Samtec电磁场仿真标准

TheSamtecGoldenStandard:AReferenceStructureforElectricalSimulationandMeasurement:PartIDanPiscottyandJulianFerry,Samtec,Inc.andRichardA.ElcoPhD,IndependentConsultantSamtecInc.(US&Canada)Fax:[e-mail]:9/11/2006©Samtec,Inc.2005Page:2AllRightsReservedIntroductionTheSamtecGoldenStandardisareferencestructureforvalidatinganddebuggingelectricalsimulationsoftwareandhighspeed/highfrequencytestmethods.Asimplemechanicaldesignmakesitquickandeasytoimportintoanalysissoftware,yetthestructureprovidesarichsetofeasilyobservedelectricalbehaviors.Severalelectricalcharacteristicscanbedefinedbyanalyticalsolutions,sopreciseperformancebenchmarkscanbeestablished.TheGoldenStandardwasdesignedwithaspecialfocusonvalidatingcoupledandlossymodels.Itcanbeusedtodeterminemodelaccuracyforsingle-endedanddifferentialimpedance,delay,S-parameters,andsingle-endednear-endandfar-endcrosstalk.TheGoldenStandardstructurehasbeenimplementedintoaphysicalPCBassemblyforuseinvalidatinglaboratorymeasurements.ThePCBassemblyincludesSMAconnectorsforattachmenttoindustrystandardinstrumentation.TheGoldenStandardboardisusefulinevaluatingnewtestproceduresandmeasurementpost-processingtechniques.TheSMAandtransitionareaoftheassemblyalsolendthemselveswelltoanalyzingde-embeddingmethods.SamtecInc.(US&Canada)Fax:[e-mail]:9/11/2006©Samtec,Inc.2005Page:3AllRightsReserved1.1ReferenceStructureOverviewLearningtousenewsimulationormodelingsoftwareisoftenadauntingproposition.Problemsetupisusuallyuniquefromtooltotool,andterminologyoftenvariesaswell.Theuserisoftenfacedwithmultipleoptionsforparameterssuchaslaunchtype,boundaryconditions,andconvergencecriteria.Insomecases,solutionruntimeversusaccuracytrade-offsmustbeconsidered.Aftertheuserbecomesfamiliarwiththesoftwareandgeneratesinitialoutput,thesolutionmustbeverified.Simulationsarepronetoawidevarietyofpotentialerrorsatmanystepsintheprocess.Someerrorsourcesareundertheuser'scontrol,andthosearebestunderstoodbyexperimentation.Othersourcesoferrorareinherentinnumericalprocessingandunderlyingassumptionsofthesoftwareandarethusbeyondtheuserscontrol.Butanunderstandingoftheselimitations,andthetypeandmagnitudeoferrorstowhichtheyarepronetointroduce,iscriticaltoefficientapplicationofvarioussoftwarepackages.Verifyingsimulationresultsagainstawellcharacterizedandunderstoodstandardallowstheusertobecomefamiliarwiththemagnitudeandsourcesoferrorinvarioussimulationmethods.Whenthesourcesandleveloferrorsareunderstood,theusercanexerciseengineeringjudgmenttodeterminewhenaparticularsimulationprocessis“goodenough”fortheproblemathand.Agoodreferencestructureshouldhaveparametersofinterestwhichcanbecharacterizedwithanalytical,closedformequations.Forthisreason,coaxialstructuresorsimpleconductorsovergroundplanesareoftenusedassimulationstandards.Suchstandardscanworkwellforaccuracydeterminationinmanycommonsituations.Coaxialstandardscanprovideabsolutenumbersforcertainparameters,andtheirgeometriesareeasytoconstructinmostmodelingsoftware.Theycanalsobemanufacturedwithgreatprecision.Suchphysicalstandardscaninterfacedirectlytocalibratedinstrumentationforcorrelationtolabmeasurements.However,coaxialstandardsarenotwellsuitedforstudyingorvalidatingsimulationsinvolvingcoupling.Andsincetheenergyisalmostentirelycontainedwithinthestructure,coaxialreferencesdonotlendthemselveswelltostudyingeffectsofvaryingboundaryconditionsinfieldsolvers.Tostudyandquantifytheseeffects,astructureconsistingofmultipleconductorsoveragroundplaneisoftenusedasareference.Absoluteaccuracyoftheanalyticalsolutionmightnotbeasgreatasinthecoaxialcase,butitisusuallysufficientforthetaskathand.Butsuchstructuresdonotlendthemselveswelltophysicalconstructionandareoftendifficulttoconnecttolaboratoryinstrumentsformodelortestprocedurevalidation.SamtecInc.(US&Canada)Fax:[e-mail]:9/11/2006©Samtec,Inc.2005Page:4AllRightsReservedWell-characterizedphysicalreferencesareusefulformorethanjustsimulationvalidation.Standardsarealsoimportantforquantifyingtheaccuracyoftestprocedures.Testablestandardsarealsonecessaryfor“closingtheloop”oncomplexmodelvalidation.Formostcomplexstructuressuchashighspeedconnectors,analyticalsolutionsdonotexist.Atypicalmethodforvalidatingamodelforsuchacomplexstructurewouldbetocomparesimulateddatawithtestdataobtainedfromaphysicalsample.However,testingsuchcomplexstructuresoftenpresentsmanypotentialsourcesoferror.Thus,anengineercanbecaughtinanendlesslooptryingtodeterminewhetherthesimulationorthemeasurementis“morecorrect”.Agoodexampleofapotentialsourceoferrorsintestingisthetransitionfixturewhichconvertsfromthecoaxialtestinstrumentinterfacetotheconnectorundertest.Theeffectsofsuchfixturescaninsomecasesbesubtractedfromthemeasurementusingmat