Thermal testing of LEDs: emerging standards

Transcription

1 Thermal testing of LEDs: emerging standards András Poppe, PhD Mentor Graphics MicReD Division, Budapest, Hungary

2 Why to deal with thermal issues in case of LEDs? Reliability is connected to thermal issues life time (failure mechanisms are thermally assisted) mechanical stress Optical properties strongly depend on temperature spectra emitted flux / efficiency / efficacy Spectral power distribution [µw/nm] 20 o C 30 o C 40 o C 50 o C 60 o C 70 o C I F = 300 ma [nm] No doubt that reliable thermal data is a must for power LEDs: widely accepted standards are needed 2 A. Poppe: Thermal testing of LEDs - emerging standards 698

3 Standardization status at CIE (from Y. Ohno) 3 A. Poppe: Thermal testing of LEDs - emerging standards

4 Approach of the JEDEC JC15 committee Overview document (not yet accepted) JEDEC JC15 committee actively works on these: OVERVIEW THERMAL MEASUREMENT THERMAL ENVIRONMENT COMPONENT MOUNTING COMPONENT CONSTRUCTION APPLICATION GUIDELINES MODELING Today we shall cover these Electrical Test Method Guideline for Estimation Measurements (as proposed by NIST) Guideline for combining CIE measurements with thermal measurements Heat Sink Natural Convection Forced Convection Drafts Heat Sink Mounting Thermal Test Board Test Luminaires??? Single Light Source LED Multi-Light Source LED?????? Terms, Definitions & Units Glossary Additional thermal guidelines for IESNA LM80 tests??? Dynamic Compact Thermal Model for single light source single heat-flow path LEDs Model validation procedures??? Measurement of AC LEDs Measurement of R th during LM80 tests Issue identified which is dealt with Recently identified issue which is dealt with Yet to be identified and/or to be dealt with Each box represents recommendations for a particular problem. New modules can be easily added 4 A. Poppe: Thermal testing of LEDs - emerging standards

5 A few words about thermal resistance of LEDs Original definition in the JEDEC JESD51-1 document Classically, for Si semiconductor diodes: R th-el = ΔT J / (I F V F ) Accurate; the questions are: what is the dissipated power of an LED? Subtract radiant flux what is the T X reference temperature Use cold plate! For LEDs, consider the radiant flux: R th-r = ΔT J / (I F V F P opt ) Both R th-el and R th-r are correct, if proper power is used to calculate T J 5 A. Poppe: Thermal testing of LEDs - emerging standards

6 Importance of the definition of R th for LEDs Traditionally: R th-el = T J / P el = T J / (I F V F ) Due to high efficiency, radiant flux must be considered: R th-r = T J / (P el P opt ) = T J / (I F V F P opt ) By neglecting P opt vendors report much nicer data than reality EXAMPLE: Let us assume two η e -s T = 50 o C, P el = 10W η e =0% (electrical only) "R th-el " = T / P el = 50/10 = 5 K/W η e = P opt /P el (radiant efficiency) η e = 25% R th-r = T / (P el P opt ) = T / [P el (1-η e )] = = 50/(100.75) = 6.67 K/W η e = 50% R th-r = T / (P el P opt ) = T / [P el 1-η e )] = = 50/(100.5) = 10 K/W 6 A. Poppe: Thermal testing of LEDs - emerging standards

7 Junction temperature performance indicator Calculation: T J = R thj-x P H + T X R thj-x junction-to-reference_x thermal resistance supplied by the LED vendor P H heating power measured/calculated by the LED user How? T X reference temperature (un)specified by the LED user Used in the design process to decide if the foreseen cooling is sufficient or not Not enough: in case of LEDs, prediction of hot lumens is also required Differential formulation of the thermal resistance R thjx T J T P H X [T P J H ] X Instead of spatial difference (temperature values at junction and reference point) temporal difference of the junction temperature can be used P H1 P H T J1 T J P H2 P H t R thj X TJ ( ) P H T J2 T J t 7 A. Poppe: Thermal testing of LEDs - emerging standards

8 How do we know T J (t)? The LEDs forward voltage under forced current condition can be used as a very accurate thermometer The change of the forward voltage (TSP temperature sensitive parameter) should be carefully calibrated against the change of the temperature (see JEDEC JESD51-1 and MIL-STD-750D) In the calibration process the S VF temperature sensitivity of the forward voltage is obtained I F V F (t) ~ T J (t) I H I M Force (current) Sense (voltage) Forward voltage change due to temperature change is measured using a 4 wire setup (Kelvin setup) 8 A. Poppe: Thermal testing of LEDs - emerging standards

9 The measurement waveforms Time window for the CIE compliant measurement of the light output I F I H heating stable cooling I M t opt t V F V H V Ff measurement V Fi V F t t H t=0 t MD t M 9 A. Poppe: Thermal testing of LEDs - emerging standards

10 Comprehensive LED testing solution: CIE compliant photometric & radiometric measurement system photometric/radiometric measurements in thermal steady-state Detector P opt (T,I F ) η e (T,I F ) V (T,I F ) steady-state electrical powering I F Test LED I F V F Temp.controlled heat sink Integrating sphere calculate R th-r and T J Aux. LED I H Force (current) I M Thermal test equipment Sense (voltage) V F (t) ~ T J (t) switching-off from I H to I M thermal resistance/impedance measurement JEDEC JSD51-1 static test method compliant thermal measurement system The JEDEC JC15 committee deals with these issues. A set of documents is prepared and is being discussed (how to apply JESD51-1 & CIE ) 10 A. Poppe: Thermal testing of LEDs - emerging standards

11 Some details of the test environment CIE TC2-63 deals with optical testing of high power LEDs, considering also the effect of the junction temperature. The JEDEC JC15 committee deals with these issues. A set of documents is prepared and is being discussed (how to apply JESD51-1 & CIE ). 11 A. Poppe: Thermal testing of LEDs - emerging standards

12 The Mentor Graphics MicReD implementation: Special LED booster: allows high voltage across a LED line (overall forward voltage can reach 280V needed for AC mains driven LEDs). V(), X long, X short, Z and flat response filters in a filter bank thermal transient tester equipment reference LED DUT LED on TEC cooled stage photodetector control electronics It can be added to the system in a plug&play manner if the voltage of the base tester is not sufficient. 12 A. Poppe: Thermal testing of LEDs - emerging standards

13 What temperature to report? The same luminous flux measurement results shown as function of reference temperature and junction temperature Φ V [lm] TERALED: Luminous Flux vs Junction Temperature I F = 150mA I F = 200mA I F = 350mA I F = 500mA I F = 700mA The JEDEC JC15 committee deals with these issues. A set of documents is prepared and is being discussed (how to apply JESD51-1 & CIE ) Φ V [lm] TERALED: Luminous Flux vs Reference Temperature I F = 150mA I F = 200mA I F = 350mA I F = 500mA I F = 700mA 0 T J [ o C] The junction temperature is the one which determines the light output, this is the relevant quantity T ref [ o C] A. Poppe: Thermal testing of LEDs - emerging standards

14 Case study: 10W white LEDs with their thermal properties and light output characteristics as function of forward current and junction temperature

15 Results for 10W white LEDs Measured at 700 ma and 85 o C Structure functions of 3 samples, power corrected with P opt TG2500 AL FSF52 R thjc real 2 K/W R thjc is identified in a way similar to the transient double interface method, a new standard: JEDEC JESD A. Poppe: Thermal testing of LEDs - emerging standards

16 The transient dual interface method for R thjc Original idea from 2005, standard JESD51-14 published in November 2010 Change of thermal interface quality at the case surface Divergence point in measured structure functions: case surface Change at the case: insulator inserted Measurement of 2 setups (2x3 min), structure functions Dirk Schweitzer - Harvey Rosten Award (24 March 2011) SEMI-THERM 2005 Best Paper Award R thjc 16 A. Poppe: Thermal testing of LEDs - emerging standards

17 Results for 10W white LEDs Measured at 350/700 ma & between 15 o C and 85 o C Structure functions of sample AL-2, power corrected with P opt What thermal resistances did we measure? LED package: no variation Al MCPCB &TIMs: temperature dependence R thjc-package + R th-mcpcb + R th-grease for "hot lumen" estimates Changes in TIM quality contribute to light output variations 17 A. Poppe: Thermal testing of LEDs - emerging standards

18 Φ V (T ref ) plots for two cases (I F =350mA) Luminous Flux [lm] Cree MCE1 Luminous Fluxes at 350mA as function of Tref 'FSF52' 'TG2500-1' T J = T ref + R th-r (I F V F - P opt ) Tref [degc] Variation of R th means, the device characteristics scaled in reference temperature will be different No need for a sophisticated control of the TEC in the integrated sphere Cree MCE1 Luminous Fluxes at 350mA as function of Tj Tj [degc] 'FSF52' 'TG2500-1' Re-scaling for junction temperature eliminates the effect of the different thermal resistance values 18 A. Poppe: Thermal testing of LEDs - emerging standards

19 Further problems: Thermal issues in short-pulse testing and in LM80 tests, AC LEDs

20 Normalized temperature rise [ C] Short pulse measurements During in-line testing photometric/colorimetric properties are measured with a short pulse T J = T ref = constant is assumed, THIS IS NOT TRUE: In 10 ms significant junction temperature change may take place dimco_kisf2 - Ch. 0 T3Ster Master: Zth 1W white LED measured in free air without any cooling assembly Question is if this causes big problems or not 10 5 Time [s] 0 1e-6 1e During 10 ms T J changes almost by 5 o C Addressed by CIE TC A. Poppe: Thermal testing of LEDs - emerging standards 21 4 August March / MEPTEC "The heat is on" Symposium

21 Example: 10W white LED Temperature rise [ C] CCT [K] Junction Temperature Response Cree_MCE1_T25_2_F1_T25_I Ch. 0 Cree_MCE1_T25_2_F1_T25_I Ch I F = 350mA I F = 700mA T CC vs. Temperature 15 T CC +50 K for T J =15 o 700 ma t = 10 ms 6000 T CC +22 K for T J =15 o 350 ma 20 T J +30 o C I F =700 ma T J +15 o C I F =350 ma Time [s] e-6 1e-5 1e T = 15 o C T [ C] P heat 350mA R th-r 20K/W V [lm] Luminous Flux vs. Temperature I F = 350mA I F = 700mA Addressed by CIE TC Slope -1.2 lm/ o C V -18 lm for T J =15 o 350 ma T = 15 o C 18.2 T [ C] A. Poppe: Thermal testing of LEDs - emerging standards 21 4 August March / MEPTEC "The heat is on" Symposium

22 In-situ thermal measurements during LM80 tests LM80 test chamber with all the LEDs assembled All measurements are done in-situ to eliminate any R th change which is NOT due to ageing In-situ thermal transient measurement In-situ light output measurement 22 A. Poppe: Thermal testing of LEDs - emerging standards

23 Recent results from LM80 test of different LEDs In cooperation with University of Pannonia, Veszprém (Hungary), prof. J. Schanda s group within the KöZLED project of the Hungarian Goverment 110% 105% 100% 95% 90% 85% 4.Osram Relative LUWV5AM Luminous 350mA Flux (Vendor O, I F = 350 ma) Structure functions taken at 0h, 500h, 1000h sample # Time [h] aver. Ligh output drop likely due to increased R th caused by TIM degradation, not by LED degradation No change inside the LED package 8 different kinds of LEDs from 4 vendors, so far 6500h burning time, processing measurement data in progress 23 A. Poppe: Thermal testing of LEDs - emerging standards

24 C th [Ws/K] C th [Ws/K] Recent results from LM80 test of different LEDs Results after 3000h: T3Ster Master: cumulative structure function(s) Vendor O, sample #44, 0h Vendor O, sample #44, 500h Vendor O, sample #44, 2000h Vendor O, sample #44, 3000h With real values of dv F /dt T3Ster Master: cumulative structure function(s) e e All HK LEDs have already died R th [K/W] TIM ageing: external to the LED LM80 measurement results must be compensated for this A. Poppe: Thermal testing of LEDs - emerging standards e-4 1e-5 Vendor HK, sample #61, 0h Vendor HK, sample #61, 500h Vendor HK, sample #61, 2000h Vendor HK, sample #61, 3000h With real values of dv F /dt R th [K/W] Delemination from the MCPCB: a failure inside the LED assembly, its contribution to light output degradation is part of the LM80 test result

25 Problems of testing AC LEDs: what Z th to use? For AC LEDs instead of the classical time-domain representation of Z th we need its frequency domain representation At higher frequency the absolute value of Z th (ω) is smaller: Z 1 jt th( ) a( t) e dt PdissDC 0 Z th ( ) R th1 Rth2 Rth3 Zthcooling _ ass jcth 1 jcth2 jcth3 25 A. Poppe: Thermal testing of LEDs - emerging standards

26 Problems of testing AC LEDs: dissipation, Z thac The AC dissipation for voltage generator driven LED: P dissac ( ) U 2 MAX I0 mu T 1 1! sin 2 ( t) U 3 MAX I0 ( mu T ) 2 1 sin 2! 3 ( t) U 4 MAX I0 ( mu Multiple frequencies T ) 3 1 sin 3! 4 ( t) U [V] AC mains I [ma] P [mw] I [ma] U LED I LED U AC [V] U_LED [V] I LED [ma] P elac P dissac I LED PelAC [mw] PoptAC [mw] PdissAC [mw] I LED [ma] SSC Archie driven by 50Hz mains P optac How to define a single number thermal metric? The first 8 harmonics of P dissac Z thac-max = T JAC-max /P dissac-mean 0.6 Z thac-mean = T JAC-mean /P dissac-mean See details in our paper in the LED session of SEMI-THERM A. Poppe: Thermal testing of LEDs - emerging standards

27 Conclusions Brief overview of standardization needs and recent activities was given Measurement setup for consistent measurement of thermal and light output metrics of power LEDs was shown Based on existing standards (JEDEC JESD51-1, CIE ) Therefore with some new measurement guidelines it is easily implemented Such guidelines are being developed by the JEDEC JC15 Committee on Thermal Standards of Packaged Semiconductor Devices Merits of such a combined thermal/radiometric LED testing station were shown by a case study Importance of changing properties of thermal interface materials and their effect on light output was shown Combined thermal transient and photometric measurements suggest that the constant junction temperature assumption in short pulse in-line testing is not valid To eliminate effect of variations (ageing) of TIM during LM80 tests, in-situ measurements are suggested, combined with thermal transient measurements Problems related to the AC thermal impedance as a single number thermal metric for AC mains driven LEDs were raised 27 A. Poppe: Thermal testing of LEDs - emerging standards 21 4 August March / MEPTEC "The heat is on" Symposium