A consortium convened by the US Department of Energy has defined and run an extreme accelerated life test on seven luminaire models, probing their modes of failure.
Ron Lenk wrote here last summer about highly accelerated life testing; he gave rules of thumb and a formula for estimating the operating lifetime of a unit under test. The testing he discussed involved high-temperature operation, which was intended mainly to simulate burn-in and to estimate luminaire lifetimes under normal operating conditions.
For a product consisting of highly reliable and long-life components, it is difficult to get an accurate estimate of its lifetime without knowing something about the failure modes of the system components and of the system as a whole. At present, not a lot is known in detail about how and why luminaires fail. (We have often discussed here circumstantial evidence of the driver as a potential weak spot.) This was the knowledge gap the LED Systems Reliability Consortium (LSRC) wanted to address. The consortium, convened under the auspices of the Department of Energy, includes lighting companies, research institutes, universities, national laboratories, and other interested parties.
The Hammer Test
The accelerated-life "Hammer Test" that LSRC researchers devised consisted of multiple "loops," each lasting 42 hours and including four stages:
Throughout each stage, power was cycled on and off at one-hour intervals. The entire test consisted of 40 loops for a total of 1,680 hours.
Seven luminaire models (not identified in the study) were tested -- one manufactured in 2009, one in 2011, and five in the first half of 2012. The luminaires were provided by different manufacturers or independently purchased. Six of the seven were 6" cylindrical downlights; the seventh was a 2' x 2' troffer. For each model, researchers subjected an average of three units to the Hammer Test. For four of the models, three control units were left on continuously for 296 days in the typical operating conditions represented by an office ceiling.
For the control units, color stability was good after the simulated 2.25 years of office operation: Δ'u'v' values ranged from 0.001 to 0.004. Lumen output for three of the control models degraded by only 1% or less; for the fourth model, lumen output degraded about 10%.
Twelve of the 17 luminaires subjected to the Hammer Test failed. The remaining five were still operating at the end. Failure was defined as either producing no light or falling below L70 after one of the loops. (Color shift was measured but was not used as a failure criterion.)
Of the 12 failed luminaires, 10 "exhibited catastrophic failure when examined after testing," according to the report. The cause of these catastrophic failures was "generally found to be associated either with the driver circuit, the LED board, or the connection between the two." The LEDs themselves were not found to be a significant source of failure. Only four of the 611 LEDs in the luminaires under test failed, having endured collectively nearly a million hours of cumulative exposure to the Hammer Test.
— Keith Dawson , Editor-in-Chief, All LED Lighting