Measuring LED and Solid State Lighting Performance
- Created on Friday, 10 July 2009
Why the large recent increase in attention to LEDs and solid state lighting? The cost of energy and the amount of carbon emissions is one answer. Lighting consumes 22% of the electricity produced in the US and 8% of the energy. The impact of SSL and other alternative lighting sources by 2025 is projected to be a 50% decrease of energy consumption by lighting and a 10% reduction in carbon emissions.1 The Department of Energy and many manufacturers are very interested in increasing the efficiency conversion of electrical power to light output. This is the key performance factor that, depending on the application, determines the level of energy-efficiency.
For applications where LEDs are used as sources of light that are viewed directly, the term luminance intensity is used (i.e. units of lumen/steradian, also known as candela). For example, in a traffic intersection signal light, the important parameter is the light emitted into a solid cone angle. For applications where LEDs are used as general illumination, the total light emitted in all directions, the total luminous flux (units of lumen), is the metric of interest. The efficiency of the conversion of electrical power into light has a specific term, luminous efficacy, and a defined relationship of measurable parameters: 1) optical power in watts, 2) electrical power in watts, and 3)luminous flux in lumens.2
New developments in the detectorbased standards have improved the ability to determine efficiency and luminous efficacy by lowering the uncertainty of easy-to-use working standards.3 These detector-based standards consist of a high-precision current-to-voltage amplifier, temperature-stabilized, silicon photodiode and temperature- stabilized spectral filter that makes the filter-detector combination very closely match the CIE 1931 standard photometric observer function. The new working standard improved the performance of components contributing to the total measurement uncertainty by factors of between 2 and 5; this has led to a reduction of total uncertainty level in the candela by a factor of 3. These detector-based standards can be used to directly determine the average LED intensity of sources using the CIE publication 127 recommended geometries.4
New spectroradiometers using backside- thinned CCD detectors provide the basis for further improvement in the uncertainty in the determination of average LED intensity and luminous flux. In the case of average LED intensity, the spectral data from a well- characterized instrument can be used to provide spectral correction factors generated from the known absolute spectral response of the detector-based luminous intensity standard and the measured relative spectral power distribution of the LED. The characterization of the spectroradiometer's performance parameters is essential for this spectral correction to be valid. It is important to characterize certain aspects of the spectroradiometer performance. Input geometry, wavelength scale assignment to the detector pixel locations, spectral band pass and scanning interval, random noise, stray light, and detector non-linearity all contribute to errors in the measurement. Once these error factors have been characterized and corrected, the uncertainty in the spectral measurement results can be combined with the detector-based standard uncertainty to give an overall uncertainty for the measurements of LEDs.
The spectroradiometer is also required to provide the color performance characteristics of the LED or SSL component and is required in the Illumination Engineering Society of North America (IESNA) LM-79 Approved Method for Electrical and Photometric Measurement of Solid State Lighting Products and the ANSI C78.377A draft of chromaticity specifications for SSL products. Spectroradio - metric data allows computation of tristimulus values, the fundamental quantities for colorimetric computation. The tristimulus values are then used to compute x and y 1931 CIE chromaticity coordinates and uâ€™ and vâ€™ CIE 1976 uniform color space coordinates. The chromaticity coordinates can then be used to compute dominant wavelength for narrow band quasi-monochromatic LEDs. For â€œwhite lightâ€ LEDs, the CIE 1931 coordinates are used to determine correlated color temperature, color rendering index (CRI). Also for white LEDs the ANSI C78.377A draft defines the quantity Duv as the closest distance from the Plankian locus on the (uâ€™, 2/3vâ€™) diagram with a + sign for above and a âˆ’ sign for below the Plankian locus.
The computation of CIE color rendering index, CRI, requires special mention. The existing method of computing color rendering was developed as a metric to help gauge the color rendering properties for fluorescent lighting products. Its application to white light LED sources created by combining three narrow band LEDs, blue, green and red, does not correspond well to the perception of color of objects when viewed with this illuminant. One solution that is actively being worked on is the development of a color quality scale (CQS), an expanded color quality metric to meet the needs of both the lighting industry and users.5
For determination of the variation of luminance intensity and color of LEDs and SSL products, some type of mechanical movement of the photometric or spectroradiometric sensor is required to map the output as a function of angle. One way is to use a traditional computer-controlled goniometer with the light source mounted to a rotation stage and the sensor mounted to a pivot arm with its center of rotation coincident with the SSL reference emitting surface. Alternatively, individual LEDs can be mounted on a computer interface controlled stage that both pivots and rotates the LED relative to a fixed sensor collection position configured for either CIE127 condition A or condition B average LED intensity measurement positions. The resolution, precision, and repeatability of both of these angular motions are important in maintaining low uncertainty in the measurement results. Another method employed by some National Measurement Institutes and commercial laboratories utilizes 6 axis industrial robots to move the sensors to the various angular positions to map the intensity and color variations. The robots employed have spatial position placement and repeatability on the order of 0.050mm over a 1.5 meter or greater reach radius. The advantage of using a robot over a traditional goniometer is the flexibility it affords to make measurements on other types of sources such as LED back-lit flat panel displays.
It should be noted that the best measurement equipment cannot give you the lowest uncertainty results without well trained and knowledgeable staff as operators. Training is a constant challenge. The continued availability of resources such as the National Institute of Standards and Technology (NIST) photometry short course, covering optical radiation measurement theory and practice, directly relates to the quantities required for LED and SSL evaluation.
- McCabe, M., DOE Director, Office of Building Technologies 2nd Annual DOE SSL program planning workshop Feb 3 &4, 2005
- Miller, C. Fundamentals of Photometry and Solid State Lighting, 2007 Council of Optical Radiation Measurement 2007 SSL Measurement Workshop
- Eppeldauer, G.P., Austin R.L. and Lustenberger, C.R. 3. New Working Standards to Disseminate NIST Radiometric and Photometric Scales, in NCSL Measure Vol. 2 No. 8 (2007), p.40
- CIE Publication 127: 2007
- Davis, W. and Ohno, Y. Toward an improved color rendering metric, in 5th Proc. International Conference on Solid State Lighting, ed. by I.T. Ferguson, J.C. Carrano, Tsunemasa Taguchi, and I.E. Ashdown, SPIE 5941, (2005), p. 59411G.