EQUIPMENTPROFILE QUADESL63 LOUDSPEAKER Manufacturer'sSpecifications SystemType:Full-rangeelectrostat- ic. NominalImpedance:8ohms. Sensitivity:86dBSPLfor2.83V rmsat 1 meter. MaximumInput:40V,peak. FrequencyResponse:35Hzto20 kHz,±6dB. A.c.Supply:240-200/120-100V,50- 60Hz,5VA. Dimensions:36%in.Hx26in.W x105/8in.D(92.5cmx66cmx 27cm). Weight:41.1lbs.(18.7kg). Price:$2,450perpair. CompanyAddress:695OakGrove Ave.,MenloPark,Cal.94025. Forliterature,circleNo.93 Quadisanamelong-knownandhighlyrespectedinhigh- fidelitysoundreproduction.Notonetorushintoproduction withanewwidgetsimplybecauseeveryoneelsemakes one,PeterWalkerhasquietlymaintainedaproductlineof superbqualityandessentiallyunchangingdesign.Ina businesswhere"breakthrough"designsaremorelikea fashionfad,havingahalf-lifemeasuredinmonths, it is refreshingtoseethattrueandunchangingqualitypersists. TheQuadElectrostaticLoudspeaker63isanewerprod- uctofPeterWalker'spersistentsearchforquality;rumorhas it thatthe63standsfortheyearwhenthedesignwas begun.TheESL63isafull-range,electrostaticloudspeaker doublet,thatis,thediaphragmradiatessoundfreelytothe rearaswellastothefront.Full-rangeelectrostaticloud- speakersneedalargesurfaceareatoradiateanyapprecia- blesoundoutputatthelowerfrequencies.Thisnormally causesdispersiondifficulties atthehigherfrequencies, wherealargediaphragmcanbemanywavelengthsin extent.Peterseemstohaveveryeffectivelysolvedthis problemwiththeinventiveuseofwave-controllingpatterns ontheconductivesurfaceswhichdeterminetheactive radiationfromthediaphragm.Thisremarkablefeatisdis- missedincompanyliteraturewithtypicallyBritishunder- statement:"Signalisfedtotheelectrodesviasequential delaylines,andthemotionofthediaphragmproducesa soundpressurepatternwhichisanexactreplicaofthat fromanidealsourceplacedsome30cmsbehindtheplane ofthediaphragm." Sinceitisanelectrostatictransducer,theESL63mustbe poweredfromthea.c.line.Here I mustinsertawordof warning:InternalprotectioncircuitrypreventstheESL63 frombeingdamagedbyexcessivepower-amplifiersignal- wellandgood,butthe63sprotectthemselvesbyshort- circuitingthespeakerline. IftheESL63isenergizedfrom thea.c.line,thisshort-circuitprotectioncomesintoaction onlyathighpower-amplevels.Butifthea.c.isswitchedoff, theESL63protectsitselfatlowsignallevels.Thismeans thatunlessyouramplifierisprotectedagainstdeadshort circuits,youcoulddamageitbyattemptingtolistento musicwhenoneorbothQuadsareunpluggedfromthea.c. lineorareturnedoff.Thisis,inmyopinion,aboobytrapfor theunwaryuser.Thespeakerisprotected,butyoumight blowyouramplifier.Ofcourse,youmightneverdosucha thing,butperhapsababysitterorchildmightturnthe systemonwhenyouarenotthere.Irecommendleavingthe ESL63sonatalltimes(thedrainissmall)and,ifyouhave 110 AUDIO/JUNE1985 Thisspeakerisadipole radiator:Almostasmuch soundcomesouttheback ascomesoutthefront. 20 In 015 Q 10 O a. 7 5 0 10 A 100 Fig.1-Impedance. 1k FREQUENCY-Hz V5 10, < abliLINDUCTANCE- 30/4z N301, SIEMENS AMPSPERVOLT IaG. LOAlg. " ^.tYir IOW IIW 9H, Fig.3-Complex admittancefor2Vrms drive. 10k 20k 100k Fig.2- Complex impedance. +6 .4 +2 0 2 4 6 20 RESISTANCE-OHMS1 IWA TT 10WAT TS 10 100 1k FREQUENCY-Hz Fig.4-Changeof admittance,at1wattand 10watts,relativeto admittanceat0.1watt. i0k 20k 100k anydoubtsaboutyouramplifier'svulnerability,placinga3- A,slow-blowfuseinthespeakerline. Standing93cmhigh,witha15-cmbase,theESL63 shouldbeplacedonareasonablyfirmsurface;atoddler couldreadilypullthemoverifthespeakerswereplacedon asoftoryieldingsurface.Therearterminalsarewell marked,andnoproblemsshouldbeencounteredincon- nectingthespeakertoagoodamplifier. Measurements Comparedtoaconventionalmoving-coilloudspeaker,the ESL63presentsanunusualloadtothepoweramplifier.The measuredmagnitude(modulus)ofimpedanceisplottedin Fig.1.Althoughratedasan8-ohmsystem,theESL63's impedanceremainswellbelowthatvalueovermostofthe audiofrequencyrange.Thismeasuredcurveagreeswell withthatsuppliedbyQuadintheusermanual;however,the curvesuppliedwiththesystemstopsat20kHz,indicating, byinference,acontinuingriseofimpedancewithfrequen- cy.ThemeasurementinFig.1showsthatthisriseisdueto aresonanceat22kHz,andthattheimpedancefallsrapidly athigherfrequencies. Thisismoreclearlyseeninthecompleximpedanceplot, Fig.2.Thehigherfrequencyresonanceisnowseentobeof suchanaturethattheimpedanceabove25kHzisapoten- tiallymoredifficultloadtodrivethanthatforanyfrequency lyingbelow20kHz.At30kHz,forexample,theESL63 presentsa6-ohmvalueatacapacitiveangleof60°. It is alsoclearfromthisplotthattheimpedancewillcontinueto fallwithincreasingfrequencyabove30kHz.Careshouldbe takenthatthepoweramplifierusedtodrivetheESL63is capableofdrivingsuchaloadatveryhighfrequencies. AlthoughproperlytracedLPrecordsmaynothaveultrason- iccomponents,andCDplayersmostcertainlywillnot,a slightlymistrackingcartridgecouldgeneratedistortionin thisrange,whichadistortingamplifiercouldcrossmodulate downtotheaudiblerange.TheESL63mightthenbe improperlyblamedforbadsoundcausedbycertaincombi- nationsofcartridgesandamplifiers. Thecomplexadmittancecurve,plottedinFig.3,also showsthishigh-frequencyeffect.Theadmittancecurveisa measureoftheamperesdrawnpervoltofamplifierdrive. Becausetheactualimpedanceiscloserto4ohmsrather than8ohms, Irecommendthatthe63sbeconsidered4- ohm(or0.25-siemens)systems.Accordingly,thedataofFig. 3aretakenatanactualdrivelevelof2Vrms(correspond - 112 AUDIO/JUNE1985 TheESL63comesthe closesttoperfectphase responseofanyspeaker I'vetestedinnearly20 yearsofmeasurement. 100 90 80 70 60 10 100 1k FREQUENCY-Hz Fig.5-One-meteron-axis soundoutputlevelwitha constantdrivelevelof 2Vrms. 0 -0 a. CD t 5dB +90 0 90 10k 20k 100k 10 100 1k FREQUENCY-Hz Fig.6-One-meteron-axis phaseresponse. 1 l ill ROTATEDTOWARD LISTENING POSTION ROTATED30 FROMLISTENING POSITION 10 100 1k 10k 2 FREQUENCY-Hz Fig.7-Three-meterroom response. 100k ID 20 0 -30 40 10k 20k 0 5 TIME-MILLISECONDS Fig.8-Energy-timecurve for3-meterroomresponse. 100k ingto 1 averagewattinto0.25siemens).Measurement extendsfrom1.26Hzto32kHz.Astheadmittancecurveof Fig.3travelsawayandupwardfromtheorigin,theloud- speakerbecomesharderforaconventionalconstant-volt- agepoweramplifiertodrive. TheadmittanceoftheESL63changeswithdrivelevel. Figure4isaplotofthischange.Relativetoadrivelevelof 100mW,the1-wattadmittancedecreasesby3dBat30Hz, thenrisesby0.5dBat80Hz.The10-wattadmittancedrops by6dB(afullhalf)at35Hz,increasesby0.5dBat80Hz, andthendropsby0.75dBat250Hz.Thismeansthatthe ESL63isaneasierbassloadtodriveathigherpower,but alsothatitissomewhatnonlinear. Anechoicfrequency-responsemeasurementswereper- formedatanactualdistanceof1.5metersandcorrectedfor anequivalentreferencedistanceof1meter.Figure5isthe measuredamplituderesponse,andFig.6isthemeasured phaseresponse,bothforanaxialmicrophoneposition. Thesemeasurementsareforaconstantvoltagelevelof2V rms,sinceIamtreatingthesystemasa4-ohmload.These measurementshavebeencorrectedforair-pathtimedelay, sothataperfectphaseresponsewouldcorrespondtothe 0°axisinFig.6.TheESL63comestheclosesttothis perfectionofanyspeakersystemwhich I havetestedin nearly20yearsofsuchmeasurement.Apositive-going voltageappliedtothe"+"terminalproducesapositive- goingpressureatthelisteningposition.Boththeamplitude andphaseresponseareexceptionallygoodthroughoutthe audiofrequencyrange.Theripplesabove4kHzarecaused byinternalgrillereflections,whicharedownabout15dB fromthedirectsound.Usableresponseextendsfrom30Hz to20kHz. The3-meterroomresponseisshownin Fig. 7.This measurementwasmadeinthesameposition,andwiththe actualpoweramplifierandspeakercableconfiguration,that Iusedforthelisteningtests.Themicrophonewasplaced whereI sat,3metersfromtheloudspeakerand1meter aboveacarpetedfloor.Myonlyconcessiontomeasure- mentwastoremovethelisteningchairand,ofcourse, myself,fromthemicrophoneposition.Thespeakerwas1 meterinfrontofadrapedwall,androtatedsothatback radiationdidnotreflecttowardthelisteningposition;as recommendedbyQuad,theESL63wasplacedonthe carpetedfloor.TwomeasurementsareshowninFig.7;both correspondtothefrequencyresponseofthefirst13mSof soundtoarriveatthelisteningposition.Theuppercurve 1 14 AUDIO/JUNE1985 Inlistening,Iheardan unusualcharacteristic,a fuzzonupper-register transients,andIhadto createanewtestto measuretheeffect. 20 -30 10 F Fig.9-Expandedenergy- timecurvefor3-meter roomresponse. 100 ONAS FREQUENCY-Hz Fig.11-FFTfortransient toneat3meters. 10k 20k 100k _J to 10dB ) 1k ',EQUENCY-Hz Fig.10-FFTforsteady- statetoneat3meters. 100 10k 206 I I I 111111111 IIItll11It C 2 4 6 8 10 TIMEAFTERRECEIPTOFTONE-MILLISECONDS Fig.12-Timesignalfor FFTofFig.11. 1006 showsresponsewhentheESL63isrotatedtowardthe listeningposition,andthelowercurveshowsresponse whenthespeakerisrotatedtodirectitssound30°off-axis fromthelisteningposition.(Duringtheearlierlisteningtests, I hadthespeakersdirectedtowardme.)Thecurvesare displacedby10dBonthisplotforclarityofpresentation. Ihavedevotedthissubstantialdiscussiontotheroomtest forthesimplereasonthatpreviously,duringthelistening tests, IhadheardanunusualdistortionwhichIwasnotable tomeasureduringsubsequentlaboratorytests.What I heardwasa"fuzz"onupper-registertransients,inthe1to 5-kHzrange.Thinkingitmightbedistortionintheprogram material,whichtheclarityoftheESL63wasrevealing, I borrowedahigh-qualityCDplayerandmadecomparative listeningtestsusingidenticalprogrampassagesfromdirect discandCD-andI stillheardthefuzz. Thesoundwassimilartocertaintypesofharmonicdistor- tion,yettheESL63provedtohaveexceedinglylowlevelsof suchdistortioninthelaboratorytests.The3-meterroomtest providedaclue. I regularlyperformmanymoremeasure- mentsthanaresubmittedforthesereviews.Onesuch measurementisthefullenergy-timecurve(ETC)ofthe3- meterlisteningcondition.Figure8istheETCcorresponding totheuppercurveinFig.7,reflectingtheconditionIused whenlistening,andherewastheclue:Thefirst3mSof soundhadamostunusualarrivalpattern.Thiscausedthe irregularresponse in thefrequencyspectrumofearly sound,whichisevidentinthe3-meterroommeasurementof Fig.7.Figure9isanexpandedETC,showingthefirst4mS ofsound. ButwhereisthedistortionIheard?Itseems,fromboththe frequencyresponseandthetimeresponse,thatthefirst3to 10mSmaybewheretheproblemlies. Iheardthedistortion onsibilantsinfemalevocalsandintoneswhichwerechar- acterizedbyrapidattackfollowedbysustain. Ifound(or believethat I found)thedistortionbymeasuringtransient tonesintheactuallisteninglocation.ButIhadtocreatenew softwareandanewtestinordertodoit. Figure10isanFFTmeasurementof16mSofsounddue toasine-wavesignalat2.65kHz.Thisisanactualmicro- phonemeasurementforthesustaintoneat93dBSPLatthe 3-meterlisteningposition,anditwascaptured250mSafter thetonewasapplied,correspondingtosteadystate. It is nearlyperfect,withlessthan0.5%second-harmonicdistor- tion.Thisisthefrequencyhavingthelargestmeasured harmonicdistortionatalevelof93dBatthelistening 116 AUDIO/JUNE1985 Horizontaldispersionis morerestrictedthan verticaldispersion.For uniformsound,aimthem atthelistener. 10 -30 -40 A IM 9 10 II TIME-MILLISECONDS Fig.13-Expanded energy-timecurvefor 3-meterroomresponse, speakeronfloor. 12 5d8 4 10 100 1k S PE FREQUENCY-Hz Fig.15-Three-meteroff- axisfrequencyresponse forspeakerraised60cm (uppertrace)andonfloor (lowertrace). A' ERRAISED60 I0k 2Cik 100k 0 -10 20 co -30 -40 8 9 10 II TIME-MILLISECONDS Fig.14-Expanded energy-timecurvefor 3-meterroomresponse, speakerraised60cm. REAR AXIS Odk FRONT AXIS Fig.16-Horizontal polar-energyresponse. 12 locationandinthe1to10-kHzrange;quitelow,evenatits worst.Bycomparison,atoneof1kHzhaslessthan0.15% distortionatthissamelevel.What I heardisclearlynot steady-stateharmonicdistortion. NowlookatFig.11.Thisisthesame16-mSwindow,with thesameHanning-weightedFFT,butnowmeasuredimme- diatelyafterthefirstsoundarrivedatthelisteninglocation. Suddenly,thereisfuzz.Theotherwisepuretoneisaccom- paniedbyanonsymmetricaldistributionofsoundextending from100Hztobeyond15kHz.Figure12,whichisthetime signalwhoseFFTisgivenin Fig. 11,revealswhythis occurs.Theoscilloscopedisplayshowsthattheearlyreflec- tionsinterferedwiththedirectsound.Theresultofthis ensembleofsoundsisacombinationofamplitudemodula- tionandphasemodulationofthe2.65kHz"carrier."Weno longerhaveonesource;wehaveanensembleofsources fortheearlysound. ButwhatisthereabouttheESL63whichmadethiseffect soprominentinmylisteningsituation?Comparisonofthe roomETC(Fig.9)withtheanechoicETC(Fig.20)showsthe existenceofearlyscatterfromthefloor.TheESL63istoo closetothefloor.Inordertoverifythishypothesis,andto exoneraterearward-radiatedsoundfromthisdipolesource, measurementswereperformedinaworst-casecondition withthespeaker'ssoundaxisoffset30°inthelistening room.Figure13istheETCforthefloor-mountedconfigura- 118 AUDIO/JUNE1985 The63sgiveagoodsense ofprogramdynamics. Hornsarewellarticulated, andcertainpercussive materialcomesalive. REAR AXIS Fig.17-Vertical polar-energyresponse. S 7 6 5 4 3 2 0 TOP FRONT AXIS 008 BOTTOM 0.1 POWER-WATTS Fig.19-IMdistortionon 440Hzproducedby 41.2Hz(El)whenmixed inone-to-oneproportion. 10 100 tion.Figure14isthesamemeasurementwiththespeakeris raised60cmabovetheflooronaflatstool.Figure15shows the3-meter,off-axisresponsewhenthespeakerisraised60 cm(uppercurve)andwhenitisplacedonthefloor(lower curve).Theearlysoundismuchsmootherintheraised position.And,yes,muchofthefuzz(butnotallofit)went awaywhenlisteningtotheraisedESL63s. Finalnote:TheESL63sareverysensitivetotheroomand whereyouplacetheminthatroom.Theyshouldneverbe placeddirectlyonahardfloor; I recommendelevating them,ifpossible.However,thesespeakersareheavyand canbetippedoverbyavigoroustoddler,sousecautionin -10 -20 -30 -40 92 0 IVV 0. MS= M1111111111111111111111111111NINIIIII 111111111111111111111111111111111 "lull" == 1111111111111111111111111111111111111111.1111111111=1W MIMEMIMINIMMINIMMIrdiaM 111011MEMINKTIE 1111111111111111PMAII 1 awl.=WO' MOMM...7="MEM/1.4INI/1111.6,1MII00 ar:00...41.1ftr -ii11114,7ar<atZliii11.1111111b .1==874=71:ot:4=2=1;ft.M=1% r-- At= .Nomm =ma. 1.11, MINN0.111 1111111111111111111 Ea -2nd 82.3rd A3.2nd E23rd Agz2rd1 En.3d A2 A2 POWER-WATTS 10 100 BO ' 90 SPLE,(41.2Hz) ;c. tdo SPLA2(110Hz) 100 SPLA4(440Hz) Fig.18-Harmonic distortionforthetonesEl (41.2Hz),A2(110Hz),and A4(440Hz). 30 3.5 4.0 TIME-MILLISECONDS Fig.20-Energy-time curvetakenat1meter withgrilleinplace. 45 placingthemathigherelevations ifchildrenorpetsare goingtobearound. Figures16and17,thehorizontalandverticalpolar - energypatterns,arerelevanttotheprecedingdiscussion onplacingtheESL63sforbestsound.Thesearemeasure- mentsoftheintegralofthesquareofthelinearamplitude frequencyresponseforallfrequenciesfrom20Hzto20 kHz.Theseshowthatthe63isadipoleradiator;almostas muchsoundenergycomesoutofthebackascomesout thefront,andwithessentiallythesamepolarpattern.How- ever,thehorizontaldispersion(theleft-rightresponse)is morerestrictedthantheverticaldispersion. Ifyouwantto 120 AUDIO/JUNE1985 Stereoimagingisexcellent andmidrangetimbral balanceisverygood,but reproductionofpianoand humanvoicefallsshort. Groupedatthespeaker'sbasearethe powerleads,fuse,on/offswitch,voltageselector, andinputterminals. obtainthemostuniformdirectsound,aimthespeakersat thesespeakersneedtoberaisedabovethefloorand shouldneverbeplaceddirectlyunderashelforanyother overhangingreflectingsurface.Thesespeakerscannotbe safelytipped;otherwise I wouldrecommendraisingthe speakersoffthefloorandtwistingthem90°(sothattopand bottombecomesides)forbestlateralizationofstereoimag- ing.Careneedstobetakenthattherearward-radiated sounddoesnotreflectoffnearbysurfacesinsuchaway thatitcomesdirectlybacktothelisteningpositionwithless thanabout20mSoftimedelay. HarmonicdistortionfortonesofE,(41.2Hz),A2(110Hz), andA4(440Hz)isshowninFig.18,with1wattcorrespond- ingto2Vrmsdrive.Whatisremarkableaboutthese measurementsisthatnotonlyisthedistortionquitelow throughoutthefullpowerrange,butdistortion isalmost independentofpowerlevelforthetwohighertones.Thisis anextremelycleanresponse. ThemeasuredIMdistortiononA4(440Hz),causedby simultaneousreproductionofEl(41.2Hz)atthesamedrive level,isplottedinFig.19.Thedistortionisquitelowrightup toanequivalentdrivelevelof100watts.Thisisanextremely cleansound.Thenatureofthedistortionisprincipallyphase modulationonA4causedbyEl. Inthecrescendotest,aninnermusicalvoiceofA4was completelyunaffected(lessthan0.02-dBchange)bysu- perimposedbroad-bandnoisewhichhada20-dBhigher averagelevel,evenuptoacombinedvoltageof80Vpeak topeak,atwhichpointthespeaker'sprotectivenetwork nippedoffthesignal. Intransferlinearity,soundpowerat100averagewatts driveshowedlessthan0.1-dBcompressionforatoneof 440Hz,comparedtoadriveof0.1watt.MiddleC,262Hz, wascompressedbyabout0.3dBathigherdrive,whileEl (41.2Hz)increasedby1dBat10watts,relativeto0.1watt. Theaxial1-meterETCisshowninFig.20.Itisapparent fromthismeasurementthattheearliest-arrivingsoundis virtuallyflawlessupto20kHz.Butlow-levelinternalreflec- tions,fromwithintheenclosure,causeinterferencethat persistsforabout0.75mS. I suspectthattheprotective grilleanditssupportingstructurearethesourceofthese earlyreflections.ThenatureofthisETCsuggestsanaccu- ratereproductionofpercussivesoundsinthehighestregis- ters,butaslightblurringoftransientswhoseenergyis concentratedinthe2to5-kHzrange. Thecombinationofthedistortionmeasurements(HD,IM, crescendo,andtransferlinearity)andtheETCindicatethat stereoimaging,bothlateralizationanddepth,shouldbe excellentandremainstableoverfullprogramdynamics. UseandListeningTests I hadagreatdealofdifficultyplacingtheESL63sfor acceptablestereosound,Theconfiguration I finallychose wastheoneusedinthe3-meterroomtest(described above),withthetwospeakerssubtendingslightlymorethan 60°atmylisteninglocationandbothrotatedtofacedirectly towardme. I placedthem1 meterinfrontofadraped surfaceinordertolessentheeffectoftherearwardsound whichcontributestothereverberantfield. Mycommentsinthispartofthereviewarebasedonmy bleevidencesavemylisteningexperience. Iwasnotfavor- ablyimpressedwiththelisteningqualitiesoftheESL63.In myopinion,theirsounddoesnotliveuptotheirhigh pedigree. Stereoimagingisexcellent,andmidrangetimbralbal- anceisverygood,but I couldnotachievetheaccurate illusionofpianoorhumanvoice,nomatterhowIpositioned thesespeakersinmyroom.Inaddition, Iwasbotheredby anupper-midrangeharshnessandfuzz,which Ihavede- scribedabove. The63sdogiveagoodsenseofprogramdynamics. Trumpetandotherhornsarewellarticulated,andcertain percussivematerialliterallycomesalivewiththesespeak- ers.Thisaccuracyofdynamicsissomewhatflawedbythe speakers'inabilitytohandleveryhighpeakSPLssincethe protectivecircuitryshutsthespeakerdownbeforedamage canoccur,butwellbelowthepeaksoundlevelthatsome listenersmayprefer. AsI saidattheoutset, Iwasabitapprehensiveabout Quad'stechniqueofspeakerprotection.Theterminalsare shuntedwheneverdamagingvoltagesareappliedtothe speakers.Puttingashort-circuitacrosstheterminalsofa high-poweramplifierwhenthesignalgetstoohighforthe speakerseemstomeabitlikedroppingasteelplateinthe pathofafastautomobilewhenitsspeedgetstoohighfor theroadconditions.Inthatcase, Iamnotsurewhois prbtectingwhom.However, I diddeliberatelytriptheESL protectionquiteanumberoftimeswhenplayinghigh- qualityCDsandLPs,withabsolutelynoproblems;the offendingchannelsimplywentmuteforafewsecondsand thencamebackon,nonetheworseforwear. Myprejudicesnotwithstanding,theESL63sareaccurate intheirimagingoforchestraandareexemplaryreproducers ofpercussiveprogrammaterial. RichardC.Heyser 122 AUDIO/JUNE1985