Lens for uniform LED illumination: an example of automated optimization using Monte Carlo ray-tracing of an LED source

Abstract

We present preliminary results of optimizing a T1 LED integral dome lens for efficient and uniform illumination. The automated optimization process is based on a geometrical model of the LED die radiation inside the encapsulant. A commercial nonsequential ray-tracing program is combined with an external global optimization engine to maximize both efficiency and uniformity. The LED model is based on surface radiance images of an LED package sectioned with a diamond saw, and validated by comparison to measured near-field irradiance patterns. The resulting lens design is compared with the original design based on calculated irradiance patterns. The optimized lens substantially improves calculated uniformity, and somewhat improves efficiency, at the expense of a more complex lens profile, including multiple ripples on the surface. Robustness of the new design is evaluated by ray-tracing the lens with the LED displaced by typical manufacturing tolerances for low-cost LED's, resulting in significantly degraded uniformity. The results are still better than the original dome lens, but clearly indicate that manufacturing variabilities may place an upper bound on the uniformity achievable, at least with our preliminary approach.