是个简单的介绍，有兴趣的人看看，有一些HALT/HASS基本概念：
HALT/HASS—QualMark
————HASSFromConcepttoCompletionbyDavidRahe,QualMark
Whenamanufacturerofpowerconditioningequipmentwantedtodeterminewhyproductswereexperiencing
amajorfailuremodeinthefield,hecontactedoneofQualMark’sAcceleratedReliabilityTestCenters(ARTC)forhelp.AHASSprocesswasproposedandimplementedtoprovideasolution.
ThegoalofHASSistodevelopascreenthatiseffectiveincausinglatentfailurestobecomeactualfailureswithoutremovingasignificantamountoftheproduct’slife.AtARTC,theHASSprocessusesrapidthermaltransitioningcombinedwithmultilevelvibrationperformedoverafrequencybandwidthfrom2Hzto10kHz.Inadditiontothesestresses,product-specificstresses,suchasvoltagemargining,frequencymargining,andpowercycling,areapplied.
HALTpreviouslywasperformedonthisproductbyQualMarkARTC,andsubsequentlythecustomerrequested
aHASSscreentobedeveloped.TheHASSprocessincludedbothHASSdevelopmentandproductionHASS
andusedthesameacceleratedtechniquesasHALT,butderated.
ProductDescriptionandChronology
Thedetailsoftheuninterruptiblepowersupply(UPS)productwillnotbedisclosedtoprotectthecustomerandproductidentity.Theunitconsistedofametalchassiswiththemaincomponentsbeingaprintedcircuitboard(PCB),alargetransformer,andabattery.
Thetopplateofthechassiswasleftofftoallowfasterthermalchangeratestotheproduct.ThePCBwasmountedtothechassisbaseviaseveralstandoffs.
Thesystemtestedwasamatureproductwithmanyhundredsofunitsinthefield.Theproductwas
experiencingrandomfailuresoccurringattheend-usersite.ThefailuremoderesultedinseveralFETs
exploding,renderingtheUPSinoperable.
AnESSprogramhadbeenperformedoneachUPSpriortoshipment.Theprogramconsistedofa50￥Cburn-inatalmostfulloutputloading.Thiswasineffectiveinprecipitatingthisfailuremodeorothers.
HASSEquipmentTheHASSprofilewasdeveloped,tested,andproven(HASSdevelopment)inourlabusingtheOVS2.5HPHALT/HASSsystem.TheHASSdevelopmentprocessincludedfixturedesign,fabrication,qualification,profile
development,andproof-of-screen.Afterthedevelopmentprocess,thescreenwasimplemented(production
HASS)atthecustomer’sfacilityonitsOVS2.5HPsystem.
2AdditionalfunctionaltestequipmentwasusedtoverifytheoperationoftheUPStoitsperformance
specification.Thistestequipmentwaslocatedoutsideofthechamberandcabledviathechamberaccessporttotheunits-under-test.
FixtureDesignandQualification
TheHASSfixturewasdesignedandfabricatedtosupporttheproductthroughputtestingandprovidepropervibrationtransmissibility,thermaluniformity,andabalancedthermalrateofchange.Twofunctionalunitswere
securedtothetwofixturelocationswiththermocouples,andaccelerometerswereattachedtoeachunittoperformthermalandvibrationsurveys(Figure1).
ProfileDesignTheHASSprofilewasbasedontheHALTresults,alongwithotherproduct-relatedvariablessuchasfunctionaltestduration,thermallong-termdegradationeffects(long-termfailuremodesthatmayrequireareductionofstresslevels),product-specificstresses,andproductionthroughputtestrequirements.Theprofileconsistedof
twocombinedenvironmentcyclesperformedwithintheproduct’soperatinglimits.TheUOL,LOL,andDLvalueswereacquiredfromtheHALTresults.
Theinitialprofilelevels(rule-of-thumbestimates)were80%ofthecumulativerangeoftheUOLandtheLOL.
Forexample,withanUOLof100°CandLOLof-70°Cequalinga170°Crange,85°Cand-53°Cwereused.
X=[(LOL-UOL)(0.20)]/2
Where:XequaledthemargintemperaturesubtractedfromboththeLOLandUOL.Vibrationsetpointlevels
were50%oftheDL,butwithintheOL,asmeasuredontheproduct.Ifthe50%valuehadexceededtheOL,then80%oftheOLwouldhavebeenusedasthesetpoint.
Vibrationwasmodulatedthroughout,beginningat3to5Grmsandslowlyrampedtothemaximumlevel.Thisprocesswasrepeatedinreverse,fromthemaximumlevelto3to5Grms.
Dwelltimesateachtemperatureextremebeganwheretheproductsattainedthedesiredtemperature,and
wereheldaminimumof5minutesduringwhichtimethefunctionaltestswererun.Thevibrationmodulationrampswereslow,witha5-minutedwellminimumrequirementatthemaximumvibrationlevel.
Thesefactorswereconsideredinthedesignoftheprofile:
ThermalUOL:100°C.
ThermalUDL:100°C.
ThermalLOL:-70°C.
ThermalLDL:
[b]Proof-of-Screen（POS）[/b]
Proof-of-screenwasatwo-stepprocess.Itdeterminedhoweffectivethescreenwasindetecting
manufacturingflawsandprovedthatthescreenhadnotremovedsignificantlifefromthescreenedproducts.
ScreenEffectiveness
Thescreen’seffectivenesswasmeasuredbyitscapabilitytoprecipitatelatentdefects,suchascomponentweaknesses,PCBflaws,circuittimingproblems,mechanicaltoleranceproblems,solderdefects,andothermanufacturingassembly-relatedissuesassociatedwiththemanufacturer’sorvendor’sprocessgoingoutofcontrol.
Forthisprocess,unitsclassifiedasNTForparametricallymarginalfunctionalunitswouldhavebeenpreferred.
SinceNTFunitswerenotavailable,productionunitswereused.
Twofunctionalproductionunitswereseededwithflawsrepresentativeofthemanufacturingprocessgoingoutofcontrol,suchasapoorsolderprocessordamagedorincorrectcomponentinsertion.Theseededflawsweremicroprocessorpin24bentunderonbothunits,capacitorC618withaleadnickedonbothunits,andacoldsolderjointonathrough-holeleadofcapacitorC303onjustoneunit.
Thescreenshoulddetecttheseflaws.Ifnot,itshouldbemodified(increaseordecreaseseveritylevels)untilitdoes.
InitialScreenResults
Theseededtestunitswerescreenedusingtheprogrammedprofile(Figure2).Thisprofilesubjectedtheunitstotemperaturecyclesfrom+83°Cto-53°Catanaveragetransitionrateof45°Cperminutecombinedwithvibration.Vibrationwasmodulatedfromasetpointof5Grmsto30Grms.Thisinitialscreenproducedthefollowingfailures:
Theunitshadathermalfaultoccurringattemperaturesbelow-40°C,withtheunitlosingserial
communications.ThefailurealsooccurredduringHALT,butwasnotcarefullyanalyzedduringtheroot-causeanalysis.
ThefailurewasthoughttobecausedbyasmallPCBthatwasgoingtobereplacedwitharedesign.However,theresultsduringHASSdevelopmentindicatedthatthesourceofthisfailurewasnotthesmallPCB.Thiswasdeterminedbydisconnectingtheboardandverifyingthatthefailurecontinuedtooccur.Thetruerootcauseofthisfailurewasundeterminedatthistime.
Inunit#1,thebatterychargecircuitwasnotfunctioning,andtheredLEDindicatedbatteryservicewasneeded.Thisoccurredduringtherampfrom-53°Cto+83°Catapproximately-35°Conthesecondcycle.Thishardfailureremainedevenat25°C.
Theunitwasremovedandtestedwithabatteryonthebench.Onpower-up,itblewFETsQ6,Q11,Q22,andQ26,theidenticalfailurethatwasoccurringinthefield.
Uponcompletionofthescreen,avisualinspectionrevealedthattheseededflawC303haddislodgedfromthethrough-holefilletatthecoldsolderjoint.
Asaresultofthesefindings,thedamagedUnit#1wasreplacedwithUnit#3,andasecondscreenwas
performedusingthesameprofile.
SecondScreenResults
ThesamethermalissuesoccurredonUnit#2astheydidduringtheinitialscreenat
-53°C.
TheloadLEDsonUnit#2wereflashingonandoffrandomlyat-53°Cduringthesecondcycle.Thiswasahardfailurethatremainedat25°Cwithnovibration.Avisualinspectionoftheunitrevealedthatametal-oxidevaristor(MOV)componenthadbrokenfreefromtheboard.
ThedamagedUnit#2wasreplacedwithUnit#4,andathirdscreenwasperformedusingaslightlymodifiedprofile.Thisprofilewasprogrammedwithreducedlevels.Thethermallowerlimitwasincreasedto-30°C,andvibrationwasreducedto20Grms.Thisprofilesubjectedtheunitstotemperaturecyclesfrom+83°Cto-30°Catanaveragetransitionrateof45°Cperminutecombinedwithvibration.Vibrationwasmodulatedfromasetpointof5Grmsto20Grms.
ThirdScreenResults
Bothunitspassedthescreenwithnofailuresoranomalies.
ProductLifeValuation
ThisprocessdeterminedthedegreeofappreciableliferemainingintheproductsafterexposuretotheHASS
profile.Theconceptwastorepeattheprofilemanytimes,withoutfailuresoccurring,toshowthatproductionunitsexposedtoonlyonepassoftheprofilewouldstillhave90%minimumoftheusefulliferemaining,or10%maximumofliferemoved.Forgreaterconfidence,additionalrepeatcyclescouldhavebeenperformedontheseunits;however,timedidnotallow.
Production-levelunitswereusedforthisprocess,withbothfixturelocationspopulated.Thetestunitswerescreenedusingthelatestprofile,thenrepeated10times.Thisprofilesubjectedtheunitstotemperaturecycles
from+83°Cto-30°Catanaveragetransitionrateof45°Cperminutecombinedwithvibration.Vibrationwasmodulatedfromasetpointof5Grmsto20Grms.
ProductLifeValuationResults
Aninspectionoftheunitsfollowingcompletionofthe20cycles(oneprofile=twocycles)revealedthattheMOVcomponentRV301hadbrokenfreefromtheboard(bothleadswerebrokenattheboard)ontheupperunit.Thisfailurewasnotconsideredofconsequencebecausethecustomerplanedtouseroom-temperaturevulcanizing(RTV)adhesivetosupporttheMOVonfutureproductionunits.
SummaryTheHASSdevelopmenteffortbeganwithdesigningandqualifyingthefixturebymeasuringtheunit’sthermalandvibrationresponses.Then,theHASSprofilewasdevelopedusingtheunitlimitsdeterminedduringHALT,alongwithotherproduct-specificfactors.Remember,HALTwasrequiredbeforeHASScouldbeproperlyperformed.
DuringthedevelopmentoftheHASSprofile,adjustmentsweremadetotheprofiletooptimizeitsefficiency.
ThesechangeswerereflectedinthefinalprofileshowninFigure3.
Thecoldthermalfaultproblemcausingtheunittoloseserialcommunicationsattemperaturesbelow-40°C
neededmoreevaluationtodeterminetherootcause.Oncedetermined,theprofilewasexpandedandthe
proof-of-screenrerun.ThesamewastrueregardingtheRTVadhesiveontheMOV.TheRTVhadtobeappliedbetweentheMOVandPCBandtheproductlifevaluationrerun.
SuccessfulimplementationofHALTcandramaticallyreduceproductdevelopmentcycletimeandcostswhile
providingtheanswersneededtoproduceamorerobustdesign.SuccessfulimplementationofHASScan
eliminateinfantmortalityfailures,reduceproductsupportcosts,andincreasecustomersatisfaction.
Asuccessfulmanufacturingprogramincludes:
PerformHALTtounderstandandgaininformationoftheproduct’scapabilities.
PerformrootcauseanalysisontheHALTfailuresandimplementcorrectiveaction.
PerformHALTagainwiththenewfixesinplace.
RepeatSteps2and3untiltheproblemsareresolvedsatisfactorily.
PerformHASSdevelopment.
PerformHASSonproductionproducts.
ThemethodofHASSdevelopmentincludes:
Designandbuildafixturethatsupportsthroughputandotherrequirements.
Performfixturequalification.
CreateaHASSprofileofthermalandvibrationstimulibasedontheHALTresults.Inadditiontothese
stresses,applyotherproduct-specificstressesthatwilltestthefunctionalintegrityoftheproduct.
Performaproof-of-screenprocess.
Releasetheprovenprofiletobeginproductionscreening.
MonitortheHASSeffectivenessovertime.
HASSdevelopmentwasnotintendedtobearigidprocesswithanendpoint.Rather,itisalivingprocessthatmayneedmodificationoradjustmentoverthelifeoftheproduct.Asmoreislearnedabouttheproductovertime,includingtheHASSresults,theprofilemaychangeandevolveintoanevenbetter,moreeffectivescreen.
Acknowledgment
IwouldliketoacknowledgeDr.GregHobbsastheinventoroftheHALTandHASStermsandtechniques.
AbouttheAuthor
DavidRahe,directoroftechnicalsupportatQualMark,hasmorethan16yearsofexperienceinthe
environmentaltestindustry.HejoinedQualMarkin1996asthemanagingengineeroftheDenverARTCand
laterservedasdirectorofoperationsforfourwestern-regionARTCs.Mr.RahealsohasheldpositionsofengineeringmanagerandgeneralmanageratTest2.QualMark,AcceleratedReliabilityTestCentersDivision,1329W.121stAve.,Denver,CO80234,(303)254-8800.
GlossaryofTerms
DL—DestructLimit
ESS—EnvironmentalStressScreening
HALT—HighlyAcceleratedLifeTest
HASS—HighlyAcceleratedStressScreen
NTF—NoTroubleFound
LDL—LowerDestructLimit
LOL—LowerOperatingLimit
OL—OperatingLimit
ROC—RateofChange
UDL—UpperDestructLimit
UOL—UpperOperatingLimit