Thermal testing of LEDs: emerging standards - … - MENTOR GRAPHICS - POPPE.pdfJEDEC JSD51-1 static test method compliant thermal measurement system CIE 127-2007 compliant photometric

Thermal testing of LEDs:

emerging standards

András Poppe, PhD

Mentor Graphics MicReD Division, Budapest, Hungary

2 © 2011 Mentor Graphics Corporation

www.mentor.com/micred

A. Poppe: Thermal testing of LEDs - emerging standards

21 March 2011 / MEPTEC "The heat is on" Symposium

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

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

560 570 580 589 599 609 619 629 639 649 659 669 678 688

[nm]

Spectral power distribution [µW/nm] 20 oC

30 oC

40 oC

50 oC

60 oC

70 oC

IF = 300 mA

698

No doubt that reliable thermal data is a must for

power LEDs: widely accepted standards are

needed





Standardization status at CIE (from Y. Ohno)





Overview document (not yet accepted)

Each box represents recommendations for a particular problem. — New modules can be easily added

Approach of the JEDEC JC15 committee

Measurement

of AC LEDs

Measurement

of Rth during

LM80 tests

OVERVIEW

THERMAL MEASUREMENT

THERMAL ENVIRONMENT

COMPONENT MOUNTING

COMPONENT CONSTRUCTION

APPLICATION GUIDELINES

Electrical Test

Method

Guideline for

Estimation

Measurements

(as proposed

by NIST)

Additional

thermal

guidelines for

IESNA LM80

tests

???

???

???

MODELING

Model

validation

procedures Guideline for

combining CIE

127-2007

measurements

with thermal

measurements

Natural

Convection

Heat Sink

Forced

Convection

Heat Sink

Mounting

Thermal Test

Board

Test

Luminaires

Single Light

Source LED

Multi-Light

Source LED

Terms,

Definitions &

Units Glossary

???

Dynamic Compact

Thermal Model for

single light source

single heat-flow

path LEDs

??? Drafts

Issue identified which is dealt with

Recently identified issue which is dealt with

Yet to be identified and/or to be dealt with

JEDEC JC15 committee

actively works on these:

Today we

shall cover

these





A few words about thermal resistance of LEDs

Original definition in the JEDEC JESD51-1 document

Classically, for Si semiconductor diodes: Rth-el = ΔTJ / (IF×VF)

Accurate; the questions are: — what is the dissipated power of an LED? Subtract radiant flux

— what is the TX reference temperature Use cold plate!

For LEDs, consider the radiant flux: Rth-r = ΔTJ / (IF×VF – Popt)

Both Rth-el and Rth-r are correct, if proper power is used to calculate TJ





Importance of the definition of Rth for LEDs

Traditionally: Rth-el = TJ / Pel = TJ / (IF VF)

Due to high efficiency, radiant flux must be considered:

Rth-r = TJ / (Pel – Popt)

= TJ / (IF VF – Popt)

By neglecting Popt vendors report

much nicer data than reality

ηe = Popt/Pel (radiant efficiency)

EXAMPLE: Let us assume two ηe -s

T = 50oC, Pel = 10W

— ηe =0% (electrical only) "Rth-el" = T / Pel = 50/10 = 5 K/W

— ηe = 25%

Rth-r = T / (Pel – Popt) = T / [Pel (1-ηe)] = = 50/(100.75) = 6.67 K/W

— ηe = 50%

Rth-r = T / (Pel – Popt) = T / [Pel 1-ηe)] = = 50/(100.5) = 10 K/W





Junction temperature – performance indicator

Calculation: TJ = RthJ-X PH + TX — RthJ-X junction-to-reference_X thermal resistance supplied by the LED

vendor

— PH heating power measured/calculated by the LED user – How?

— TX 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

TJ

t

TJ1

TJ2

TJ

PH1

PH2

t

PH

PH

H

J

XthJP

TR

)(

Differential formulation of the thermal resistance

H

XJ

H

XJXthJ

P

T

P

TTR

][

Instead of spatial difference (temperature values at

junction and reference point) temporal difference of the

junction temperature can be used





How do we know TJ(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 SVF temperature sensitivity of the forward

voltage is obtained

Forward voltage change due to temperature change is measured using

a 4 wire setup (Kelvin setup)

IF VF(t) ~ TJ(t)

IM IH

Force (current) Sense (voltage)





The measurement waveforms

tMD

t

t=0

VF

IH

IM

VH

VF

IF

tH tM

heating

measurement

t

topt

VFf

VFi

cooling stable

Time window for the CIE 127-2007 compliant measurement of the light output





Comprehensive LED testing solution:

Popt(T,IF)

ηe(T,IF)

V(T,IF)

photometric/radiometric

measurements in thermal

steady-state

JEDEC JSD51-1 static test method compliant thermal measurement system

CIE 127-2007 compliant photometric & radiometric measurement system

steady-state

electrical

powering

thermal resistance/impedance

measurement

IM IH

Force

(current)

VF(t)

~

TJ(t)

Sense

(voltage)

IF V

F

Temp.controlled

heat sink

Detector

Aux. LED

Test LED

IF

Integrating

sphere

Thermal

test

equipment

switching-off

from IH to IM

calculate Rth-r and TJ

The JEDEC JC15 committee deals with these issues. A set of documents is prepared and is being discussed (how to apply JESD51-1 & CIE 127-2007)





Some details of the test environment

The JEDEC JC15 committee deals with these issues. A set of documents is prepared and is being discussed (how to apply JESD51-1 & CIE 127-2007).

CIE TC2-63 deals with optical testing of high power LEDs, considering also the effect of the junction temperature.





The Mentor Graphics MicReD implementation:

photo-

detector

V(), Xlong, Xshort, Z and flat response filters in a filter bank

DUT LED on TEC cooled

stage

control electronics

thermal transient

tester equipment

reference LED

It can be added to the system in a plug&play manner if the voltage of the base tester is not sufficient.

Special LED booster: allows high voltage across a LED line (overall forward

voltage can reach 280V – needed for AC mains driven LEDs).





What temperature to report?

The same luminous flux measurement results shown as function of

reference temperature and junction temperature

20 40 60 80 100 120 0

50

100

150

200

250

TJ [oC]

ΦV [lm] TERALED: Luminous Flux vs Junction Temperature

IF = 150mA IF = 200mA IF = 350mA IF = 500mA IF = 700mA

10 20 30 40 50 60 70 80 90 0

50

100

150

200

250

TERALED: Luminous Flux vs Reference Temperature

Tref [oC]

ΦV [lm]

IF = 150mA IF = 200mA IF = 350mA IF = 500mA IF = 700mA

The junction temperature is the

one which determines the light

output, this is the relevant

quantity.

The JEDEC JC15 committee deals with these issues. A set of documents is prepared and is being discussed (how to apply JESD51-1 & CIE 127-2007)

Case study: 10W white LEDs with their thermal

properties and light output characteristics as function of forward current and junction

temperature





Measured at 700 mA and 85 oC

— Structure functions of 3 samples, power corrected with Popt

— RthJC is identified in a way similar to the transient double interface method, a

new standard: JEDEC JESD51-14

FSF52 AL TG2500

RthJC real ≈ 2 K/W

Results for 10W white LEDs





The transient dual interface method for RthJC

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

RthJC

SEMI-THERM 2005 Best Paper

Award

Dirk Schweitzer - Harvey Rosten

Award (24 March 2011)





Results for 10W white LEDs

Measured at 350/700 mA & between 15 oC and 85

oC

— Structure functions of sample AL-2, power corrected with Popt

LED package:

no variation

Al MCPCB &TIMs:

temperature

dependence

What thermal

resistances did we

measure?

RthJC-package + Rth-MCPCB + Rth-grease for "hot lumen" estimates

Changes in TIM quality contribute to light output variations





ΦV(Tref) plots for two cases (IF=350mA)

360

370

380

390

400

410

420

430

440

450

10 20 30 40 50 60 70 80 90

Lum

inous F

lux [lm

]

Tref [degC]

Cree MCE1 Luminous Fluxes at 350mA as function of Tref

'FSF52'

'TG2500-1'

Variation of Rth means, the device

characteristics scaled in reference

temperature will be different 360

370

380

390

400

410

420

430

440

450

20 30 40 50 60 70 80 90 100 110

Lum

inous F

lux [lm

]

Tj [degC]

Cree MCE1 Luminous Fluxes at 350mA as function of Tj

'FSF52'

'TG2500-1'

TJ = Tref + Rth-r (IF VF - Popt)

Re-scaling for junction temperature eliminates the

effect of the different thermal resistance values

No need for a sophisticated control

of the TEC in the integrated sphere

Further problems: Thermal issues in short-pulse testing

and in LM80 tests, AC LEDs




21 March 2011 / MEPTEC "The heat is on" Symposium 4 August 2010

Short pulse measurements

During in-line testing photometric/colorimetric properties are measured

with a short pulse — TJ = Tref = constant is assumed, THIS IS NOT TRUE:

In 10 ms significant junction temperature change may take place

During 10 ms TJ changes almost by 5

oC

1e-6 1e-4 0.01 1 100 10000

0

5

10

15

20

25

Time [s]

Norm

aliz

ed

te

mp

era

ture

ris

e [°

C]

T3Ster Master: Zth

dimco_kisf2 - Ch. 0

Question is if this causes big problems or not…

1W white LED measured in free

air without any cooling

assembly

Addressed by CIE TC 2-64





Example: 10W white LED

Pheat 3W @ 350mA

Rth-r 20K/W

1e-6 1e-5 1e-4 0.001 0.01 0.1 1 10 100 1000 0

20

40

60

80

100

120

140

Time [s]

Tem

pe

ratu

re r

ise [°

C]

Junction Temperature Response 0.01

15.0

15.3

Cree_MCE1_T25_2_F1_T25_I0350 - Ch. 0

Cree_MCE1_T25_2_F1_T25_I0700 - Ch. 0

TJ +15oC IF=350 mA

TJ +30oC IF=700 mA

t = 10 ms

15 20 25 30 35 40 5800

5900

6000

6100

6200

6300

T [°C]

CC

T [

K]

TCC vs. Temperature 15

22.80

48.53 IF = 350mA IF = 700mA

TCC+22 K for TJ=15 oC @ 350 mA

TCC+50 K for TJ=15 oC @ 700 mA

T = 15 oC

24 26 28 30 32 34 36 38 40

350

400

450

500

550

600

650

700

750

800

T [°C]

V [lm] Luminous Flux vs. Temperature 15

18.2

IF = 350mA IF = 700mA

V -18 lm for TJ=15 oC @ 350 mA

Slope -1.2 lm/oC

T = 15 oC Addressed by CIE TC 2-64





LM80 test chamber

with all the LEDs

assembled

In-situ light output

measurement

In-situ thermal transient

measurement All measurements are done in-situ to eliminate any Rth change which is NOT due to ageing

In-situ thermal measurements during LM80 tests





4.Osram LUWV5AM 350mA

85%

90%

95%

100%

105%

110%

0 200 400 600 800 1000 1200

41

42

43

44

45

46

aver.

Structure functions taken at

0h, 500h, 1000h

Relative Luminous Flux (Vendor O, IF = 350 mA)

Time [h]

sample #

In cooperation with University of Pannonia, Veszprém (Hungary), prof. J. Schanda’s group within the KöZLED project of the Hungarian Goverment

8 different kinds of LEDs from 4 vendors, so far 6500h burning time,

processing measurement data in progress

No change inside

the LED package

Ligh output drop likely due

to increased Rth caused by

TIM degradation, not by

LED degradation

Recent results from LM80 test of different LEDs





Results after 3000h:

All HK LEDs have already died

Recent results from LM80 test of different LEDs

0 2 4 6 8 10 12

1e-5

1e-4

0.001

0.01

0.1

1

10

100

Rth [K/W]

Cth

[W

s/K

] T3Ster Master: cumulative structure function(s)

15.1

Vendor O, sample #44, 0h




0.9

With real values of dVF/dT

0 2 4 6 8

5.6 4.7 0.6

1e-5

1e-4

0.001

0.01

0.1

1

10

100

Cth

[W

s/K

] T3Ster Master: cumulative structure function(s)

Rth [K/W]

TIM ageing: external to the

LED – LM80 measurement

results must be compensated

for this

Delemination from the MCPCB: a failure inside the

LED assembly, its contribution to light output

degradation is part of the LM80 test result

Vendor HK, sample #61, 0h Vendor HK, sample #61, 500h

Vendor HK, sample #61, 2000h Vendor HK, sample #61, 3000h

With real values of dVF/dT





Problems of testing AC LEDs: what “Zth” to use?

For AC LEDs instead of the classical time-domain representation of Zth we need its frequency domain representation

At higher frequency the absolute value of

Zth(ω) is smaller:

0

)(1

)( dtetaP

Z tj

dissDC

th

asscoolingthth

th

th

th

th

th

th ZRCj

RCj

RCj

Z _3

3

2

2

1

1

111)(





Problems of testing AC LEDs: dissipation, ZthAC

The AC dissipation for voltage generator driven LED:

...)(sin!3

1

)()(sin

!2

1

)()(sin

!1

1)( 4

3

043

2

03202 tmU

IUt

mU

IUt

mU

IUP

T

MAX

T

MAX

T

MAXdissAC

SSC Archie driven by 50Hz mains -350

-250

-150

-50

50

150

250

350

0 5 10 15

-15

-10

-5

0

5

10

15

U AC [V]

U_LED [V]

I LED [mA]

U [V] I [mA]

AC mains

ULED

ILED

0

500

1000

1500

2000

2500

3000

0 5 10 15

-15

-10

-5

0

5

10

15

PelAC [mW]

PoptAC [mW]

PdissAC [mW]

I LED [mA]

0

0.2

0.4

0.6

0.8

1

1 2 3 4 5 6 7 8

The first 8 harmonics of PdissAC

P [mW] I [mA]

ILED

PelAC

PdissAC

PoptAC

Multiple frequencies

How to define a single number thermal metric?

ZthAC-max = TJAC-max /PdissAC-mean

ZthAC-mean = TJAC-mean /PdissAC-mean

See details in our paper in the LED session of SEMI-THERM





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 127-2007)

— 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

Documents

Thermal testing of LEDs: emerging standards - … - MENTOR GRAPHICS - POPPE.pdfJEDEC JSD51-1 static test method compliant thermal measurement system CIE 127-2007 compliant photometric