Practical Photometric Testing
of SSL Products
March 26, 2009
Chris Durell, VP Sales – Sphereoptics, LLC
1
Practical Lumens Photometric Testing
Integrating Spheres
•
OVERVIEW OF MEASUREMENT: Place lamp inside an integrating
sphere or at port of sphere and sample total light output of the DUT.
•
DERIVED VALUES: Lumens, Spectrum (W/nm), Chromaticity, CRI
– One Measurement for Single Lumens Value
– Easy to Monitor/Control Ambient Temp. and Light Conditions
ADVANTAGES:
– Self Contained, Fast Measurements, Easy to Use/Repeat, Robust
Measurement, Relatively Low Cost.
– Can be used at all stages of product development (raw LED to
finished product).
DISADVANTAGES:
– No quantization of spatial/angular distribution
•
•
2
•
Practical Lumens Photometric Testing
Integrating Spheres
Integrating Sphere
– 2pi – Directional Sources
– 4pi – Omni-Directional Sources
3
Source: IESNA LM-79
Internal Sphere Measurements (4pi)
1
SStandard
φStandard
3
Calibration Lamp
βStandard
βSample
Sample
SSample
4
2
Auxiliary Lamp
4 (λ )
S Sample
β S tan2 dard (λ )
φ Sample (λ ) =
* φ S tan1 dard (λ )
*
3
S S tan1dard (λ ) β Sample
(λ )
4
External Sphere Measurements (2pi)
1
SStandard
φStandard
3
Calibration Lamp
External or Internal
βStandard
2
βSample
Sample
SSample
4
Auxiliary Lamp
Plug for external port
4
S Sample
(λ )
β S tan2 dard (λ )
φ Sample (λ ) =
* φ S tan1 dard (λ )
*
3
S S tan1 dard (λ ) β Sample
(λ )
5
Variations on Sphere Designs
Customization for…
• Device/Port Size
• Orientation
• Testing Needs (thermal, etc)
6
Spectral vs. Photometric V(λ) Filters
•
•
•
•
•
LED Standards all recommend spectral measurements for absolute
metrology.
Spectral Measurement are more accurate and gives much more info:
– Lumens
– Color (x,y, u’, v’)
– Radiometric distribution (W/nm)
– CRI
– Monochromatic LED Values (Dominant Wavelength, Peak
Wavelength, Purity, FWHM, etc)
– Peak Wavelength can be used to help in determining Junction
Temperature (Tj)
Photometric Gives ONLY Lumens and has absolute accuracy issues.
Likewise, Three Filter or Four Filter Colorimeters should NOT be used
for LED or SSL Color measurements due to spectral mismatch errors.
Relative metrology (KGD) may be less restrictive (spectral not needed).
7
Spectral vs. V(λ) Filters
•
Go Spectral!
Deviation from CIE
Photopic V-Lambda
Curve of a real
photopic detector
3/26/2009
Corporate Overview
8
Practical Sphere Considerations
•
•
•
•
Device in the sphere should be <2% of the total surface area of the
sphere or longest dimension should be < 2/3 of diameter (per LM-79)
Sphere Coating reflectance should be as high as possible to ensure
good throughput and good integration.
Use a spectrometer of <5nm bandwidth for best results.
– Not all spectrometers are adequate – stray light and other
considerations can be problematic.
• Sphereoptics uses a NIST stray light reduction on our
spectrographs – only company in the industry to do this.
– Reference: Zong, et al, NewRAD, October 17-19, 2005,
Davos, Switzerland
External Standards for 2pi are being promoted by both IESNA and DOE
as best way to do calibrations for reflectorized sources (LM-20).
– Internal standards can be used for both 2pi and 4pi if you are careful.
9
Uncertainty of Sphere Based Measurements
•
•
•
•
•
•
NIST Uncertainty in Sphere-Based Measurements (Best on Earth)
– +/-1% (2σ) Lumens for White LED products
– +/-0.001 x and y (2σ) Chomaticity for White LED Products
Best Practical Secondary Laboratory Uncertainty
– +/-1.5-2% (2σ) Lumens for White LED products
– +/-0.002 x and y (2σ) Chomaticity for White LED Products
Commercial Laboratories
– +/-3% (2σ) Lumens for White LED products
– +/-0.003 x and y (2σ) Chomaticity for White LED Products
Good Industry Laboratories and QC
– +/-5% (2σ) Lumens for White LED products
– +/-0.005 x and y (2σ) Chomaticity for White LED Products
Practical Production Testing <10%
All above numbers are based on spectral measurements.
Source: Y. Ohno from LED 2008 Conference
10
Practical Lumens Photometric Testing
Goniometers
•
OVERVIEW OF MEASUREMENT: Light source is placed at the center of an angular
scanning device that pans a detector or mirror over the 2pi or 4pi steradians of light
output with a specified angular resolution
•
DERIVED VALUES: Candela/m2 (Lumens/m2-sr), Derive Candela (Lumens/sr), Lux
(Lumens/m2), ray tracing, Lumens. With Spectrometer: Spectrum (W/nm),
Chromaticity, CRI, monochromatic values by angle.
ADVANTAGES:
– Lots and lots of data – complete profile of source output.
– Required for Final Luminaire Testing and IESNA Specs (and DOE…)
DISADVANTAGES:
– Expensive Equipment (and testing services)
– Requires large dedicated room
– Long measurement period (spectral – even longer)
– Many different types (A,B,C) and some geometries may not be appropriate for your
measurement.
•
•
11
Practical Lumens Photometric Testing
Goniometer
Spectral Illuminance Head
Spectral Illuminance Head
Center of DUT
is Point of
Rotation for θ
and Φ
θ
DUT
Calibration Irradiance Lamp: W/m2-nm
Steradians derived from distance between lamp
and size of irradiance sensor
Φ
12
Practical Lumens Photometric Testing
Goniometer
NIST Goniometer – All Sources
Only
Goniometer – Flood or Directional
Hoffman AGS 1118
13
Sources: www.nist.gov, www.hoffmanengineering.com
Practical Lumens Photometric Testing
Goniometer – Auto Headlamp
Isocandela Plot – Taken on a Type A Gonio
14
Source: http://www.tfhrc.gov/safety/hsis/pubs/04148/character.htm
Practical Lumens Photometric Testing
Goniometer – Luminares
“Cobrahead” Outdoor Fixture
1. Downward Street Side
2. Downward House Side
15
Source: http://www.efficiencyfixturecorp.com/Portals/3/images/EFC_LID-CH-80w-120w-150w_CobraHead.pdf
Practical Goniometer Considerations
•
•
•
Measurement equipment for most luminaires runs minimum of $50k
and more typically in excess of $100k.
– Operating and maintaining the equipment requires a full time
employee.
– Only justified in high volume or R&D facilities.
Services from qualified labs are a good alternative.
Use manufacturer provided data on LEDs and LED modules.
16
Other Useful Optical Test Equipment
Luminance Meter
- Cd/m2, Color
Spectrometer
- Spectrum and Color
Illuminance Meter
- Lux, Ft-Cd, Color
Spectroradiometer
- Cd/m2, Color, Spectrum,
W/m2-sr-nm, Illuminance(?)
SSL Standards do not address use of these
instruments as proper uses are application specific.
Very easy to get wrong values if you don’t
understand proper set-up and use!
17
Electrical Considerations
•
•
•
•
•
•
AC Power: <0.2% (2σ) Regulation on Voltage or Current
– Calibration of AC power meter <0.5% (2σ)
DC power: <0.2% (2σ) Regulation on Voltage or Current
– Calibration of DC power meter <0.1% (2σ)
Measurement of Voltage and Current to provide Wall Power and Luminous
Efficacy (Lumens/Electrical Watts).
May require split sockets (Kelvin Sockets) and/or calibrated shunt for proper
current measurements.
Vf (Forward Voltage) of SSL lighting should be better than 1mV for single
high power devices to match new LM-XX high power LED standards coming
out. Vf needed to derive Junction Temperature (Tj)
PWM and AC-Driven LEDs treated as “Steady State” – For now…
Example: Keithley 2400
18
Thermal Considerations
•
•
•
•
•
Anyone who has designed LED products knows that thermal control is essential
to lumen output and product stability (Lifetime).
LM-79 and LM-80 make reference to controlling temperature on final products:
– Device Temperature – measured at Case Temperature (Ts)
– Ambient Temperature – measured at sphere or within 1m device (for
goniometer) and held constant at 25C+/-1C
At a minimum, suppliers have to measure Case Temperature of DUT after
device stabilization:
– Stabilization Defined in LM-79: “…at least 3 readings of light output and
electrical power over a period of 30 min. taken 15 minutes apart [with
stability] less than 0.5%.” This is part of the required report.
Ambient Air temperature specified is 25C, but actual environment of device may
not be 25C (ex: ASSIST in-ceiling temps of 30-40C). Need to control temp?
A more proactive approach is to control DUT temperature (and ambient?)
during product development testing and characterization
19
Sphereoptics
TOCS System
20
Sphereoptics
TOCS System
21
Thermal – Optical – Electical
•
- SLM Component
- TOCS Component
•
Complete
System can be
formed around a
sphere or
goniometer
Ambient Air
Temperature
Control also
Possible
22
Thermal – Optical - Electrical
Sphereoptics TOCS
(Thermal Optical
Characterization System)
Added or Retro-fit to any
Sphere System
23
Lifetime Testing (LM-80)
•
•
•
•
•
Test your product to L70=70% of Lumen Efficiency.
– Three different burn temperatures: 55C, 85C and user selected to
predict L70 success at 35,000hrs minimum.
– Hold temperatures of DUTs to +/-2C during life testing
– Hold Ambient Air to +/-5C during life testing
– Testing in <65% RH
– Thermocouple needs to be <+/-1.1C or +/-0.4% accuracy for Case
Temperature during test.
– On finished products (luminaires) only – subassemblies not
addressed.
Test minimum of 6000hrs (minimum increment of every 1000hrs).
Use LED manufacturer’s data?
Custom systems can be developed for life test.
Photometric Testing is always conducted at ambient temp (25C +/-2C)
24
New Standards Example: LM-XX
• High Power LEDs and
Devices.
• Thermal DUT control
essential to <0.1C.
• Accurate Vf
(<0.001mV) reading
required.
•Vicarious control &
measurement of Tj
• Requires pulsing and
DC (or steady state
operation)
25
Source: CORM 2008, Zong and Ohno, Rensselaer Polytechnic Institute (RPI), Troy, NY, June 9-12, 2008
Practical Thoughts
•
•
•
What do IESNA, DOE and CALiPER “really” want?
– Objective Standards (referee) to enable fast LED technology development
– Protect marketplace from inferior quality products
– DOE is placing burden on testing on the LED component, subsystem and
system manufacturers
• Final test is DOE Lab or other Certified Lab
• Everything before that is YOU – test early in your development
– Expect more test standards incorporating optical, electrical and thermal
integrated testing.
– Obtain NVLAP or DOE Lab Certification? …maybe not…
Rely on LED Manufacturer’s Data?
– Standards are new, most people are not ready or equipped to deal with
testing requirements
• Hard to force your suppliers to test for you – may not be relevant
• Do It Yourself?
Weigh the Costs of Testing Services vs. Costs of Testing Equipment and
iteration in your product cycle.
26
What Test Equipment Should LED Users have?
Testing: Incoming QC -> R&D -> Compliance -> Product QC
27
More Optical Training Opportunities…
•
•
•
•
OSW – Optical Sensing Workshop Lecture
– Radiometry / Photometry Units
– Integrating Sphere Applications
– Spectroscopy, Colorimetry & Spectrometer Applications
OSW – Optical Sensing Workshop Hands-On Testing
– LED Lumens, Color and Goniometry
– Colorimetry
– Reflectance / Transmittance
– Thin Film Analysis
Locations
– Tucson, AZ – May 11
– Albuquerque, NM – May 12
– Denver, CO – May 14
Contact: Justin Jacobs – 603-738-6910, [email protected]
28