Information on LED lifetimes and factors
LEDs are long-life light sources which generally do not suddenly fail, but gradually lose their brightness. Their luminous flux – measured in lumen – becomes smaller. This is known as degradation. The rate at which an LED loses brightness depends on the operating conditions and also on external factors such as temperature, relative humidity and changes in the thermal load. And how exactly can lifetime be predicted?
In contrast to conventional light sources, an LED still emits light at the end of its life, just not enough. In order to predict the life of an LED a failure criterion must first be defined. L70/B50 is the criterion most often used. L stands for the relative brightness level at end of life (70 percent) and B stands for the relative number of LEDs (50 percent) whose brightness at this point in time is less than 70 percent.
The industry standard of general lighting for determining the life of LEDs consists of test specification IES LM-80 and the calculation method based on IES guideline TM-21. LM-80 prescribes how the data has to be recorded. The test must run for at least 6000 hours at two different package temperatures. TM-21 describes the calculation method for arriving at lifetime forecasts from the LM-80 test results. Extrapolation is limited to six times the duration of the LM-80 test.
In addition to high temperatures, high currents and relative humidity will also speed up the degradation process. To obtain a more precise lifetime forecast, manufacturers such as OSRAM Opto Semiconductors operate LEDs not only at different temperatures but also at different currents and different humidity levels, and measure the brightness at regular intervals. At first glance it appears that the drop in brightness is the same for all LEDs. The reduction can be defined by mathematical functions which extrapolate the fall in brightness into the future. It is not immediately obvious, however, which function is the most accurate. To find the most suitable function, we can use a mathematical trick, namely the smallest quadratic deviation method.