Semiconductor and IC Package Thermal Metrics …

1 1 SPRA953C December2003 RevisedApril2016 SubmitDocumentationFeedbackCopyright 2003 2016,TexasInstrumentsIncorporatedSemicon ductorand IC PackageThermalMetricsApplicationReportSP RA953C December2003 RevisedApril2016 Semiconductorand IC PackageThermalMetricsDarvinEdwards,HiepN guyenABSTRACTM anythermalmetricsexistfor semiconductorand integratedcircuit(IC) packagesrangingfromR JAto JT. Often,thesethermalmetricsare misappliedby thosewhotry to use themto estimatejunctiontemperaturesin documentdescribestraditionaland new thermalmetricsand puts theirapplicationin perspectivewith respectto JAJunction-to-Ambientand R JT, Junctionto Top of (JB) of Figures11s vs 2s2pPCBfor SizeImpacton a JAvs (JB) of Tables1 FactorsAffectingR JAfor a JTfor a registeredtrademarkof othertrademarksare the propertyof theirrespectiveowners.

2 ()TTRP owerJAJA=+ qR JAJunction-to-Ambientand R December2003 RevisedApril2016 SubmitDocumentationFeedbackCopyright 2003 2016,TexasInstrumentsIncorporatedSemicon ductorand IC PackageThermalMetrics1R JAJunction-to-Ambientand R JMAJ unction-to-MovingAirThe junction-to-ambientthermalresistance,R JA, is the mostcommonlyreportedthermalmetricand is JAis a measureof the thermalperformanceof an IC packagemountedon aspecifictest intentof R JAis to give a metricby whichthe relativethermalperformanceof apackagecan be ,the thermalperformanceof a TI devicecan be comparedto a is true whenbothcompaniesuse a standardizedtest to measureR JA, suchas that specifiedby JEDECin the EIA/JESD51-xseriesof ,however,JEDEC conditionsare not followedand the excursionsfromthe standardsare not havea dramaticeffecton the measuredvaluesof R JA.

3 Therefore,unlesstest conditionsare reportedwith the R JAvalue,they shouldbe measurementof R JAis performedusingthe followingsteps(summarizedfromEIA/JESD51- 1,-2, -5,-6, -7, and -9):Step1. A device,usuallyan integratedcircuit(IC) packagecontaininga thermaltest chip that canbothdissipatepowerand measurethe maximumchip temperature,is mountedon a The temperaturesensingcomponentof the test chip is The Package -and test-boardsystemis placedin eithera still air (R JA) or movingair (R JMA) A knownpoweris dissipatedin the test Aftersteadystateis reached,the junctiontemperatureis The differencein measuredambienttemperaturecomparedto the measuredjunctiontemperatureis calculatedand is dividedby the dissipatedpower,givinga valuefor R JAin ,R JAhas oftenbeenusedby systemdesignersto estimatejunctiontemperaturesof theirdeviceswhenusedin equationusuallyassumedto be validfor calculatingjunctiontemperaturefromR JAis.

4 (1)This is a misapplicationof the R JAthermalparameterbecauseR JAis a variablefunctionof not just thepackage,but of manyothersystemlevelcharacteristicssucha s the designand layoutof the printedcircuitboard(PCB)on whichthe part is effect,the test boardis a heatsink that is solderedtothe leadsof the designor configurationof the test boardchangesthe efficiencyofthe heatsink and thereforethe measuredR JA. In fact, in still-airJEDEC-definedR JAmeasurements,almost70% 95%of the powergeneratedby the chip is dissipatedfromthe test board,not fromthesurfacesof the systemboardrarelyapproximatesthe test couponusedto determineR JA, applicationof R JAusingEquation1 resultsin lists factorsthat can influenceR JAfor a givenpackageoutlinewhenall materialsare first columnlists the factorwhilethe secondcolumngivesarule of thumbestimateas to theimpactof the FactorsAffectingR JAfor a GivenPackageOutlineFactorsAffectingR JAStrengthof Influence(ruleof thumb)PCBdesignStrong(100%)Chipor pad sizeStrong(50%)Internalpackagegeometrica lconfigurationStrong(35%)AltitudeStrong( 18%)ExternalambienttemperatureWeak(7%)Po werdissipationWeak(3%)

5 01020304050607080 Dip 20 CuDip 24 CuQFP 100 CuQFP 120 A 42 QFP 132 A 42 QFP 132 CQFP 160 A 42 QFP 208 CuQFP 208 + 240 CuPLCC 20 CuPLCC 28 CuPLCC 68 CuPLCC 84 CuSOJ LOC 32 A42 SOJ LOC 32 CuSOIC 8 CuPackage TypeJ/ A (5C/W)01020304050601s2s2p% Shift216108% JAJunction-to-Ambientand R JMAJ unction-to-MovingAir3 SPRA953C December2003 RevisedApril2016 SubmitDocumentationFeedbackCopyright 2003 2016,TexasInstrumentsIncorporatedSemicon ductorand IC PackageThermalMetricsIn light of the fact that R JAis not a characteristicof the packageby itselfbut of the Package ,PCB,andotherenvironmentalfactors ,it is best usedas a comparisonof example,if TI reportsan R JAof 40 C/Wfor a packagecomparedto acompetitor'svalueof 45 C/W,the TI part will likelyrun 10%coolerin an applicationthanthe competitor' establisheda set of standardsfor measuringand reportingthe thermalperformanceof underthe has a set of thermalstandardsthat are substantiallydifferentfromthe JAis not a constant.

6 Therefore,it iscriticalto determinethe standardsthat wereusedto calculateor measureR JEDEC specification,two test boardtypesare 1s (singlesignallayer)configurationgivesaty picalusagevaluefor a moderatelypopulated, 2s2p(doublesignallayer,doubleburiedpower plane)configurationgivesa best caseperformanceestimateassuminga sparselypopulated,high-trace-densityboar ddesignwith buriedpowerand showsmodeledR JAdifferencesfor thesetwo boardsfor 17 allthe materialsand packagegeometrieswereheld constantfor 1s vs 2s2pPCBfor VariousPackagesAs shown,as muchas a 50%R JAvariationcan be expectedas a functionof 1s vs 2s2ptest SizeImpactThe chip or die pad insidea packagecan performthe samefunctionas a heatspreaderif the chip or padis functionof the heatspreaderis ,it spreadsenergyfromthe hot spot ofthe chip overa widerareaon the packagesurface, ,itincreasesheattransferfromthe pad to the lead fingersor to the packageballs.

7 Whichthenconducttheheatto the showsthe impactof die size on R JAfor a tape-basedareaarraychip scalepackage(CSP).As shown,the R JAfor this packagechangesalmost8 with die is importantto re-measureor recalculateR JAfor a packageif a die shrinkis Distance(mm)202224262830 ( C/W)0204060801001201401601802004 Die Size (mm)R ( qJAC/W)1s2s2p681012R JAJunction-to-Ambientand R December2003 RevisedApril2016 SubmitDocumentationFeedbackCopyright 2003 2016,TexasInstrumentsIncorporatedSemicon ductorand IC PackageThermalMetricsFigure2. Die SizeImpacton a topicrefersto the layoutwithina Package ,be it a traditionallead framepackage,smallpad (S-pad) Package ,lead-on-chip(LOC),or ball grid array(BGA) also havea majorimpacton the includethe distancebetweenthe tips of the lead in the packageand the die pad as shownin Figure3, or eventhe downsetbetweenthe pad and lead latteris an especiallyimportantthermalcriterionin thin BGAs,designof the interposertraceconfigurationis importantin spreadingheatfromthe die to thepackageballswhereit is conductedinto the R JAvs Pin-to-PadDistance()()

8 JAJA effectiveTTRP owerq=+ JAJunction-to-Ambientand R JMAJ unction-to-MovingAir5 SPRA953C December2003 RevisedApril2016 SubmitDocumentationFeedbackCopyright 2003 2016,TexasInstrumentsIncorporatedSemicon ductorand IC the air pressureof the ambientenvironmentchangeswith altitude,the coolingefficiencyof the air [1] showedthat a deviceis expectedto run 20%hotterat 8000ft comparedto the samedeviceoperatingat sea fan performanceand internalchassisair flow whenusedat considered,especiallywhenthesystemdesign is marginalfroma are usedto test theirsystemsat ,thesecompaniesinstrumenttheirdesignsto measureinternalcomponenttemperatureswhen operatingat lists multiplicationfactorstakenfromthe IBM workto derateR JAvaluesdeterminedat sea MultiplicationFactorsAltitude(ft) a greateffecton convectiveand 4th powerof temperature(T4)

9 ,radiativeheattransferis the contrary,convectiveheattransfersufferswi th increasingtemperatureasthe air is less denseat ,the effectof radiationis muchhigherthanthat offree the TI thermallab showabouta 10% 20%improvementin R JAwhenmeasuredbetweenan ambientof 0 C 100 C; that is, R JAat 100 C ambientis about20%lowerthantheR JAat 0 C surfacetemperatureof the devicedrivesbothconvectionand radiationenergyloss fromthe hotterthe packagesurfacebecomes,the moreefficientconvectionand radiationheatloss to ,R JAdecreaseswith increasingpowerdissipationfroma low powerdissipationwhichresultsin minimalincreasein surfacetemperature,R JAis sometimesfoundto be 2 3x higherthanat JAEffectiveTheta-ja(R JA) is a system-levelparameterthat dependsstronglyon systemparametersas describedinthe previoussections.

10 Therefore,it is sometimesusefulto definean R JA(effective), whichis simplythe R JAofthe deviceoperatingin the systemof R JA(effective)can be estimatedfromthermalmodelingormeasuremen tsin the system,it is possibleto use Equation1 to calculatethe junctiontemperatureassumingthe powerof the surroundingcomponentson the systemdoesnot thenbecomes:(2)The systemconditionsleadingto an R JA(effective)valueshouldalwaysbe definedwhenreportingR JA(effective).JAJCCARRRqqq=+Cold PlateCoolant InCoolant OutThermal GreaseTest CouponThermocoupleICR December2003 RevisedApril2016 SubmitDocumentationFeedbackCopyright 2003 2016,TexasInstrumentsIncorporatedSemicon ductorand IC PackageThermalMetrics2R JCJunction-to-CaseThe junction-to-casethermalresistancemetricw as originallydevisedto allowestimationof the thermalperformanceof a packagewhena heatsink was R JCis, the thermalresistancefromthe operatingportionof a semiconductordeviceto outsidesurfaceof the Package (case)closestto the chip mountingareawhenthat samesurfaceis properlyheatsunkso as to minimizetemperaturevariationacrossthat surface.