«SOLID-STATE LIGHTING PRODUCT QUALITY INITIATIVE THIRD EDITION SEPTEMBER 2014 Next Generation Lighting Industry Alliance LED Systems Reliability ...»
4 Develop, implement, and adhere to design and engineering recommended best practices and document them in a form suitable for general lighting users.
Achieve industry consensus on screening tests which can identify product designs that are unlikely to meet vendor claims of reliability and lifetime (may be for specific product groups and may involve specific performance ranges).
Continue to improve and develop industry-consensus methods for measuring and reporting key LED product attributes, such as lumen and color maintenance, particularly efforts to accelerate such testing.
Recommendations for Buyers:
Focus on qualifying suppliers. Understand what methods the vendor is using to support reliability or lifetime claims, and require data. As they become available, require compliance with industry-consensus robustness tests.
Understand the warranty. What is covered and what is not? Is the warranty period a reasonable fraction of the claimed lifetime? Will the manufacturer have compatible replacement parts as applicable, when needed?
Avoid using products for which reliability claims are based on unreliable proxies for luminaire lifetime, such as the lumen maintenance of the LED package. Require and examine additional luminaire product or subsystem data from your qualified manufacturer to support any such claim.
2.1 COMBINING ABRUPT AND GRADUAL FAILURESIn earlier editions of this guide, “standard” or “default” lifetime of an LED luminaire (or lamp) was defined only in terms of lumen output and specified as the time when light output of half the product population has fallen below 70% of average initial light output for any reason (B50/L70). This definition thus encompasses gradual lumen depreciation of the LED sources, depreciation due to interaction with other components or materials in the luminaire, and catastrophic failure of any component or subsystem, ranging from total failure with no light output to the failure of a subset of the LEDs leading to luminous flux below a specified threshold. We continue to recommend this definition.
For a limited number of applications, excessive color shift may also be considered a failure. It is then natural to extend the definition of lifetime: when light output of half the product population has fallen below 70% of average initial light output or has shifted color beyond a specified limit (B50/L70/Czz). Czz depends on specific application needs; there is not a generally accepted level that can be applied to all products. There is also no accepted industry standard for projecting color shift past the test period. For most products, a color shift requirement is probably neither necessary nor appropriate.
For certain LED products in which the LED sources are individually visible (“direct view”), failure of a certain percentage of the LEDs, while not necessarily leading to total depreciation of 30%, may constitute an “aesthetic” failure, similar to color shift. This requirement is very design-specific, and again is not appropriate for the majority of products and should be left as a job specification where needed.
Finally, in some situations the specific limits proposed above may not suffice. These recommendations are not intended to exclude different job-specific requirements, but the same basic definition still applies. For example, the specification may be described as Bxx/LXX, Byy/CYY, Bzz/F for a specific case, where Bxx/LXX means xx% of the product is below XX% lumen depreciation, yy% of the product is below YY% of color shift, and zz% of the product has catastrophic failure (F). There may also be a specification that xx + yy + zz cannot exceed some limit, e.g., 50%. The following sections expand on the various types of failure.
2.1.1 ABRUPT FAILURE TO LIGHT
A critical part can fail to cause an LED luminaire to stop generating light altogether (catastrophic failure).
Examples include a power supply failure, corrosion of an electrical connection that stops the electrical flow to critical components, or breakage of a critical part due to vibrations or stresses beyond what the luminaire can handle. Such failure modes are not limited to LED systems. Much has been learned from reliability studies of other electronic systems that is directly relevant to this sort of failure, so it may be possible to apply this knowledge in developing a model of such failures in LED systems.
Mean time to failure (MTTF) can be determined for key components using established acceleration methods for parts other than the LED. Operating temperature is often a key item in calculating mean time to failure for electrical parts. For certain electronic parts, operational voltage and operational current can also impact MTTF and standard models have been proposed for calculating the impact of temperature, voltage, and current on the
For LED packages, the level of drive current applied to the LED as well as junction temperature are critical determinants and will affect the expected MTTF for those components. One abrupt failure mechanism at the LED package level is the loss of electrical connections through mechanisms such as a broken wire bond or an open solder joint. Other potential failure modes at the LED level include loss of emitter sites due to defect propagation, die cracking, and contact or silver mirror corrosion. All of these mechanisms are impacted by temperature, humidity and current. Finally, electrostatic discharge (ESD) and electrical overstress (EOS) can also produce abrupt failures at the LED package level through damage to the emitter layer or device interconnections. Manufacturers usually can provide extensive data on the performance and reliability of LED packages. They have made great progress in minimizing abrupt failures in LEDs, and field experience suggests this is not a significant factor in overall product performance provided good manufacturing practices are followed and the product is designed within the limits of the LED package. Examples of good manufacturing practices include proper thermal management of the LEDs to achieve controllable junction temperatures, avoiding the use of adhesives and materials which may outgas volatile organic compounds or corrosive materials (e.g., sulfur) that can damage LEDs, proper mechanical protection of the LEDs, and proper grounding practices during assembly.
The mean times to failure for components that are expected to have a significant contribution (“principal components”) can be combined to estimate the mean time to abrupt failure of the LED-based luminaire. If the product design is such that reliability is determined by a small number of principal components, and if established best practices learned from conventional lighting product manufacture are followed, it is possible to estimate overall product reliability with a reasonable confidence level. That is the fundamental basis for manufacturers’ proprietary methods of estimating reliability for specific products.
2.1.2 LUMEN MAINTENANCE AND LIMITATIONS
Absent abrupt failure, or as a contributor to combination failure, the light output of the LED luminaire will fade to the point that the end user concludes the luminaire has “failed” and has reached the end of its useful life.
Although a reduction in luminous flux by 30% seems to be an accepted value for many applications and standards, and is therefore the default, it is beyond the scope of this document to prescribe that as a universal standard. Rather, the intent is to assist in the understanding of how lumen depreciation fits into a more general definition of lifetime.
LED lumen depreciation information and data gathered from LM-80 testing by the manufacturer of the devices provide a baseline for luminaire system depreciation, which is typically greater than LED lumen depreciation alone. While there are efforts underway to develop standard measurement methods for entire luminaires, product size, sample size requirements, and the time of measurement will often conspire to make these tests impractical for other than the smallest examples, such as replacement lamps. While it may be possible to usefully accelerate such tests on full luminaires or lamps, such as by a significant increase in temperature, one 7 IEC International Standard 61709:2011, Electric components – Reliability – Reference conditions for failure rates and stress models for conversion, ed. 2.0, http://webstore.iec.ch/Webstore/webstore.nsf/ArtNum_PK/45258!opendocument&preview=1.
7 must take care to choose experimental conditions that avoid introducing new failure modes not relevant to normal operation. One alternative is to develop accelerated tests for subsystems and components that can be combined with source depreciation data to provide an estimate of luminaire reliability, thus reducing test time and expense. The work of the LSRC provides some guidance as to how such tests might be devised and also helps to establish best practices within the industry, but it is beyond the scope of this effort to develop specific test protocols or requirements.
2.1.3 HOW COLOR SHIFT FITS IN
Color change can be dissatisfying and the limiting factor of product lifetime in some cases. As color shift is monitored in LM-80 measurements, some information is available for any LED that has been tested according to this standard. LEDs are only one of the possible sources of color shift; plastic optical components or reflectors may be another. However, no prediction method for the evolution of chromaticity with time (analogous to TMfor lumen maintenance) is currently available.
Acceptable limits for color shift depend on the application. Users of a space lit by a single light source will be less sensitive to color shift than users of a space lit by two or more light sources that shift differently over time. Even LED products that have identical color shift behavior can lead to color shift complaints. Consider a space lit by lamps on two independent circuits: one circuit is operated for eight hours per day, and the other is on an emergency circuit that operates the lamps for 24 hours per day.
Over time, the color difference between the lamps on the two circuits will grow, and may lead to user dissatisfaction at some point.
There is no intrinsic reason that color cannot be quite stable, as demonstrated by the L Prize®-winning replacement lamp from Philips.8 But there are many factors that can contribute to color change including LED package design, LED operating temperature, methods and materials used to assemble an LED into a lamp or luminaire, the secondary lens material and its temperature, and the operating environment.
In addition to color shift over time, which is a reliability concern, there is reversible color shift with temperature and LED current (during dimming, for instance). These color shifts may be intrinsic, based on semiconductors physics. NEMA’s LSD 60-2012 standard, The Effects of Dimming on Color and Efficacy of LED Lamps,9 gives an overview of these color shifts.
Because of these complexities, we recommend that including color shift in an estimate of lifetime or specific limits on color shift not be applied for standard LED luminaire products for consumer or for general commercial use. Limits may be appropriate for products for specific applications that require color stability or for designerspecified products.
To a much greater extent than other lighting technologies, it is possible to classify products according to their reliability requirements. The design space is sufficiently broad that products with nominally the same appearance, such as replacement A-lamps, may have dramatically different lifetimes and reliability. The differences may be a result of operating temperature, LED drive current, materials and assembly, electronic component choices, or another aspect of the design. For example, a product can be designed to minimize cost, maximize efficacy, maximize light output, minimize color shift, operate in a harsh environment, or address any other performance aspect for a particular application and customer set. For this reason, the usual market segmentations of lighting products10 may have products within them that have different classes of reliability and lifetime performance requirements.
While it is beyond the scope of this document to define reliability requirements, it may be useful for the industry to move in the direction of defining reliability classes that may be addressed with similar protocols for testing and specifying reliability and lifetime, rather than requiring all products to undergo an extensive “one size fits all” testing regimen. For example, a reflector lamp product might be made offered in a residential model (“standard”), a commercial version (“long life”), and a retail display (“color stable”) model. Each would have a different testing protocol and different reliability specifications. A protocol for testing the “long life” product might then apply to other "long life" classified products.
Some examples of design objectives are:
Products Driven by Initial Cost (“Standard”) The primary objective is to provide reasonable performance at the lowest possible initial cost. There is less emphasis on lifetime and reliability, although products must meet customer expectations. Products would likely be found in residential or consumer market segments and may include replacement lamps for consumer use;
however, there may be examples in other segments as well. This class would have relatively short operating hours as compared with commercial or industrial applications. The environmental stress is generally low (clean power, controlled environment, etc.). As a consequence of the shorter lifetime objective, reliability testing may require shorter times and smaller sample populations in order to establish a reasonable performance.