CP-CPK-PP-PPK.ppt

ProcessCapability(Cp/Cpk/Pp/Ppk)GlobalTrainingMaterial,Creator:GlobalMechanicsProcessManagerFunction:MechanicsApprover:GaryBradley/GlobalProcessTeamDocumentID:DMT00018-ENVersion/Status:V.1.0/ApprovedLocation:Notes:NMPDOCMANR4PCPPCProcessLibraryDocManChangeHistory:IssueDateHandledByComments1.021stDec01JimChristy&SrenLundsfrydApprovedforGlobalUse1.103rdFeb05YrjoSaastamoinenMiniCpk&ConfintervalsaddedCpktemplateupdatedNOTEAllcommentsandimprovementsshouldbeaddressedtothecreatorofthisdocument.,Contents,SectionHeading/DescriptionPage1Variation,TolerancesandDimensionalControl42Population,SampleandNormalDistribution153CpandCpkConcept28ConfidenceInterval37ProcessCapabilityAnalysis47MiniCpk537CpkDataCollectionSpreadsheetDMM0002460,ProcessCapability-EvaluatingManufacturingVariation,AcknowledgementsBennyMatthiassen(NMPCMT,Copenhagen,Denmark)FrankAdler(NMPAlliance,Dallas,USA)JoniLaakso(NMPR&D,Salo,Finland)JimChristy(NMPSRC,Southwood,UK),Section1Variation,TolerancesandDimensionalControl,TwoTypesofProductCharacteristics,Variable:Acharacteristicmeasuredinphysicalunits,limetres,volts,amps,decibelandseconds.,Inthistrainingwedealwithvariablesonly,TheSourcesofProcess/SystemVariation,Process,TwoTypesofProcesses,Allprocesseshave:Natural(random)variability=duetocommoncauses,StableProcess:Aprocessinwhichvariationinoutcomesarises“onlyfromcommoncauses,UnstableProcess:Aprocessinwhichvariationisaresultofbothcommonandspecialcauses,Unnaturalvariability=duetospecialcauses,Shewhart(1931),TheTwoCausesofVariation,CommonCauses:Causesthatareimplementedintheprocessduetothedesignoftheprocess,andaffectalloutcomesoftheprocessIdentifyingthesetypesofcausesrequiresmethodssuchasDesignofExperiment(DOE),etc.SpecialCauses:Causesthatarenotpresentintheprocessallthetimeanddonotaffectalloutcomes,butarisebecauseofspecificcircumstancesSpecialcausescanbeidentifiedusingStatisticalProcessControl(SPC),Defect,USL,LSL,nominalvalue,Tolerances,LSL(lowerspecificationlimit)10,7,USL(upperspecificationlimit)10,9,Acceptablepart,RejectedPart,RejectedProduct,Nominal10,80,1,RejectedPart,Atoleranceisaallowedmaximumvariationofadimension.,MeasurementReport,ForMeasurementreport(FAI)wenormallymeasure1-5partspercavitydependingonresultandthetypeofthedimension,ExampleofCapabilityAnalysisData,Forsomecriticaldimensionsweneedtomeasuremorethan1partForcapabilitydataweusuallymeasure5pcs2times/hour=100pcs(butsamplingplanneedstobemadeonthebasisofproductionquantity,rundurationandcycletime),ProcessCapability-Whatisit?,ProcessCapabilityisameasureoftheinherentcapabilityofamanufacturingprocesstobeabletoconsistentlyproducecomponentsthatmeettherequireddesignspecifications,ProcessCapabilityisdesignatedbyCpandCpk,ProcessPerformanceisameasureoftheperformanceofaprocesstobeabletoconsistentlyproducecomponentsthatmeettherequireddesignspecifications.ProcessPerformanceincludesspecialcausesofvariationnotpresentinProcessCapability,ProcessPerformanceisdesignatedPpandPpk,WhyMakeProcessCapabilityStudies,Aprocesscapabilitystudywouldrevealthatthetoolshouldnotbeaccepted,Whenadimensionneedstobekeptproperlywithinspec,wemuststudytheprocesscapability.butstillthisisnoguaranteefortheactualperformanceoftheprocessasitisonlyaninitialcapabilitystudy,TheNokiaProcessVerificationProcess,1-5parts/cavitymeasuredformeasurementreport(seeDMY00019-EN),Section2.Population,SampleandNormalDistribution,TheBellShaped(Normal)Distribution,Symmetricalshapewithapeakinthemiddleoftherangeofthedata.Indicatesthattheinputvariables(Xs)totheprocessarerandomlyinfluenced.,“PopulationParameters”=Populationmean=Populationstandarddeviation,PopulationversusSample,PopulationAnentiregroupofobjectsthathavebeenmadeorwillbemadecontainingacharacteristicofinterest,SampleThegroupofobjectsactuallymeasuredinastatisticalstudyAsampleisusuallyasubsetofthepopulationofinterest,TheNormalDistribution,WhatMeasurementsCanBeUsedtoDescribeaProcessorSystem?,mean(average)ordescribesthelocationofthedistribution,(m),ameasureofcentraltendency,isthemeanoraverageofallvaluesinthepopulation.Whenonlyasampleofthepopulationisbeingdescribed,meanismoreproperlydenotedas(x-bar):,Themostsimplemeasureofvariabilityistherange.Therangeofasampleisdefinedbyasthedifferencebetweenthelargestandthesmallestobservationfromsamplesinasub-group,e.g.5consecutivepartsfromthemanufacturingprocess.,WhatMeasurementsCanBeUsedtoDescribeProcessvariation?,sST-oftennotatedasorsigma,isanothermeasureofdispersionorvariabilityandstandsfor“short-termstandarddeviation”,whichmeasuresthevariabilityofaprocessorsystemusing“rational”sub-grouping.,whereistherangeofsubgroupj,Nthenumberofsubgroups,andd2dependsonthenumberofsamples(n)withinasubgroup(seenextslide).,WhatMeasurementsCanBeUsedtoDescribeProcessvariation?,d2valuesforSST,n=no.ofsamplesineachsub-group,Typical:n=5,Example:,WhatMeasurementsCanBeUsedtoDescribeProcessvariation?,TheDifferenceBetweenSSTandsLT!,ThedifferencebetweenthestandarddeviationssLTandsSTgivesanindicationofhowmuchbetteronecandowhenusingappropriateproductioncontrol,likeStatisticalProcessControl(SPC).,Short-termstandarddeviation:,Long-termstandarddeviation:,ThedifferencebetweensSTandsLT,ThedifferencebetweensSTandsLT,ThedifferencebetweensLTandsSTisonlyinthewaythatthestandarddeviationiscalculatedsLTisalwaysthesameorlargerthansSTIfsLTequalssST,thentheprocesscontroloverthelonger-termisthesameastheshort-term,andtheprocesswouldnotbenefitfromSPCIfsLTislargerthansST,thentheprocesshaslostcontroloverthelonger-term,andtheprocesswouldbenefitfromSPCThereliabilityofsLTisimprovedifthedataistakenoveralongerperiodoftime.AlternativelysLTcanbecalculatedonseveraloccasionsseparatedbytimeandtheresultscomparedtoseewhethersLTisstable,DefiningCpandPp,Thetoleranceareadividedbythetotalprocessvariation,irrespectiveofprocesscentring.,Section3.CpandCpkConcept,DefiningCpkandPpk,CpkandPpkIndexesaccountalsoforprocesscentring.,WhatistheDifferenceBetweenCpandCpk?,TheCpindexonlyaccountsforprocessvariabilityTheCpkIndexaccountsforprocessvariabilityandcenteringoftheprocessmeantothedesignnominalTherefore,CpCpkNOTE:SameappliesalsoforPpandPpk,WhatDoTheseIndexesTellUs?,SimplenumericalvaluestodescribethequalityoftheprocessThehigherthenumberthebetter,RequirementforCpandCpkis1.67min.RecommendationforPpandPpkis1.33min.,Thisleavesussomespaceforthevariation,i.e.asafetymargin,AreweabletoimproveourprocessbyusingSPC?,Ifindexislow,followingthingsshouldbegivenathought:,IstheproductdesignOK?,Aretolerancelimitssetcorrectly?,Istheprocesscapableofproducinggoodqualityproducts?Processvariation?DOErequired?,Isthemeasuringsystemcapable?(SeeGageR&R),Cpk-Witha2-sigmasafetymargin,RequirementforCpandCpkis1.67min.1.67isaratioof=5/3or10/6.,2*standarddeviation,2*standarddeviation,Cpk=1.67theprocessisCAPABLE,Cpk=2.0theprocesshasreachedSixSigmalevel,WhatDoTheseIndexesTellUs?,IfCp=Cpk,IfPp=Ppk,IfCpkCp,IfPpkPp,IfCp=Pp,IfCpk=Ppk,IfPp1,33,Correctiveactions,No,StabilizeProcess&VisualQuality,Step1-Preparation,ManufacturejigsGaugecalibrationGaugeR&R,Collectandmark5consecutivepartseveryhrsuntil20setshavebeencreated,PrepareCpkSpreadsheetfor100parts,Step2-Measureparts,Fillinmeasurementdata,Seechartspageforwarningpoints,CorrectiveActionPlanFlow,Mean,CheckforData(typing)Error,NoError,Error,Range(orVariation),ChecklastCalibration,NotOK,EvaluatewithMechanicalDesignifSpecificationcanbeadjusted?,CheckGR&Rforthisorsimilardimension,NotOK,BacktoStep3,No,BacktoStep2forthatpart/dimension,CheckTypeofError,Yes,Section6.MiniCpk,MiniCpk,TheMiniCpkStudymethodfollowsthesameprinciplesoftheProcessCapabilityStudymethodofconfidenceintervalsrelationstosamplesize.ThismeansthataCpk=1.67forasamplesizeof100pcsisstatisticallyequivalenttoaCpk=1.87forasamplesizeof40pcs.TheuseofMiniCpkstudymethodmaywellreducethemeasurementeffortatthesupplierandfastentheapprovaltime.ItisrecommendedtousetheMiniCpkstudymethodinthefollowingcases:Pre-qualificationofprocessese.gbeforemovingthetooltothefinalmoulderlocationRe-qualificationofprocessesaftermassapprovale.gwhenaftermajormaintenanceorcolourchangeCopytoolapprovalsRegulartoolapprovalswhenagreedbetweentheprogramandthesupplierNote!Forre-qualification(whenthetoolhasnotchanged)measurementsfromonly1cavitymaybeacceptable,MiniCpk&ConfidenceInterval,ConfidenceIntervalonCpk/PpkwithN=100,lowerCIonPpk1.33isPpk1.14ifN=40,lowerCItogetPpk1.14,needPpk1.47therefore:forN=40mustsettargetof1.47(lessvariationtooffsetforsmallersample)WhySmallerSampleLessmeasurementtime/cost(60%less)Fasterapproval(another60%?)RewardformorecapableprocessANDSTILLSTATISTICALLYSOUNDDECISION,ProcessCapabilityTrial40/100parts,Collectandmark20setsof5partsfromeachcavityatregularintervalsafterprocesshasstabilized.Theintervallengthwillbetheproductionruntimedividedby20Measurediamonddimensionsfrom1standlastpartsofeachsubgroupandrecordresultsincpktemplate(intotal40parts).IfCpk1.87andPpk1.47approve“MiniCpk”studyIfCpk1.67andPpk1.33orRerunCpktrialNote:TheclosertheCpkvalueisto1,87themorelikelyitistoachieve1,67bymeasuringallparts,ProcessCapabilityTrial40/100parts/10hrs,Step3aAnalyzeData,Measure1standlastpartafterpartshavestabilized,Step1-Preparation,ManufacturejigsGaugecalibrationGaugeR&R,PrepareCpkSpreadsheetfor40samples,Step2a-Measureparts,Fillinmeasurementdata,Yes,Stabilitywarnings,ProcessCapabilityOK,Yes,No,Cpk1,87Ppk1,47,CorrectiveActions,No,Seechartspageforwarningpoints,Measure2nd,3rdand4thpart,Fillinallmeasurementdata,Step2b-Measureparts,Step3bAnalyzeData,Yes,Stabilitywarnings,ProcessCapabilityOK,Yes,No,Cpk1,67Ppk1,33,CorrectiveActions,No,Seechartspageforwarningpoints,Cpk1,35Ppk1,05,GotoStep2b,PrepareCpkSpreadsheetFor100samples,Note!AsCpk/PpkvaluesdecreasesodoesthepropabilityforsuccesfulStep2b,ProcessCapabilityTrial40parts/10hrs,Yes,Yes,Step3AnalyzeData,Measureallpartsafterpartshavestabilized,ProcessCapabilityOK,Yes,No,Cpk1,87Ppk1,47,Correctiveactions,No,StabilizeProcess&VisualQuality,Step1-Preparation,ManufacturejigsGaugecalibrationGaugeR&R,Collectandmark2consecutivepartseveryhrsuntil20setshavebeencreated,PrepareCpkSpreadsheetfor40parts,Step2-Measureparts,Fillinmeasurementdata,Seechartspageforwarningpoints,Cpktrial25parts,Collect25consecutivepartsfromeachcavityafterprocesshasstabilizedMeasurediamonddimensionsfromallpartsIfPpk2,approve“MiniCpk”studyIfPpk2,Correctiveactions,No,StabilizeProcess&VisualQuality,Step1-Preparation,ManufacturejigsGaugecalibrationGaugeR&R,Collectandmark25consecutiveparts,PrepareCpkSpreadsheetfor25parts,Step2-Measureparts,Fillinmeasurementdata,Seechartspageforwarningpoints,Section7.DataCollectionSpreadsheetDMM00024,DataCollectionSheet(DMM00024-EN-8.0),Goto:NGSW,Suppliermanual1.3ComponentQualityPlanning&MonitoringDMM00024-EN-8.0CpkDataCollectionSpreadsheet,Open,enablemacrosandsavetolocalharddriveNamingofthespreadsheetdefaultiscpk-drawing#-tool#.xls(cpk-12345T03.xls)Thetemplatewillgivethatautomaticallybutwillnotsavewiththatname,DataCollectionSheet(DMM00024-EN-8.0),DataCollectionSheet(DMM00024-EN-8.0),DataCollectionSheet(DMM00024-EN-8.0),DefaultTableSizeis20subgroupsand5samplesinasubgroup(Whenrunning“minicpkthetablemustbesetaccordinglyUnusedcellswillbedeleted,DataCollectionSheet(DMM00024-EN-8.0)Histo