LED photometry presents a different set of challenges from that of other sourcesdevices may be small or large, or highly directional, and can exhibit thermal effects that affect their performance and spectral output. Accurate measurement requires careful technique.
Integrating spheres provide the most efficient method for measuring the flux of devices and device arrays, but are not without challenges. Integrating spheres have been typically used to measure more Lambertian (diffuse) sources than highly directional LEDs. Careful sphere design, particularly in size and position of baffles, along with positioning of the detector and the LED, alleviates most of these problems. It is imperative that the detector not capture LED output directly, or even capture "first bounce" light. Appropriately placed baffles ensure that the light is integrated by many bounces before it reaches the detector. An alternative is using a cosine-corrected detector, either directly at the sphere surface or coupled to the detector with an optical fiber.
To ensure accurate measurements, use baffles to protect the detector from direct illumination by the device under test.
Many applications use only the forward output of a lensed LED. Evaluating such sources requires positioning the LED such that only the desired angular output enters the sphere. With standardized geometry these partial flux measurements represent an intensity measurement. A potentially more accurate technique, albeit slower and more expensive, involves characterizing the LED with a goniophotometer and integrating the flux over the solid angle of interest.
Inserting an LED into a port at the side of the integrating sphere is a common method for capturing the power spectral distribution from the device; it requires exact positioning of the device at the sphere wall for repeatable results. This technique does not weight the back radiation of the LED uniformly, as compared to mounting the LED at the center of the sphere. For that reason, side mounting the LED is sometimes referred to as 2 pi (steradian) flux, and center-mounted measurements are referred to as 4 pi flux to distinguish the two geometries.
LED arrays, with their expanded source size, require a larger sphere for accurate measurements. Placing the source at the center of the sphere improves measurement reproducibility (see figure). An effective design combines the sample holder and the array assembly with a large baffle to shield the detector from first-bounce radiation. For large test samples it is best to add an auxiliary lamp to calibrate the self-absorbance effects of the large baffle/sample assembly. Light in the integrating sphere undergoes hundreds of collisions with the sphere wall, baffle, and sample before being absorbed or collected at the measurement port. Interactions of the light at the sample-specific baffle and the test article will, in general, reduce the measured flux and can change the spectral distribution; lack of self-absorbance correction can introduce errors of 10 to 20% for some lamp assemblies.
LED color and power can change until the device achieves thermal equilibrium, affecting measurements. Passive or active temperature stabilization is one approach; another is to extrapolate performance from a low-power data point. An LED operating at 1 kHz pulse modulation with a 1% duty cycle experiences little heating, allowing manufacturers to capture a data point and then multiply the power by 100 to obtain the effective output of the device when operated with the diode junction temperature at ambient. The results for 1% duty cycle are pretty much invariant upon the heat-sink used and can be repeated by other labs, but the approach may not be representative for the performance at the maximum-rated DC current values.
LEDs are normally tested at a set current value. However, the compliance voltage depends on temperature and aging of the device. For that reason, it is common practice to use four-wire measurements and log the set current and voltage compliance of the device under test.
As LEDs and LED arrays become more prevalent in illumination and communications, precise metrology of these devices will become increasingly important. While international organizations such as CIE and ISO continue to discuss standardizing test methods, good metrological methodology will allow laboratories to supply consistent and accurate device measurements. oe
Art Springsteen, Will Weber
Art Springsteen is president and Will Weber is CTO of Avian Technologies LLC, Wilmington, OH and Wallkill, NY.
Tim Moggridge is president of Lumetrix Ltd. and Instrument Systems Canada, Ottawa, Ontario.