«SOLID-STATE LIGHTING PRODUCT QUALITY INITIATIVE THIRD EDITION SEPTEMBER 2014 Next Generation Lighting Industry Alliance LED Systems Reliability ...»
High Efficiency, Long Lumen Life, and Low Maintenance (“Long Life”) Products Energy efficiency is a significant factor because of longer operating hours. Outdoor and industrial products may have this objective. Environmental requirements are more severe, especially including humidity or extreme temperature. The products are expected to be in a place for a long time with minimum attention, especially for industrial applications in hard to reach places, roadway lighting, and so forth. Higher costs of reliability testing could be justified in part by lower maintenance costs, for example, but there may be more tolerance for greater color shift and the definition of what constitutes lumen output failure. Longer test times with larger samples would be required. In some cases additional stressors may be appropriate.
“Color Stable” Products Modest shifts in color may qualify as a failure for these products; beam quality and stability is an issue. Such “designer” products are found in retail, hospitality, or other public spaces, where good lumen maintenance and
As noted, these reliability categories may cross market segment lines which are highly varied. However, it may be reasonable to establish a more limited set of standard pass/fail reliability protocols or robustness tests along these lines.
2.2.1 SERVICEABILITY Another issue that must be addressed with respect to LED-based luminaire lifetime is that some LED luminaires have been designed to be serviceable and some have not.
Non-serviceable LED-based luminaires fail similarly to traditional incandescent light bulbs. If one part fails (for example, the filament or the outer glass bulb breaks), the entire unit no longer works. This is the case with non-serviceable LED-based luminaires. If a critical part fails or the light output falls below the needed light output, the entire luminaire has reached its end of life.
Serviceable LED-based luminaires are more similar to fluorescent luminaires used today. If a fluorescent lamp fails, it is replaced and the luminaire becomes operable again. Similarly, if an LED power supply or LED array fails and it can be replaced, the luminaire becomes operable again. Therefore, the lifetime of a serviceable LED-based luminaire is when a major mechanical or optical part fails that is not serviceable, or the time when replacement parts are no longer available, or the time when luminaires that are more energy efficient or have additional features and benefits can be economically justified to replace the current one.
Serviceability does not preclude the use of reliability classes. The replaceable components of a serviceable product can still be classified according to some scheme such as that above for purposes of defining required testing and the results combined with other components to provide an overall performance estimate of the luminaire product.
A warranty may be written for the life of a serviceable product that assumes some servicing paid by the user, or possibly for the life until service. It is prudent for the buyer to understand what the warranty will cover as the costs could be quite variable.
In order to accurately characterize the reliability performance of any product, it is important to identify and understand those failure modes that materially affect it. In the case of LED lighting products, we are generally familiar with the lumen depreciation of LED packages which will eventually result in the light no longer being useful. We are also aware through experience with traditional lighting as well as LED lighting that another gradual change, color shift, may provide a limit to lifetime as well.
The LSRC has reviewed studies intended to identify potential failure modes and provide additional understanding of product life. We reviewed the results of some highly accelerated multi-variant tests and other available data to learn which failures may be significant and how those failures might be accelerated. Some of the information on failure modes comes from a series of highly accelerated tests executed by RTI International with the help of DOE funding on a limited number of product examples.11 Other information can be derived from the DOE testing associated with the Philips L Prize-winning LED A-lamp.12 Systematic field data is of very limited availability (and tilted towards reported failures) but does provide some additional insight into those areas that should receive some attention. Still further non-public information comes from the experience of members of the LSRC and informs the discussions about important failure mechanisms. Members were asked which failure modes they most frequently observed; the results are summarized in Figure 2.
Figure 2. The most commonly observed failures from LSRC member survey.
“Times Referenced” means the number of respondents who cited this failure mode.
Accelerated testing is an engineering tool that uses carefully selected environmental conditions to speed up the aging and/or degradation processes associated with components in an SSL device. Among the conditions (i.e., stress levels or stressors) that may be varied in a properly designed accelerated test are temperature (including both high and low temperature extremes), humidity, vibration, electromagnetic irradiation, outdoor UV exposure, chemical exposure, electrical power, and ripple.
Most SSL products are designed to operate under specified ranges for temperature, humidity, electrical power, and other parameters. These products will still function outside these design ranges, but their lifetimes will typically be shorter than under normal operating conditions. For example, LM-80 test data for LEDs has demonstrated that operation at high temperatures (e.g., 125oC), for a given current level will result in faster lumen depreciation than operation at a lower temperature (e.g., 55oC). This is a natural consequence of the chemical kinetics of the degradation of LED components, and since higher temperatures will increase the rate of the processes responsible for degradation, the performance of a device decreases faster under these conditions.
The primary advantage of performing accelerated testing at high stressor levels is that the aging process occurs at a faster rate, so test time is reduced. Properly designed accelerated test methods seek to build a correlation between lifetime under elevated stress levels and normal operational levels. As a result, accelerated testing will typically be performed until failure of the device(s) under test (“DUT”). When failure occurs, an examination of the failed part is performed and the determination of the failure mechanism is critical to understanding the test results and to building scientifically sound correlations to normal operating conditions. Accelerated test methods should also recreate failure mechanisms that are observed under normal operation and not create new failure modes for the test to have the greatest meaning.
There are many different types of accelerated test methods that can be applied, and these methods can be roughly grouped by the number and level of stresses applied to the DUT. In some cases it may be appropriate to
apply several of these stresses in succession. Examples of typical accelerated test methods include:
Constant environmental accelerated tests. In this test procedure, a constant environmental stress is applied to the DUT. An example of this type of test is a temperature bake at a value higher than the normal operational value.
Cycling environmental accelerated tests. In this test procedure, the environmental stress is cycled between two or more levels (usually a high state and a low state). Two common examples are temperature shock, which involves rapid exposure of the DUT to two temperature extremes (e.g., -50oC and 125oC) with sufficient time for equilibration at each endpoint, and temperature cycling, which involves a more gradual cycling of the DUT between the temperature extremes.
Multiple environment stress tests. In this test procedure, two or more environmental stress levels are controlled and the DUTs are subjected to their combined influences. A common example is temperature and humidity testing often performed at 85oC and 85% relative humidity (85/85).
Step-stress tests. In this test procedure, one or more environmental stress levels are held constant for a period time and then increased by a set amount. The process is repeated until failure of the DUT occurs.
Highly accelerated life tests (HALT). In this test procedure, environmental stress levels well beyond those expected during normal operation are applied to the DUT to promote failure in a short period of time. In many instances HALT methods will use multiple environmental stressors to further reduce test time. The use of step-stress and HALT methods in testing SSL devices is explored further in a 2013 book edited by LSRC members Dr. Willem van Driel and Dr. Xuejun Fan.13 For tests used to verify product performance, an important consideration in designing accelerated tests is that the failure modes produced under accelerating conditions should mimic those occurring in normal operation.
This consideration often places a limit on environmental stress levels that can be used in accelerated testing.
Extreme environmental conditions, such as highly elevated temperatures or vibration levels that may occur in step-stress or HALT methods, can introduce new failure modes. These tests, sometimes tests to failure, can be useful for identifying potential failure modes in operation, but proper evaluation of these new modes is critical to understanding how or if the test results apply to the product.
3.1.1 EXPERIMENTS ON ACCELERATING FAILURES
In the context of the U.S. Department of Energy’s L Prize competition to design a 60W A-lamp equivalent, Pacific Northwest National Laboratory (PNNL) extensively tested the Philips submission entry, including long-term operational performance. As of July 2013, 200 sample lamps had been continuously tested at a 45oC ambient temperature for over 25,000 hours. This is perhaps the largest publicly available data set on LED product reliability, and it is significant that there have been no failures, lumen depreciation is negligible, and average chromaticity shift is less than.002. These tests continue. Figure 3, from PNNL’s lumen maintenance testing report,14 summarizes the results. While this product is a good example of the ability of LED lighting to perform consistently over a long period of time, the data also show that the moderately elevated temperature provides little if any acceleration of lumen depreciation, illustrating the challenge of reducing the test times for such products. Other means will need to be explored to achieve that goal.
Figure 3. Comparison of best, worst, 13th-worst, and average L Prize lumen maintenance after 25,000 hours.
Source: PNNL 13 W.D. van Driel and X.J. Fan, eds., Solid State Lighting Reliability: Components to Systems. Springer, 2013.
14 U.S. Department of Energy, Lumen Maintenance Testing of the Philips 60-Watt Replacement Lamp L Prize Entry.
13 RTI International, in association with the LSRC, tested commercially available, mass-produced indoor luminaires using a HALT also known as the “Hammer Test.”15 The intent of the Hammer Test was not to develop a new robustness test, but rather to accelerate luminaire failure to less than 1,000 hours by subjecting them to environments outside their design range. Once failure is induced, subsequent tests will be needed to further investigate failure modes and determine acceleration factors.
In the Hammer Test, 6" LED downlights were subjected to a series of sequential environmental stresses including temperature cycling (-50oC to 125oC), wet high temperature operational life test (WHTOL) at 85oC and 85% relative humidity (RH), and high temperature operational life test (HTOL) at 120oC. In addition to these multiple environmental stressors, electrical power to the luminaires was cycled on and off at one-hour intervals to provide electrical stress as well. The study found that such SSL luminaires can exhibit exceptional durability even under the extreme stresses of the Hammer Test. All luminaires examined in the study survived more than 100 cycles of temperature shock (-50°C to 125°C) and nearly half survived more than 300 cycles. The failures that were observed typically occurred in the driver circuit, with board-level failures being most common. The 611 LEDs contained in these luminaires endured nearly one million hours of cumulative exposure to the Hammer Test, and only four failures (1%) were observed during the testing. These findings reinforce the high reliability of LEDs in lighting systems, even under extreme conditions, and suggest that other elements of the luminaire are more likely to fail first. The level of performance demonstrated by the luminaires examined in this Hammer Test protocol suggests that SSL luminaires will have a low probability of random failure in the field during normal use.
Tests such as the Hammer Test and the step-stress testing on L Prize lamps have demonstrated that welldesigned SSL devices are exceptionally robust and can operate for substantial periods of time well outside their
specified operational environments. Key findings include:
Catastrophic failure of LED package failures continues to be rare in testing.
Lumen depreciation of LED sources operated under proper thermal control and moderate current drive is low, and the device can operate for extended periods without failure or significant lumen depreciation.
Yellowing of some polymeric optical components may occur and contribute to lumen depreciation and color shift.
Driver failure is likely to be an eventual cause of catastrophic or abrupt failure in SSL devices, but longlived drivers are available.
Failure of electrolytic or film capacitors is a leading cause of driver failure, but de-rating of electrolytic capacitors will extend product life.
Impedance increases in electrical components (especially capacitors) with aging and can also cause an increase in power dissipation in the driver and a reduction in power factor.