An accelerated aging test is the main method in evaluation of the reliability of light-emitting diodes (LEDs), and the first goal of this study is to investigate how the junction temperature (Tj) of the LED varies during accelerated aging. The Tj measured by the forward voltage method shows an upward trend over the aging time, which gives a variation about 6°C–8°C after 3,000 h of aging under an ambient temperature of 80°C. The second goal is to investigate how the variation of Tj affects the lifetime estimation. It is verified that at a certain aging stage, as Tj increases, the normalized luminous flux linearly decreases with variation rate of microns () (1/°C). Then, we propose a method to modify the luminous flux degradation with the Tj and to meet the requirements of a constant degradation rate in the data fitting. The experimental results show that with the proposed method, the accelerated lifetimes of samples are bigger than that of the current method with increment values from 8.8% to 21.4% in this research.
Although the predicted lifetime of the classical 6000 h test given by Energy Star is taken as the normal lifetime of LED products in most research and applications, the aim of this study is to explore the error in lifetime prediction of LED lamps based on the 6000 h test. A non-accelerated aging test with 10 LED lamps is conducted for 20,000 h (from March 2016 to now) under room temperature, which is long enough for this kind of lamp reaching the real lifetime with the normalized luminous flux dropping to 70% naturally. At different aging periods, the correspondent lifetime of each sample is predicted by the lumen degradation, and the median lifetime of 10 samples is obtained by applying the Weibull distribution. Result shows that the of the real lifetime is 16,867 h in this work, and the aging time should be at least 9000 h to make the error in predicting the lifetime less than 3%. On the other hand, the values of 0.006, 0.007, and 0.008 are taken as the three thresholds for predicting the lifetime by color shift. For the case of 0.008, the calculated shape parameter of 8.4 in Weibull distribution is similar with that of the real lifetime, which means the of 0.008 for this kind of lamp gives the same failure mechanism as that of lumen degradation of 70%.
The lumen degradation of LED lamps undergoing an accelerated aging test is investigated. The entire LED lamp is divided into three subsystems, namely, driver, lampshade, and LED light source. The parameters of output power [Watts (W)], transmittance (%), and lumen flux (lm) are adopted in the analysis of the degradation of the driver, lampshade, and LED light source, respectively. Two groups of LED lamps are aged under the ambient temperatures of 25°C and 85°C, respectively, with the aging time of 2000 h. The lumen degradation of the lamps is from 3.8% to 4.9% for the group under a temperature of 25°C and from 10.6% to 12.7% for the group under a temperature of 85°C. The LED light source is the most aggressive part of the three subsystems, which accounts for 70.5% of the lumen degradation of the LED lamp on average. The lampshade is the second degradation source, which causes 21.5% of the total amount on average. The driver is the third degradation source, which causes 6.5% under 25°C and 2.8% under 85°C of the total amount on average.
Accelerated aging tests are the main method used in the evaluation of LED reliability, and can be performed in either online or offline modes. The goal of this study is to provide the difference between the two test modes. In the experiments, the sample is attached to different heat sinks to acquire the optical parameters under different junction temperatures of LEDs. By measuring the junction temperature in the aging process (), and the junction temperature in the testing process (), we achieve consistency with an online test of and and a difference with an offline test of and . Experimental results show that the degradation rate of the luminous flux rises as increases, which yields a difference of projected life of 8% to 13%. For color shifts over 5000 h of aging, the online test shows a larger variation of the distance from the Planckian locus, about 40% to 50% more than the normal test at an ambient temperature of 25°C.