《Red X问题解决策略.docx》由会员分享，可在线阅读，更多相关《Red X问题解决策略.docx（42页珍藏版）》请在三一文库上搜索。
1、2009 World Conference on Quality and ImprovementA Convergent Approach toProblem SolvingHa DaoBill MaxsonMay 20, 2009Minneapolis, MNASQ Automotive Division1Ha Dao & Bill MaxsonFlowchart For Advanced Problem SolvingYESNOIs It Working?Dont Mess With It!YESDid You MessWith It?YOU IDIOT!NOAnyone ElseYESY
2、ESWill it Blow UpKnows?Youre SCREWED!In Your Hands?NONO Can You BlameNOHide ItSomeone Else?Look The Other Way YesNO PROBLEM!Copyright 2007Six Sigma Academy International,LLCPage 1A convergent 5-why ExampleProblemCookies taste really badCookies are undercookedIngredients are wrongUsed goose eggs rath
3、er thanchicken eggsRoot CauseRecipe did not specify birdtypeASQ Automotive Division3Ha Dao & Bill MaxsonProblem Solving & Prevention JourneyProblemProblem-FreeFree ProductsProducts andand ProcessesProcesses ThroughThrough ProactiveProactiveProblemProblem PreventionPreventionPreventingPreventing Prob
4、lemsProblemsProblemProblem SolvingSolvingFireFire FightingFightingWheredo youthinkyouaretoday?ASQ Automotive Division4Ha Dao & Bill MaxsonPage 2Presentation OutlineA Convergent Approach to Problem Solving Comparative Analysis Simple Explanation Brake Leaks Case StudyASQ Automotive Division5Ha Dao &
5、Bill MaxsonDifferent Paradigms Lead To Different Problem Solving Approaches Engineering Based Math Based Evidence BasedASQ Automotive Division6Ha Dao & Bill MaxsonPage 3Engineering BasedThe Strategy Talk to experts Insights into product or process (experience) Developing and testing a theory Insight
6、s into basic principles:- F=ma- V=IR- PV=nRTProblemPossible CausesSolutionEngine does not startFuel tank is empty or throtle valve is not primed.Fill fuel tank with fuel and prime throttle.Bad glow plug or dead plug booster battery.Replace glow plug and recharge or replacebattery.Fule lines, air cle
7、aner, or muffled is clogged.Clean or replace clogged parts.Eingine is flooded due to over priming.Remove glow plug and discharge fuel.Also, test the glow plug. Replace the glowplug if it is defective.Throttle valve not adjusted properly.Set idle and needle valve to standardstarting position.Servo li
8、nkage not adjusted properly.Move servo to neutral then readjust.Engine starts but thenFuel tank is empty.Fill fuel tank with fuel.stallsFuel lines, air cleaner, or muffler is clogged.Clean or replace clogged parts.ASQ Automotive Division7Ha Dao & Bill MaxsonEngineering BasedThe Tools Engineering bas
9、ed techniques: Fishbone Diagrams Brainstorming FMEAs 8 D Design Changes Trial and Error“We never have trouble identifying the root cause.We do struggle with implementing the solution.” Manufacturing ExecutiveASQ Automotive Division8Ha Dao & Bill MaxsonPage 4Math BasedThe Strategy Collect lots of dat
10、a Talk to experts y = f(x)y = a + bx1, + cx2 + dx3 + Modeling Simulation Knowledge of key inputs is importantMathematical relationships are moreimportant then physical relationships.ASQ Automotive Division9Ha Dao & Bill MaxsonMath BasedThe Tools Math Based Techniques: Designed Experiments Analysis o
11、f Variance Multiple RegressionAnalysis Monte Carlo Techniques Scatter Plots Box Plots SPC Etc.ASQ Automotive Division10Ha Dao & Bill MaxsonPage 5Evidence BasedThe Strategy Talking to the parts y = f ( x) y = a + b x1, + c x2 Finding extremes Process of eliminationA statistically sound approachwithou
12、t all the statisticsASQ Automotive Division11Ha Dao & Bill MaxsonEvidence BasedThe Tools Evidence Based TechniquesSystem Level Response - Multi VariCustomer Compliant- Concentration Diagram- Event to Energy TransformSub-assy. Level Response - Component SearchPerformance Output- Operations Search- Sc
13、atter PlotsDetailed Component - Paired ComparisonFeature- Group ComparisonManufacturing Process - etcProcess ParameterASQ Automotive Division12Ha Dao & Bill MaxsonPage 6Problem Solving StrategyUnderstanding the PhysicsConverging with astructured approach.EffectCauseLooking for a SolutionDiverging wi
14、th astructured approach.CauseEffectASQ Automotive Division13Ha Dao & Bill MaxsonTalking to the PartsTo converge on the root cause, you must eliminate suspects.ASQ Automotive Division14Ha Dao & Bill MaxsonPage 7The Pareto Principle and Problem SolvingPareto Means Just a Few Xs Cause Most of theYVaria
15、tion in Ycaused byvariation in each XXBXA XK XP XT XG XQ X FRed XASQ Automotive Division15Ha Dao & Bill MaxsonComparing the ExtremesFinding the Red X for Unit to Unit VariationWorst of the WorstBest of the Best(WOW)(BOB)Dont fall into the trap of comparing parts out of spec to those in spec.You will
16、 struggle and be disappointedASQ Automotive Division16Ha Dao & Bill MaxsonPage 8No Variation ?Most Technical Problems can be Solved as Variation ProblemsPrototype FailuresValidation FailuresA Single Field ReturnDamaged ProductError CodesASQ Automotive Division17Ha Dao & Bill MaxsonFinding ContrastCh
17、aracteristicPart to PartCharacteristicTime to TimeCharacteristicModel to ModelASQ Automotive Division18Ha Dao & Bill MaxsonPage 9Why Did the Liberty Bell Crack?CrackNo CrackToo MuchWeakEnergyPartPass / Fail is not adequateASQ Automotive Division19Ha Dao & Bill MaxsonEvent to Energy TransformTest par
18、ts to failure measure the energyEnergy to CrackWOWBOBASQ Automotive Division20Ha Dao & Bill MaxsonPage 10Solving a 10 Year Old Field Failure ProblemGold Wire Bond FailuresNine bond pads areThe wire bonder places a ball bondelectroplated with gold.on the ASIC, then puts a wedge orLead Framestitch bon
19、d on the bond pad.ASIC chipgoes here.The remaining traces are tin plated.Lifted wire bonds on pads 1 and 2 (vs. 3-9) based on field return analysis.ASQ Automotive Division21Ha Dao & Bill MaxsonSorting through the ConfusionWire Bond Failures By MonthWhat changed in the process?Tin contamination on fa
20、ilures higherthan experts recommend.FailureNon-failureAre they really extremes?XxxxxXxxxxxxxxxWhat gets us to an answer andCertain pads failing,are they weaker?solution the fastest?Lets find out!ASQ Automotive Division22Ha Dao & Bill MaxsonPage 11Pull Tester to Check Bond StrengthEvent to Energy Tra
21、nsformPullBall BondWedge BondASICBond fails here, results inintermittent or no connections.ASQ Automotive Division23Ha Dao & Bill MaxsonMulti-Vari of Current ProductionPull Strength for Pad Location27.525.0Reading22.520.017.515.0Pad12789ASICConclusion: Current production units exceed the minimum pul
22、l strength of 5 grams. Pad location 2 is consistently the weakest. Further analysis does not show any tin contamination.ASQ Automotive Division24Ha Dao & Bill MaxsonPage 12Comparing ExtremesTaking Advantage of Within Unit VariationDont ask, “What is wrong with the worst bond pads?” Ask, “What is dif
23、ferent between the best and worst?”Strengthof BondVariation inInefficientInput EnergyPoor AdhesionBond (Pad(Bonder)(Contamination)Stiffness)ASQ Automotive Division25Ha Dao & Bill MaxsonFinding the Red X for Within Unit VariationSEM of a Bond Wire withHigh Pull StrengthIdealBadLarge, matte-finishLarg
24、e area of streaks,area of deformation.indicative of wire-slip.SEM of a Lifted Bond WireSometimes has“rainbow” appearance.Conclusion: Gold wire isslipping relative to bond site.Ultrasonic witness waves (chatter) and streaksfrom ceramic capillary throughout wedge face.ASQ Automotive Division26Ha Dao &
25、 Bill MaxsonPage 13Confirming Stiffness Affects Pull StrengthASICWorst SiteBest SiteCurrent SiteIllustration of Pad #2Pull Strength (grams)Bond Strength by Location8 TrialsCurrent25BestWorst20151050 12345678TrialConclusion: The bond site furthest from the ASIC (Best site) has the higher pull strengt
26、h. This indicates that the lead is acting like a diving board. Changed bond site for pads #1 and #2.ASQ Automotive Division27Ha Dao & Bill MaxsonKnowledgeHighLow Deflection in the Z AxisPullStrength(grams)Minimum SpecificationZero% TinLowHigh( 8%)Conclusion: High Tin + High Deflection = Low Pull Str
27、ength.ASQ Automotive Division28Ha Dao & Bill MaxsonPage 14Strategy Drives your Tool SelectionUpperExtremeCharacteristicLowerExtremeRed XASQ Automotive Division29Ha Dao & Bill MaxsonPossible causes for airplane crashes?ASQ Automotive Division30Ha Dao & Bill MaxsonPage 15Brake Leaks Case StudyWheel Cy
28、linder Project Definition Convergence ConfirmationASQ Automotive Division31Ha Dao & Bill MaxsonBackground Customer complaining of brake leaks at vehicle assembly plant. Problem is found on15-20% of brakes tested. Increasing torque at connector port helps but does not eliminate all leakers. Initial c
29、ontainment was for corrosion. Did not eliminate the problem. Replacing the wheel cylinder eliminates the leak. Takes 4 hours to repai r each vehicl e.Connector PipeTubeNutConnectorWheelPortCylinderASQ Automotive Division32Ha Dao & Bill MaxsonPage 16Project Definition TreeEliminate Leaks on Drum Brak
30、e AssembliesFeatureEventDefectPropertyMalfunctionDestructiveEventEventLeaks Leaks M&DEvac andFillLeaks atOtherConnectorFittingsHole InterfaceConnectorPortFACTUAL - ApproachRationale: Leak is an event. Leaks occur when energy is applied Leaks detected at both M&D and Evac and Fill Confirmed fluid lea
31、king at Connector Hole InterfaceASQ Automotive Division33Ha Dao & Bill MaxsonWheel Cylinder Part SchematicFACTUAL - ApproachWheel CylinderConnector PipeLeaks at Brake Pipe & WheelTube NutCylinder Connector HoleConnector PortWheel CylinderASQ Automotive Division34Ha Dao & Bill MaxsonPage 17Cross-Sect
32、ion of JointFACTUAL - ApproachWheelTube NutCylinderFlared TubeLeak PathThru HoleThe joint consists of a tube nut and a flared tube. Thetube nut is torqued to the wheel cylinder, whichcompresses the flared tube to create a sealed joint.ASQ Automotive Division35Ha Dao & Bill MaxsonSolution TreeGreen Y
33、 = Leak Rate (ccm)FrequencyRequiredNoLeaksCurrentCustomer Enthusiasm LimitLeaksBOBLimitWOW0 ccm2 ccm20 ccmLeak RateFACTUAL - ConvergeRationale:M = 3.6 ccmP = 61.8 ccmASQ Automotive Division36Ha Dao & Bill MaxsonPage 18IsoplotFACTUAL - ApproachAir Leak Rate (ccm) tested at 55 psiMeasurement Systempas
34、sed Isoplot and caneffectively discriminatebetween BOBs and WOWs.ASQ Automotive Division37Ha Dao & Bill MaxsonSolution TreeGreen Y = Leak Rate (ccm)FrequencyRequiredNoLeaksCurrentCustomer Enthusiasm LimitLeaksBOBLimitWOW0 ccm2 ccm20 ccmLeak RateFACTUAL - ConvergeRationale:M = 3.6 ccmP = 61.8 ccm Pas
35、sed IsoplotASQ Automotive Division38Ha Dao & Bill MaxsonPage 19Solution TreeFACTUAL - ConvergeGreen Y = Leak Rate (ccm)FrequencyRequiredNoLeaksCurrentCustomer Enthusiasm LimitLeaksBOBLimitWOW0 ccm2 ccm20 ccmRationale:Leak RateM = 3.6 ccmP = 61.8 ccm Passed IsoplotJoint toVehicle toModel toJointVehicleModelASQ Automotive Division39Ha Dao & Bill MaxsonSolution TreeFACTUAL - ConvergeGreen Y = Leak Rate (ccm)FrequencyRequiredNoCurrentLeaksCustomer Enthusia