Light output characteristics of power LEDs considering their real thermal resistance

Transcription

1 Light output characteristics of power LEDs considering their real thermal resistance András Poppe 1,2 Gábor Molnár 2 Albin Szalai 1 1 Budapest University of Technology & Economics, Department of Electron Devices, Hungary 2 Mentor Graphics MicReD Division, Budapest, Hungary Lux et Color Vespremiensis '09

2 Spectral intensity [W/nm] Why to deal with thermal issues? 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 distribution of light output of a 1W red LED at different current levels and different temperatures T25_I200 T25_I500 T50_I200 T50_I500 T75_I200 T75_I Wavelength [nm] No doubt that reliable thermal data is a must for power LED based lighting design 2

3 Main drive for thermal characterization from a manufacturer point of view Fair comparison with competitors' data In an ideal world: provide real support for customers from a customer point of view Are temperatures within specs? Starting point: T J = R th J-refX P + T X T X from (often unspecified) measurements (own responsibility) P from estimated power dissipation (own responsibility) R th J-refX from component datasheets (other s responsibility) If T J calculated > T specification Redesign! In case of lighting: specified lumens at operating temperature 3

4 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-real = 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 WPE = P opt /P el 4

5 Example: Let us assume two WPE-s (wall plug efficiency: P opt /P el ) T J = 50 o C, P el = 10W "R th-el " = T J / P el = 50/10 = 5 K/W WPE = 25% R th-real = T J / (P el P opt ) = T J / [P el (1-WPE)] = = 50/( ) = 6.67 K/W WPE = 50% R th-real = T J / (P el P opt ) = T J / [P el (1-WPE)] = = 50/(10 0.5) = 10 K/W By neglecting P opt vendors report much nicer data than reality 5

6 Where do we need real R th of LEDs? Once thermal performance is known, light output characteristics can also be identified for a given temperature. The question is: Shall I obtain specified luminous flux from my luminaire at a given ambient temperature? Many times computer simulation is used to answer this question. 6

7 Characterization of a 1W LED lamp The complete lamp With lense removed: Internal details of the assembly: 7

8 Characterization of a 1W LED lamp Thermal transient measurement with a thermal transient tester (T3Ster) in 1ft 3 JEDEC standard still-air chamber: resembles real application 8

9 Normalized temperature rise [ C] Normalized temperature rise [ C] Results for the 1W LED lamp Z th curves: R thja 22K/W (independent of fixture) T3Ster Master: Zth 25 T3Ster Master: Zth 24 dimco_kisf2 - Ch. 0 dimco1_nagyg - Ch dimco_kisf2 - Ch. 0 dimco1_nagyg - Ch Time [s] 0 1e-6 1e Time [s] 9

10 Cth [Ws/K] Details of the heat flow path T3Ster Master: cumulative structure function(s) dimco_kisf2 - Ch. 0 dimco1_nagyg - Ch. 0 LED on MCPCB grease luminaire e-4 radial spreading in the * shaped MCPCB glue / solder between MCPCB and bottom of heat slug conical heat-spreading in the Cu heat-slug die attach between the Si submount and the heat-slug Si submount bumps between the Si submount and the saphire subtstrate with the active layer (flip-chip) LED chip on saphire substrate flip-chip assembled onto a Si submount 45 o C 22 o C junction Rth [K/W] This R th -C th map is derived from the measured T J transient. It helps identify the details of the junction to ambient heat flow path. ambient 10

11 Cth [Ws/K] Cth [Ws/K] Decomposition of the problems For modeling / design of lighting applications LED packages are replaced by their compact models These models must be scaled in real thermal resistance T3Ster Master: cumulative structure function(s) T3Ster Master: cumulative structure function(s) dimco_kisf2 - Ch. 0 LED dimco1_nagyg on MCPCB - Ch. 0 grease luminaire dimco_kisf2 - Ch. 0 LED dimco1_nagyg on MCPCB - Ch. 0 grease luminaire e-4 1e Rth [K/W] Rth [K/W] Luminaires can be described by a detailed model e.g. for CFD analysis 11

12 Measuring consistent metrics: JEDEC JSD51-1 static test method compliant thermal measurement steadystate electrical powering for CIE compliant photometric & radiometric measurement DUT LED on cold-plate detector record thermal data, calculate T J and R th-real switching-off photometric/radiometric measurements in thermal steady-state P opt (T,I F ) WPE(T,I F ) V (T,I F ) 12

13 Such a system in practice Such a setup can be used to identify the real thermal resistance of LEDs 13

14 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 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 14

15 Luminous Flux [Lm] Temperature rise [ C] T [K] Example: 10W Cree MCE white LED TeraLed: Color Temperature T3Ster Master: Smoothed response Cree_MCE1_T25_2_F1_T25_I Ch. 0 Cree_MCE1_T25_2_F1_T25_I Ch I = 350mA I = 700mA T J +15 o C for t=10ms e-6 1e-5 1e Time [s] P heat 350mA R threal 20K/W TCC +48 K for T J =15 o 700mA TeraLed: Luminous Flux vs Ambient T [ C] Temperature TCC +22 K for T J =15 o 350mA I = 350mA I = 700mA V -18 lm for T J =15 o 350mA T [ C] Slope -1.2 lm/ o C 15

16 Cth [Ws/K] R th of 10W Cree MCE white LEDs Measured at 350/700 ma & between 15 o C and 85 o C Corrected with emitted optical power T3Ster Master: cumulative structure function(s) Corr_CREE_MCE_AL_F1_T15_I Ch. 0 Corr_CREE_MCE_AL_F1_T15_I Ch. 0 Corr_CREE_MCE_AL_F1_T25_I Ch. 0 Corr_CREE_MCE_AL_F1_T25_I Ch. 0 Corr_CREE_MCE_AL_F1_T40_I Ch. 0 Corr_CREE_MCE_AL_F1_T40_I Ch. 0 Corr_CREE_MCE_AL_F1_T55_I Ch. 0 Corr_CREE_MCE_AL_F1_T55_I Ch. 0 Corr_CREE_MCE_AL_F1_T70_I Ch. 0 Corr_CREE_MCE_AL_F1_T70_I Ch. 0 Corr_CREE_MCE_AL_F1_T85_I Ch. 0 Corr_CREE_MCE_AL_F1_T85_I Ch. 0 4 different thermal management solutions were studied e Rth [K/W] Temperature dependence of R th means, the device characteristics scaled in reference temperature will be different even for the same device. 16

17 Test environment Due to high dissipation, liquid cooled cold plate was used: Fixture on the cold plate accommodates the integrating sphere Also very useful in long-term stability studies 17

18 Φ V (T ref ) plots for two cases (I F =350mA) Cree MCE1 Luminous Fluxes at 350mA as function of Tref 'FSF52' 'TG2500-1' Tref [degc] Variation of R th means, that the device characteristics scaled in reference temperature will be different 18

19 Φ V (T J ) plots for two cases (I F =350mA) Cree MCE1 Luminous Fluxes at 350mA as function of Tj 'FSF52' 'TG2500-1' T J = T ref + R th-real (I F V F - P opt ) Slope -1.1 lm/ o C Tj [degc] Re-scaling for junction temperature eliminates the effect of the different thermal resistance values 19

20 Φ V (I F,T J ) plots for 7 different samples Cree MCE1 Luminous Fluxes at 350mA and 700mA, 1st & 2nd series mA 350mA '350-Lum/Al-1' '350-Lum/Al-2' '350-Lum/Cu-1' '350-Lum/FSF52' '350-Lum/TG2500-1' '350-Lum/TG2500-2' '700-Lum/Al-1' '700-Lum/Al-2' '700-Lum/Cu-1' '700-Lum/FSF52' '700-Lum/TG2500-1' '700-Lum/TG2500-2' Here we see the real variation among the LED samples Tj [degc] Re-scaling for junction temperature eliminates the effect of the different thermal resistance values 20

21 η V (I F,T J ) plots for 7 different samples Cree MCE1 Efficacies at 350mA and 700mA, 1st & 2nd series mA 350mA '350-Eff/Al-1' '350-Eff/Al-2' '350-Eff/Cu-1' '350-Eff/FSF52' '350-Eff/TG2500-1' '350-Eff/TG2500-2' '700-Eff/Al-1' '700-Eff/Al-2' '700-Eff/Cu-1' '700-Eff/FSF52' '700-Eff/TG2500-1' '700-Eff/TG2500-2' Here we see the real variation among the LED samples Tj [degc] Re-scaling for junction temperature eliminates the effect of the different thermal resistance values 21

22 Some conclusions Temperature is a key factor in the operation of LEDs, not only in terms expected life time but also in terms of performance: lower operating temperatures ensure more light output. Despite thermal standardization activities, today s LED data sheets still do not provide the real thermal resistance values of LEDs since at most of the LED vendors thermal measurements are not yet combined with the measurement of the light output characteristics. This results in reporting smaller thermal resistance data then reality. Knowing real thermal resistance of LEDs is vital in luminaires' thermal design 22

23 Some conclusions By knowing the exact thermal resistance & heating power of LEDs, light output characteristics can be presented as functions of the real junction temperature. By doing so, variations in the actual thermal resistance in the test environment can be eliminated from the test results. This needs to be cared about in long-term stability studies as well: luminous flux measurement and thermal resistance measurement should be combined. Acknowledgments This work was partially supported by the KÖZLED TECH_08-A4/ project of the Hungarian National Technology Research and Development Office. 23