«SOLID-STATE LIGHTING PRODUCT QUALITY INITIATIVE THIRD EDITION SEPTEMBER 2014 Next Generation Lighting Industry Alliance LED Systems Reliability ...»
Driver electronics can be an important reliability concern for LED luminaire systems, sometimes more important than the LED sources. Historically, in legacy lighting systems that were designed to only function two to three years at most before maintenance was required, power supplies and their components were specified to match this maintenance cycle. With today’s SSL light sources, with lumen maintenance periods that can effectively extend into many decades, a three-year power supply spec may not make sense. Highly reliable power supplies can be designed to match the system needs while simultaneously meeting cost objectives. Power supplies can also be separately tested,16 and much information is available from manufacturers, although generally not in a standard format.
Power management within a properly designed SSL luminaire ensures that the power delivered to the LED package(s) is appropriately sized, filtered, and controlled. A contemporary “driver” (power supply) contains multiple electrical circuits (input power conditioning, AC/DC converter, power factor correction, switching mode control, current regulation, output filtering, etc.) working together to provide regulated output power to LEDs. A failure in any of these circuits can have a cascade effect on driver performance and may ultimately result in a catastrophic failure. LED drivers consist of typical power supply electrical components including capacitors, inductors, power transistors, diodes, bridge rectifiers, resistors, and transformers, and any of these components may fail during operation. Since failure rates of electronic components can be related to their operational temperature, proper thermal management techniques are important to ensure driver longevity. In addition, driver designs that de-rate critical components such as capacitors have an added operational margin.
Hallmarks of a well-designed power supply system include:
Thermal management. A general characteristic of contemporary power supplies is that, to one extent or another, they generate heat. The well-known “10-degree rule” for electronic components states that for every 10 degrees (Celsius) of operating temperature reduction, you can effectively double the life of an electronic component. While this rule of thumb greatly over-simplifies a very complex interaction between operating temperatures of electronics and their expected MTTF, it is nevertheless a generally useful and applicable rule. To carry this philosophy further, a proper SSL luminaire design will also ensure that the overall thermal management design of the luminaire accounts for the power supply thermal load, and not just the LED-generated thermal load. Additionally, the luminaire manufacturer should always ensure that the temperature-range rating of the power supply is compliant with the expected operating environment of the luminaire.
Sub-component selection. Just as appropriate management of a power supply’s thermal load will improve its MTTF, of equal if not more importance is the careful specification and selection of individual components within that power supply. Power supply components, such as capacitors, diodes, rectifiers, and transistors, all have a critical functional role within a well-designed and well-manufactured power supply. As previously stated, each component’s operating parameters and rated MTTF should be carefully matched to the overall desired MTTF/MTBF (mean time between failures) of the luminaire to optimize the overall ROI of the luminaire. In general, carefully matched components will yield improved efficiency and reliability.
16 S. Tarashioon, W.D. van Driel, and G.Q. Zhang. “Multi-physics reliability simulation for solid state lighting drivers,” Microelectronics Reliability 54 (2014): 1212–1222, doi: 10.1016/j.microrel.2014.02.019.
15 Version control. Changes in one circuit within a power supply can have a cascade effect that ultimately produces failure in the power supply. Power supply performance is only as good as its weakest link in the environment that it is used. Great care should be exercised by luminaire manufacturers as regards the version control process employed by their power supply manufacturer; any and all engineering changes that are proposed by the power supply manufacturer should be tracked and made available for review, to ensure that any proposed change has been carefully examined, documented, and—ideally— tested. Improperly managed engineering changes, particularly when not thoroughly tested, can lead to a much higher than normal failure rate; when such failure rates happen in certain customer environments—e.g., on streets, bridges and tunnels—repair and/or replacement can be very costly, both in terms of actual repair costs, and in terms of diminished brand reputation.
Transient immunity/surge protection. Solid-state lighting products are subject to the same environmental conditions as their legacy-lighting counterparts: they must contend with power sources that in exterior and/or industrial-commercial installations can subject them to line voltages that may frequently vary widely against a utility’s “normal specification.” Sometimes these variations can be extreme, and while the duration of these voltage spikes can be very brief—often measured in milliseconds—they can nevertheless have detrimental consequences on the long-term reliability and expected life of the power supply. While many older generations of legacy lighting used “iron-core” ballasts that were more forgiving of such surges and spikes, today’s electronic power supplies can be much less forgiving. To address this issue, some luminaire manufacturers will employ surge devices of some type, varying from simple off-the-shelf sacrificial capacitors, one-to-one pass-through transformers, or bespoke designs with remotely accessible health status. Regardless of the design, care should be exercised when designing the luminaire’s power system to insure that it is capable of delivering its stated reliability within its intended installation environment.
Power factor. Power factor is generally defined as the ratio of the real power flowing to the load, to the apparent (metered) power in the circuit. A power supply that has an efficient power factor rating is widely recognized as a de facto requirement by most utilities and end users. Luminaire manufacturers must therefore ensure that this performance parameter be properly specified and met by the power supply vendor.
In summary, for proper, long-term reliable luminaire operation, the power supply and related electronics must provide a well-controlled and protected drive current and possibly other control and monitoring features, and must be designed to properly function for the anticipated life of the product. Component failures due to improperly designed and executed SSL power management will usually result in a catastrophic failure of the luminaire, and subsequent costly replacement for the end user or for the manufacturer, if the failure occurs within the warranty period.
While many luminaire manufacturers have achieved good results with power supply performance and have developed good relationships with their suppliers to procure the driver performance they need, for others the performance of the electronic subsystems remains a limitation to overall luminaire reliability. One consistent request from a number of luminaire makers has been to arrive at a more standardized reporting of driver performance, including reliability, in order to better address these issues. To appropriately minimize the probability of such failures, SSL luminaire manufacturers can and should work closely with their power supply
manufacturer(s) to define and agree on key reliability factors. Such factors would include clearly specifying:
16 Designed MTTF Availability of reliability test data Operating temperature and voltage range (so-called V,I mapping) Surge rating Formal revision-control process(es) and communication and approval flows HALT testing (completed or planned) Regular in-process quality testing
In-process quality testing should also be further defined and agreed, relative to:
Quality assurance (QA) team qualifications QA system(s) and methodology in use Frequency of ongoing production testing Pass/fail criteria and “non-conforming” response policy Availability of historical QA test data How and when will QA data be communicated Some progress has been made in developing new driver measurement standards. The ANSI American National Standard Lighting Group (ANSLG) assembled an ad hoc committee in 2013 to draft a measurement standard for LED drivers.17 At this point, this standard does not address long-term changes in drivers, although that effort has been contemplated. Such specifications and QA policies or processes outlined above may increase the unit cost of a given power supply to some degree, but such additional costs should be far less than the aggregate costs of unit failures in the field.
3.3 LED PACKAGE FAILURES
The LED package today is a very reliable component and is less likely to be the dominant cause of system failure in current products than when it was first introduced. The two main concerns related to LED packages are the gradual failures due to lumen depreciation and, where applicable, color shift.
3.3.1 LUMEN DEPRECIATION
The gradual diminution of light output from the source components will eventually limit useful life, however defined, absent other, earlier failures. For an LED luminaire, the time to reach output that is too low, typically 70% of the original luminous flux, will define the longest useful life of a luminaire under the same test conditions as those for the packaged LEDs. This has been called the “entitlement” for lifetime. All other failures will reduce this time to a greater or lesser extent depending on design, operating conditions, environment, and so on.
17 Jianzhong Jiao, “ANSI Recognizes Need for LED Driver Testing Standards,” LEDs Magazine, June 2014, http://www.ledsmagazine.com/articles/print/volume-11/issue-5/features/standards/ansi-recognizes-need-for-led-driver-testingstandards.html.
17 Quite a bit of work has been done to define test methods for package lumen depreciation (LM-80), including the projection of a limited test time to longer periods (TM-21). Forthcoming standards will provide methods of measurement and projection for lamps or light engine components (LM-8418 and TM-2819), although for practical reasons (product size, sample size, test time, cost) these may be more useful for limited verification of predictions made through other means rather than for routine product life tests. We refer to the standards themselves for more details on the measurement and characterization of lumen depreciation. We also note that LED package lumen depreciation is not a proxy for luminaire life because of the presence of other failure modes.
3.3.2 LED COLOR SHIFT
Defining color shift is itself a complicated issue, and color shift remains a difficult parameter to include in a lifetime definition. The Standard Deviation of Color Matching (SDCM), also called the MacAdam Ellipse, roughly approximates the ability of the eye to distinguish color differences. In the CIE 1976 uʹvʹ color space, one SDCM is represented by an approximate circle with a radius of about 0.001 Δuʹvʹ. Such a definition, however, says nothing about the direction of color shift. Relative shifts of LEDs in the same direction are less obvious to the observer than are shifts in opposite directions, for example. And shifts in different directions may be more or less evident or distasteful to the observer as well. Measurement over time is difficult, and there is as yet no agreed method to extrapolate results to longer times.
As noted earlier, color shift is not applicable for all applications or for all products in a particular market segment. Therefore it should be considered as an “add-on” specification only for products sold as “color stable.” Additionally, for a general statement of color stability, a failure should be described as a shift outside some limit, regardless of the direction of shift. However, for a “specified product” a buyer may add additional requirements.
To expand further: The color point of a light source is generally measured using chromaticity coordinates that are defined by the Commission Internationale de l’Eclairage (CIE). While the CIE has developed several different formats for chromaticity coordinates and the corresponding color spaces, the one most commonly used by LED manufacturers is the CIELUV color space adopted in 1976,20 because it provides for a nearly uniform chromaticity scale. In the CIELUV color space, the chromaticity coordinates are designated as u’ and v’. Hence, the change in color of a light source as it ages can be calculated from its current chromaticity coordinates (i.e.,
u’,v’) and its initial chromaticity coordinates (i.e., u0’, v0’) using this formula:
Δu’v’ = [( ) ( )]
As the primary light source, the characteristics of the LED are important in understanding the potential for color shift of the lighting system. LM-80 data published by LED manufacturers provides insights into the color shift characteristics of a specific product under the test conditions used in LM-80 (i.e., constant current, controlled temperature and low humidity). These color shift values are usually given in terms of Δu’v’, although sometimes the individual chromaticity coordinates (u’,v’) are given as well. Note that values of Δu’v’ convey the magnitude of the color shift, but do not give the direction of the shift.
18 IES LM-84-14, Approved Method for Measuring Luminous Flux and Color Maintenance of LED Lamps, Light Engines, and Luminaires, http://www.ies.org/store/product/ies-approved-method-for-measuring-luminous-flux-and-color-maintenance-of-led-lamps-lightengines-and-luminaires-1339.cfm.
19 IES TM-28-14, Projecting Long-Term Luminous Flux Maintenance of LED Lamps and Luminaires, http://www.ies.org/store/product/projecting-longterm-luminous-flux-maintenance-of-led-lamps-and-luminaires-1348.cfm.
20 N. Ohta and A.R. Robertson. Colorimetry: Fundamentals and Applications. New York: Wiley, 2005.