lm-85 research€¦ · high-performance led test solutions. 1 © vektrex | info@vektrex.com lm-85...

High-performance LED test solutions.

© Vektrex | info@vektrex.com

High-performance LED test solutions.

1

LM-85 Research Preliminary Findings

© Vektrex | info@vektrex.com

Research Question – Are the three LM-85 methods equivalent?

•  All tie measurements to a specific junction temperature

•  Assumption: o  If Tj(measurement 1) = Tj(measurement 2), o  Then Φ (measurement 1) = Φ (measurement 2)

•  But is it always the same?

2

© Vektrex | info@vektrex.com

Possible LM-85 error sources

•  Phosphor temperature differences

•  Current rise/fall time

•  Current shape

•  Pulse width errors

•  Measurement timing

3

Research by Yuqin Zong of NIST indicated phosphor could be a significant error source

© Vektrex | info@vektrex.com

LM-85 research goals

•  Compare three LM-85 methods

•  Determine if phosphor produces an error

•  Expose, quantify other error sources

•  Develop guidelines for improved methods

4

© Vektrex | info@vektrex.com

Vf monitor plot shows final 0.3C of temperature stabilization

5

500µV Vf change

© Vektrex | info@vektrex.com

Methodology •  Use identical LEDs for phosphor/no phosphor tests

o  Royal blue pump o  White

•  Best-of-class instrumentation o  Back thinned, temperature stabilized spectrometer o  TEC platform with 0.01C stability o  Precision pulsed current source with microamp stability o  Electro-Optical “LED Bench” software o  Fast sampling precision digitizer

6

© Vektrex | info@vektrex.com

Testing steps

1)  DC: Platform set to 25C, Tj was allowed to rise, platform maintained at 25C, untriggered measurement with ms integration time

2)  Continuous Pulse: Platform temperature adjusted to match DC Vf, untriggered measurement with 100X DC integration time

3)  Single Pulse: Platform temperature adjusted to match DC Vf, triggered measurement with 10ms delay, DC integration time

7

© Vektrex | info@vektrex.com

Precise digitizer timing allows temperature matching within 0.03C

•  SpikeSafe precise triggering combined with highly accurate digitizer allows Vf to be compared precisely

•  Resulting measurements are taken with junction temperatures matching to within 0.03C

8

Digitizer can acquire and average Vf over a precise window

© Vektrex | info@vektrex.com

Lumileds provided sample LEDs, other samples purchased on-line

9

© Vektrex | info@vektrex.com

Samples were stabilized for 100 hours prior to testing •  Mounted on copper load boards

•  100 hour burn-in at 55C

10

Vektrex ITCS LM-80 Chamber

© Vektrex | info@vektrex.com

Photometric system uses integrating hemisphere

11

© Vektrex | info@vektrex.com

Photometric System Block Diagram

12

Instrument Systems CAS-140 Compact Array Spectrometer

Control Panel Software

Vektrex High Power TEC Controller

Vektrex SpikeSafe Pulsed Source/Measure Unit 5A Current Source,

500ksps Digitizer

Precision Temperature Control Platform

Diffuser

Instrument Rack

Load Board

Hardware Trigger

Force

Sense

Integrating Hemisphere

SpecWinPro Software

© Vektrex | info@vektrex.com

Precision pulsed source measure unit with sub-microsecond pulse accuracy allows accurate continuous pulse measurements

13

Vektrex SpikeSafe SMU 500mA 100µs pulse, note minimal overshoot, fast settle time

Keithley 2600B 500mA 100µs pulse, note large overshoot and ringing that persists for 70µs

70µs

© Vektrex | info@vektrex.com

In LM-85 Continuous Pulse method current transitions during integration time

14

DC

Single Pulse

Continuous Pulse

Spectrometer Integration

Time

Cur

rent

0

I

0

I

Cur

rent

Time

Time

© Vektrex | info@vektrex.com

Expected and Unexpected Error Sources

15

DC Single Pulse Continuous Pulse

Amplitude + or - Pulse Width N/A Usually N/A + or - Rise/fall Time N/A Usually N/A +, increases

with shorter pulse widths

Phosphor Temperature

N/A +, may be small

+

Unknown Amber LEDs

N/A -

© Vektrex | info@vektrex.com

The amp-second product of square and sloped pulses of equal width is the same

16

Actual 100µs Current Pulse With 20µs rise/fall

Ideal 100µs Current Pulse Cur

rent

(A)

Time (µs)

0

0.2

0.4

0.6

0.8

1

1.2

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120

© Vektrex | info@vektrex.com

But LEDs produce more light at lower currents

17

Lum

inou

s Fl

ux (l

m)

Forward Current (mA)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 200 400 600 800 1000

Cree XP-E Green LED

Ideal LED

64%

50%

© Vektrex | info@vektrex.com ACCURACY!POWER |!

To Minimize Errors, Use Short Pulses!with Fast Rise and Fall Times!

Measurement Error Due To Nonlinear LED Output

Pulse Rise/Fall Time (µs)

Pulse Width

20µs 4%

<0.5% 1%

8%

100µs 50µs 200µs

2%

0.5% 1% 2%

0.25%

5µs

InstrumentRiseTimeSpecification

Source1 2µsSource2 30µs

© Vektrex | info@vektrex.com

The result is a positive error that gets larger as pulses are made more narrow

19

Measurement Error Due To Nonlinear LED Output

Pulse Rise/Fall Time (µs)

Pulse Width

20µs 4%

<0.5% 1%

8%

100µs 50µs 200µs

2%

0.5% 1% 2%

0.25%

5µs

InstrumentRiseTimeSpecification

Source1 2µsSource2 30µs

© Vektrex | info@vektrex.com

Test Data

20

White RoyalBlue Amber AmberCreeDC,Initial 0% 0% 0% 0%Continuous,100us 0.73% -0.12% -0.84% -0.78%Continuous,50us 1.11% 0.06%Continuous,20us 1.82% 0.80%Continous,10us 2.69% 1.72%SinglePulse 0% -0.02% -0.63% -0.86%DC,Final 0.12% -0.16% -0.02% 0.14%

© Vektrex | info@vektrex.com

Test Data & Analysis

21

White RoyalBlue Amber AmberCreeDC,Initial 0% 0% 0% 0%Continuous,100us 0.73% -0.12% -0.84% -0.78%Continuous,50us 1.11% 0.06%Continuous,20us 1.82% 0.80%Continous,10us 2.69% 1.72%SinglePulse 0% -0.02% -0.63% -0.86%DC,Final 0.12% -0.16% -0.02% 0.14%

Initial and final DC readings indicate parts were stable during testing

© Vektrex | info@vektrex.com

Test Data & Analysis

22

White RoyalBlue Amber AmberCreeDC,Initial 0% 0% 0% 0%Continuous,100us 0.73% -0.12% -0.84% -0.78%Continuous,50us 1.11% 0.06%Continuous,20us 1.82% 0.80%Continous,10us 2.69% 1.72%SinglePulse 0% -0.02% -0.63% -0.86%DC,Final 0.12% -0.16% -0.02% 0.14%

Roughly 1% difference between white and royal blue Continuous Pulse measurements correspond to phosphor error

© Vektrex | info@vektrex.com

Test Data & Analysis

23

White RoyalBlue Amber AmberCreeDC,Initial 0% 0% 0% 0%Continuous,100us 0.73% -0.12% -0.84% -0.78%Continuous,50us 1.11% 0.06%Continuous,20us 1.82% 0.80%Continous,10us 2.69% 1.72%SinglePulse 0% -0.02% -0.63% -0.86%DC,Final 0.12% -0.16% -0.02% 0.14%

Matching DC and Single Pulse readings indicate phosphor effect does not extend beyond 10ms measurement delay used

© Vektrex | info@vektrex.com

Test Data & Analysis

24

White RoyalBlue Amber AmberCreeDC,Initial 0% 0% 0% 0%Continuous,100us 0.73% -0.12% -0.84% -0.78%Continuous,50us 1.11% 0.06%Continuous,20us 1.82% 0.80%Continous,10us 2.69% 1.72%SinglePulse 0% -0.02% -0.63% -0.86%DC,Final 0.12% -0.16% -0.02% 0.14%

Increasing values with decreasing pulse width correspond to expected rise/fall error

© Vektrex | info@vektrex.com

Test Data & Analysis

25

White RoyalBlue Amber AmberCreeDC,Initial 0% 0% 0% 0%Continuous,100us 0.73% -0.12% -0.84% -0.78%Continuous,50us 1.11% 0.06%Continuous,20us 1.82% 0.80%Continous,10us 2.69% 1.72%SinglePulse 0% -0.02% -0.63% -0.86%DC,Final 0.12% -0.16% -0.02% 0.14%

Amber LEDs show unexplained -0.8% error when measured using Single Pulse or Continuous Pulse

© Vektrex | info@vektrex.com

Dominant Wavelength Measurement Indicates Change in Junction Temperature Unlikely to be Cause of Amber Phenomenon

•  Lumileds Amber o  Flux change/degree C: 1.6% o  Dominant wavelength shift/degree C: 0.123nm o  LED dominant wavelength DC: 594.99 o  LED dominant wavelength Single Pulse: 595.0

•  Cree Amber o  LED dominant wavelength DC: 589.55 o  LED dominant wavelength Single Pulse: 589.58

26

© Vektrex | info@vektrex.com

Additional Tests Showed Smaller Effect At Low Currents As Well

27

-1.20%

-1.00%

-0.80%

-0.60%

-0.40%

-0.20%

0.00%

0.20%

0.40%

Royal Blue 0.5 Royal Blue 0.05 Amber 0.5 Amber 0.05

Deviation From DC Measurement

Single Pulse Continuous Pulse

© Vektrex | info@vektrex.com

Additional research

•  Three additional Lumileds amber samples were tested with very similar results

•  Based upon the preliminary findings, Cree will supply additional samples

•  Additional testing will try to pinpoint the source of the amber error

•  Instrument Systems will also provide an updated spectrometer

28

© Vektrex | info@vektrex.com

Measurement recommendations

•  Assess the magnitude of phosphor errors before using continuous pulse for phosphor converted LEDs

•  Correlate continuous and single pulse measurements back to DC whenever possible using a high speed sampling voltmeter or digitizer

•  If using continuous pulse ensure current source rise/fall is <1/100 of pulse width

•  Until.

29