lo as dlrecL convoluLlon). lrequency-domaln convoluLlon ls far more efflclenL buL lnLroduces slgnlflcanL laLency and ls unsulLable for ampllfler slmulaLlon. ulrecL convoluLlon has zero laLency buL requlres slgnlflcanL processlng power for even Lhe shorLesL l8s. 1herefore commerclal producLs Lry Lo keep Lhe l8 lengLh as shorL as posslble ln order Lo keep Lhe requlslLe processlng cosLs as low as posslble. l8s are now ublqulLous and many producLs even allow Lhe user Lo lnsLall afLermarkeL l8s for more Lonal cholces. 1he advanced user can even creaLe hls or her own l8s uslng wldely avallable deconvoluLlon" Lools. 1he Axe-lx ll allows Lhe user Lo capLure l8s uslng on-board slgnal generaLlon and deconvoluLlon furLher slmpllfylng Lhe process. Cne of Lhe newer slmulaLors even allows Lhe user Lo capLure l8s uslng LesL Lones ln[ecLed lnLo an acLual Lube amp and Lhen recordlng Lhe response of Lhe ampllfler. unforLunaLely Lhls approach doesn'L allow accuraLe separaLlon of Lhe ampllfler response from Lhe speaker response whlch llmlLs overall devlce flexlblllLy and ls also sub[ecL Lo slgnlflcanL very-low-frequency lnaccuracy due Lo Lhe absence of sysLem response aL Lhose frequencles. An advanced Lype of l8 processlng known as dynamlc convoluLlon holds promlse for fuLure producLs buL Lhe processlng power and sLorage memory requlred make Lhls Lechnlque prohlblLlve ln consumer producLs. lor example a Lyplcal loudspeaker would requlre many megabyLes of daLa sLorage whereas a convenLlonal l8 can be sLored ln several kllobyLes. 1hese Lechnlques are also exLremely lnpuL level senslLlve and musL be carefully callbraLed Lo work properly. lurLhermore llsLenlng LesLs have noL clearly esLabllshed any audlble lmprovemenL over classlc sLaLlc l8s.
ACL 3 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn DYNAMIC WAVESHAPERS Whlle l8s achleved a much-needed lmprovemenL ln frequency response reallsm, ampllfler slmulaLors sLlll suffered from poor dynamlc response and users complalned Lhey dldn'L feel" Lhe same as a real Lube amp. 1hls unnaLural response has been addressed ln varlous ways buL Lhe mosL common ls Lhe dynamlc waveshaper". ln lLs slmplesL form Lhe dynamlc waveshaper ls noLhlng more Lhan a sLaLlc waveshaper preceeded and followed by Llme-varylng ampllflers. 1hls Lechnlque ls wldely used and ls able Lo reallsLlcally slmulaLe Lhe sag and compresslon of an overdrlven Lube amp. MosL producLs now use a slngle dynamlc waveshaper Lo slmulaLe Lhe enLlre nonllnear response of Lhe ampllfler. 1he Axe-lx ll uses a hlghly sophlsLlcaLed approach LhaL lnvolves mulLlple Lrlode slmulaLors and a dedlcaLed power amp slmulaLor. lL ls our flrm bellef LhaL mulLlple dlsLorLlon sLages can only be accuraLely slmulaLed uslng mulLlple dlglLal sLages of dlsLorLlon, especlally separaLe preamp and power amp slmulaLlons as Lhe dlsLorLlon conLrlbuLed by each ls unlque and Lhe besL Lones are achleved Lhrough a comblnaLlon of preamp and power amp dlsLorLlon. roducLs LhaL use a slngle dlsLorLlon sLage exhlblL glare" or smear" ln Lhe mldrange as Lhe same waveshaper ls used Lo dlsLorL Lhe enLlre specLrum whereas wlLh an acLual Lube amp operaLlng ln lLs sweeL spoL" some parLs of Lhe specLrum are dlsLorLed by Lhe preamp and some parLs by Lhe power amp. reamp dlsLorLlon provldes greaLer sLrlng separaLlon buL can be harsh. ower amp dlsLorLlon Lends Lo be warmer buL can be muddy. A well-deslgned Lube amp balances ouL Lhese Lwo dlsLorLlons Lo achleve clarlLy wlLhouL harshness. PRACTICAL IMPLEMENTATION All modern dlglLal ampllfler slmulaLors use an underlylng generlc ampllfler model (or several base models) LhaL can be ad[usLed Lo repllcaLe a deslred Lube ampllfler. 1hese ad[usLmenLs lnclude varlous programmable fllLers, waveshapers and dozens, or even hundreds, of parameLers LhaL conLrol Lhe underlylng model. 1hese parameLers can, for example, conLrol Lhe shape of Lhe waveshaper(s), Lhe amounL of compresslon, Lhe shape of Lhe lnpuL fllLerlng, eLc. ln operaLlon Lhe slgnal from Lhe lnpuL source (Lyplcally an elecLrlc gulLar) ls dlglLlzed and fed Lo Lhe ampllfler slmulaLor. 1he slmulaLor applles lnpuL equallzaLlon, dlsLorLlon and Lhen ouLpuL equallzaLlon. 1he user can ad[usL varlous parameLers LhaL conLrol Lhe equallzaLlon and amounL of dlsLorLlon as well as oLher perLlnenL aspecLs of Lhe slmulaLlon. noLe LhaL Lhe Axe-lx ll uses mulLlple sLages of dlsLorLlon wlLh fllLerlng beLween each sLage Lo accuraLely repllcaLe Lhe frequency selecLlve naLure of a real amp's dlsLorLlon. 1he processed slgnal from Lhe ampllfler slmulaLlon ls Lhen lnpuL Lo a speaker slmulaLlon whlch ls slmulaLed uslng l8s (and ll8s when necessary). ln some cases Lhe user may wlsh Lo defeaL Lhe speaker slmulaLlon lf he ls uslng a LradlLlonal speaker cablneL and Lherefore does noL deslre Lhe speaker slmulaLlon.
ACL 6 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn USER CONTROL no ampllfler slmulaLor would be compleLe wlLhouL allowlng Lhe user Lo conLrol Lhe slmulaLlon. All modern slmulaLors lnclude galn, Lone and varlous oLher conLrols LhaL allow Lhe user Lo shape Lhe sound. MosL Lube ampllflers are unlque ln LhaL Lhey provlde Lone conLrol vla a passlve neLwork known as a Lone sLack. A Lone sLack ls acLually somewhaL crude buL lmparLs a parLlcular sonlc quallLy Lo a Lube amp LhaL cannoL be repllcaLed uslng convenLlonal acLlve Lone conLrols (llke you would flnd on hl-fl equlpmenL). MosL modern slmulaLors repllcaLe Lhls Lone sLack Lo varylng degrees of success whereas some make no aLLempL Lo repllcaLe Lhe behavlor and Lherefore suffer from lack of auLhenLlc conLrol. 1he Axe-lx ll has Laken Lone sLack repllcaLlon Lo Lhe nexL level and ln verslon 10.xx flrmware lncludes 32 meLlculously modeled Lone sLacks. 1hese Lone sLacks noL only repllcaLe Lhe unlque sonlc slgnaLure of Lhelr real-world counLerparLs buL even repllcaLe Lhe Laper of Lhe conLrols and Lhelr lnLeracLlon. LIMITATIONS uesplLe all Lhe lmprovemenLs ln ampllfler slmulaLlon, all modern slmulaLors suffer from varlous degrees of lnaccuracy ln comparlson Lo Lhe acLual ampllfler belng slmulaLed. 1he lnaccuracles can be caLegorlzed as follows: 1. uevlaLlons ln conLrol response, l.e. Lone conLrols don'L work Lhe same. 2. uevlaLlons ln frequency response, l.e. Lhe ouLpuL doesn'L sound qulLe Lhe same. 3. uevlaLlons ln nonllnear response, l.e. Lhe dlsLorLlon quallLy ls dlfferenL. 4. uevlaLlons ln dynamlc response, l.e. Lhe slmulaLlon doesn'L feel Lhe same. lor an ampllfler slmulaLor Lo be Lruly perfecL lL musL address all of Lhese llmlLaLlons. lorLunaLely we now have Lhe Lechnology Lo address Lhese concerns: MulLlpolnL lLeraLlve MaLchlng and lmpedance CorrecLlon (MlMlC 1M ).
ACL 7 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn Introducing MIMIC TM Technology MlMlC ls a breakLhrough ln gulLar ampllfler slmulaLlon Lechnology. MlMlC addresses Lhe four fundamenLal lnaccuracles ln slmulaLlon Lhrough a comblnaLlon of Lechnlques and renders Lhe resulLlng slmulaLlon lndlsLlngulshable from an acLual Lube ampllfler. ln Lhe followlng secLlons we wlll cover Lhe four caLegorles and MlMlC's Lechnology ln correcLlng Lhem. 1. CONTROL RESPONSE DEVIATION 1he ulLlmaLe ampllfler slmulaLlon ls one LhaL behaves [usL llke Lhe ampllfler belng slmulaLed over all posslble seLLlngs of Lhe varlous conLrols. A slmulaLlon LhaL only represenLs Lhe amp aL one parLlcular seL of conLrol poslLlons ls a poor slmulaLlon slnce Lhe ldeal seLLlngs for one user aren'L usually Lhe same for oLher users. lurLhermore mosL ampllfler conLrols are hlghly lnLeracLlve and more Lhan [usL slmple galn and Lone conLrols. lor example, Lhe drlve conLrol on many amps changes Lhe amp's lnpuL frequency response as lL ls roLaLed, Lyplcally geLLlng brlghLer as Lhe drlve ls decreased. 1hls ls deslrable as gulLars wlLh powerful plckups are usually darker and sound beLLer wlLh more lnpuL Lreble boosL. Slnce Lhe plckups are more powerful Lhe user wlll Lyplcally seL Lhe drlve lower whlch Lhen resulLs ln more Lreble boosL. Conversely gulLars wlLh weaker plckups, l.e. slngle coll gulLars, are brlghLer and sound beLLer wlLh less Lreble boosL. As Lhe drlve conLrol ls lncreased Lhe amounL of Lreble boosL decreases. uevlaLlons ln conLrol response are due Lo only one reason: lncorrecL maLhemaLlcal represenLaLlons. SlmulaLlng drlve and Lone conLrols ls very dlfflculL and laborlous. 1he equaLlons are onerous and compllcaLed, especlally for Lone sLacks. Whlle lL ls LempLlng Lo Lhrow one's hands ln Lhe alr and use generlc hl-fl sLyle Lone conLrols and a generlc galn conLrol, we belleve Lhls ls conLrary Lo Lhe ulLlmaLe goal of exacLly repllcaLlng a Lube amp. 1.1 FIXING IT unforLunaLely (for us, noL you) Lhere's only one way Lo flx lL: hard work. We exhausLlvely worked ouL Lhe equaLlons for all Lhe Lone sLacks by evaluaLlng Lhe mesh and node equaLlons. lorLunaLely we have compuLers LhaL handled Lhe polynomlal reducLlon buL lL was sLlll a Perculean efforL. We also worked ouL all Lhe equaLlons for presence and depLh conLrol calculaLlons so Lhese conLrols work llke Lhe acLual amp (excepL for Lhe Laper ln some cases). llnally we measured Lhe Laper of Lhe conLrols on Lhe acLual amps and enLered LhaL daLa. 1hls ensures LhaL Lhe conLrols on Lhe model behave ldenLlcally Lo Lhe conLrols on Lhe acLual amp. noLe however LhaL Lhe Laper of Lhe presence (and depLh) conLrol can devlaLe from Lhe acLual amp. ln our LesLs we found LhaL Lhe presence conLrol on many amps dld noLhlng for Lhe flrsL 80 of lLs roLaLlon and all Lhe acLlon occurred ln Lhe lasL 20. We feel LhaL Lhls deslgn anomaly ls undeslrable and Lherefore dld noL model LhaL aspecL.
ACL 8 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn 2. FREQUENCY RESPONSE DEVIATION ln llgure 1 we lnLroduced Lhe fundamenLal paradlgm of gulLar Lone. All gulLar slmulaLors are based on Lhls fundamenLal paradlgm and, lndeed, all real Lube amps are lmpllclLly based on Lhls same paradlgm. Powever, an ampllfler slmulaLlon wlll ofLen sound dlfferenL Lhan Lhe real amp. 1hls seems lmplauslble glven Lhe apparenL slmpllclLy of Lhe block dlagram. 1here are a varleLy of reasons why Lhls can happen. Cne reason why Lhls can happen ls slmply poor lmplemenLaLlon. 1he slmulaLlon slmply doesn'L repllcaLe Lhe real ampllfler's lnpuL and ouLpuL fllLers accuraLely. 1hls can be due Lo flawed daLa (l.e. lnaccuraLe schemaLlcs) or mlsLakes ln LranslaLlon (l.e. human error). LlLher of Lhese wlll cause frequency response devlaLlons LhaL are deLecLable Lo Lhe human ear. AnoLher cause for frequency response devlaLlon ls due Lo whaL are known as paraslLlcs". A real Lube amp doesn'L necessarlly behave exacLly llke lLs schemaLlc predlcLs. 1hls ls caused by paraslLlc elemenLs LhaL are unlnLenLlonal buL, noneLheless, lmpacL Lhe frequency response. Clder amps LhaL use polnL-Lo-polnL wlrlng are more suscepLlble Lo Lhls whereas C-board amps are more lmmune, alLhough noL LoLally. uevlaLlon can also occur due Lo ampllfler-Lo-loudspeaker lnLeracLlon dlfferlng from Lhe predlcLed lnLeracLlon. 1he ouLpuL lmpedance of a Lube ampllfler ls somewhaL hlgh and Lhls causes Lhe frequency response of Lhe ampllfler Lo follow Lhe lmpedance of Lhe speaker. MosL ampllflers lmparL some amounL of negaLlve feedback Lo reduce Lhls buL, noneLheless, some degree of lnLeracLlon ls always presenL. 1he Axe-lx predlcLs Lhls lnLeracLlon very accuraLely buL Lhere are sLlll cases where Lhere ls devlaLlon due Lo Lhls lnLeracLlon. AnoLher posslblllLy ls componenL Lolerance and drlfL. Whlle Lhe deslgn may speclfy cerLaln values for componenLs, ln pracLlce Lhe acLual componenLs wlll devlaLe from Lhese values. Cver Llme Lhe devlaLlon can lncrease due Lo componenL aglng. ln some cases Lhese devlaLlons lmprove Lhe sound of Lhe amp. ln oLher cases, however, Lhls can be deLrlmenLal. 2.1 FIXING IT MlMlC addresses frequency response devlaLlon uslng a paLenL-pendlng measuremenL Lechnlque LhaL ldenLlfles and correcLs Lhese devlaLlons. A complex serles of Lones ls ln[ecLed lnLo Lhe ampllfler belng slmulaLed and Lhe ouLpuL ls measured. 1hese measuremenLs Lhen form daLa seLs LhaL are sLored wlLh each model and correcL Lhe varlous fllLers. 1hese daLa seLs are falrly large buL, forLunaLely, Lhe Axe-lx ll was deslgned Lo be fuLure-proof and has coplous non-volaLlle memory. ?ou wlll noLlce LhaL Lhe slze of verslon 10.xx flrmware ls qulLe a blL larger Lhan prevlous verslons. 1hls ls due Lo Lhe lncluslon of Lhls correcLlon daLa. MlMlC also deLecLs when Lhere are slgnlflcanL devlaLlons beLween Lhe measured and predlcLed responses and lssues a warnlng. 1hls warnlng ls an lndlcaLlon Lo reevaluaLe Lhe model and ldenLlfy posslble daLa enLry errors. noLe LhaL MlMlC ls more Lhan [usL LC maLchlng. lL maLches all Lhe varlous fllLers ln Lhe slmulaLlon Lo Lhelr real-world counLerparLs by carefully applylng LesL slgnals of varlous ampllLudes Lo lsolaLe Lhe deslred fllLers.
ACL 9 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn 3. NONLINEAR RESPONSE DEVIATION Whlle frequency response devlaLlons are Lhe mosL noLlceable Lype of devlaLlon, dlfferences ln nonllnear response are also deLecLable by humans. 1ube ampllflers lmparL Lwo dlsLlncL Lypes of nonllnearlLy: cllpplng and crossover dlsLorLlon. Cllpplng dlsLorLlon ls Lhe prlmary source of dlsLorLlon ln an overdrlven Lube amp and ls lnLenLlonally creaLed by overdrlvlng Lhe preamp and/or power amp. Crossover dlsLorLlon occurs ln push-pull power amps and lmparLs dlsLorLlon on low-level slgnals. Whlle large amounLs of crossover dlsLorLlon can be ob[ecLlonable, small amounLs glve an amp characLer and are especlally useful ln aggresslve, modern Lones. Some modern slmulaLors do noL lnclude crossover dlsLorLlon slmulaLlon and we feel Lhls ls a glarlng omlsslon. uevlaLlon from Lhese nonllnearlLles arlses due Lo Lhe Lransfer funcLlon of Lhe slmulaLed dlsLorLlon belng dlfferenL from Lhe Lransfer funcLlon of Lhe dlsLorLlon generaLed ln Lhe Lube amp. All amp slmulaLors rely on waveshapers Lo generaLe dlsLorLlon. Cne common meLhod ls Lhe (k + x) / (k - x)" Lechnlque. 1hls meLhod lmplemenLs a waveshaper LhaL ls somewhaL programmable ln Lerms of how hard" Lhe cllpplng ls. CLher Lechnlques lnvolve polynomlals or oLher maLhemaLlcal funcLlons. 1he Axe-lx ll ls unlque ln LhaL lL lmplemenLs acLual Lrlode models for Lhe preamp slmulaLlon. 1hese models behave nearly ldenLlcally Lo an acLual Lube Lrlode (and are far more accuraLe Lhan Lhe aforemenLloned Lechnlques), compleLe wlLh frequency and ampllLude dependence. AnoLher prlmary source of nonllnearlLy devlaLlon ls blas polnL devlaLlon. 1he raLlo of odd-Lo-even harmonlcs generaLed by a nonllnearlLy ls a funcLlon of Lhe qulescenL operaLlng polnL. ln oLher words, how much Lhe slgnal ls shlfLed Lo one slde of Lhe Lransfer funcLlon. 1hls ls known as Lhe blas polnL. Whlle we can predlcL wlLh excellenL cerLalnLy Lhe blas polnL of a Lube, Lhe real world doesn'L always agree and human error can also be a facLor. 3.1 FIXING IT MlMlC addresses devlaLlons ln nonllnear response by applylng pseudorandom nolse of varylng ampllLudes Lo Lhe ampllfler belng slmulaLed and comparlng Lhe response Lo Lhe response of Lhe slmulaLed ampllfler. PlsLograms of Lhe responses are collecLed and analyzed uslng sLaLlsLlcal meLhods and any devlaLlons ln dlsLorLlon Lransfer funcLlon and blas polnL are Lhen ldenLlfled and correcLed.
ACL 10 of 10 - l8AC1AL AuulC S?S1LMS WPl1L AL8 - MuL1l-Cln1 l1L8A1lvL MA1CPlnC & lMLuAnCL CC88LC1lCn 4. DYNAMIC RESPONSE DEVIATION 1he feel" of an ampllfler ls malnly aLLrlbuLed Lo lLs dynamlc response. 1ube ampllflers compress when overdrlven due Lo sagglng of Lhe power supply. 1he amounL of sag depends on a varleLy of facLors lncludlng Lhe power Lransformer, Lype of recLlfler used, eLc. uevlaLlons ln dynamlc response can make an ampllfler slmulaLlon feel Loo sLlff or Loo spongy ln comparlson Lo Lhe real amp. 4.1 FIXING IT MlMlC addresses dynamlc response devlaLlons by applylng Lonal pulses Lo Lhe ampllfler belng slmulaLed and Lo Lhe slmulaLlon slmulLaneously. 1he perLlnenL parameLers of Lhe slmulaLlon are Lhen varled unLll Lhe dynamlc characLerlsLlcs of Lhe slmulaLlon maLch Lhose of Lhe amp belng slmulaLed. 1hese correcLed parameLers are Lhen sLored wlLh Lhe model. Summary MlMlC 1M processlng ls a ma[or advancemenL ln Lhe fleld of gulLar Lube ampllfler slmulaLlon. 1hls Lechnology ldenLlfles devlaLlons ln Lhe response of Lhe slmulaLed ampllfler Lo Lhe acLual ampllfler and generaLes correcLlve daLa brlnglng a level of accuracy LhaL has been hereLofore unachlevable. An lnherenL advanLage of Lhls Lechnlque ls Lhe exacL separablllLy of Lhe ampllfler slmulaLlon from Lhe cablneL slmulaLlon. 1hls allows Lhose LhaL deslre amp-ln-Lhe-room sound" Lo achleve LhaL wlLh no compromlse as compared Lo compeLlng Lechnologles LhaL cannoL fully separaLe Lhe ampllfler slmulaLlon from Lhe cablneL response. Slmply bypass Lhe cablneL slmulaLlon and connecL Lhe Axe-lx ll Lo a quallLy power amp and gulLar cablneL of your cholce. lor Lhose comforLable wlLh lull-8ange llaL-8esponse monlLorlng (l8l8), MlMlC achleves pollshed, sLudlo-quallLy Lones wlLh ease. A Lremendous amounL of work wenL lnLo Lhls Lechnology and we are pleased Lo offer Lhls upgrade free-of- charge Lo all owners of Lhe Axe-lx ll ulglLal CulLar rocessor.