Considerations in Building an SSL Specification Boiler Plate

Posted: November 16, 2014 in Uncategorized

Solid-state lighting presents many opportunities to create fidelity in specification of lighting not practical or available before. Further, the blend of aggressive marketing, hype, and deception on the part of solid-state product manufacturers demands more diligent specification than ever before. The days of the conventional mode of specifying luminaires and lamps as two separate components, with experiential trust in both, is over. Today, luminaires and light sources are integral to one another, often offered up by those who have marginal experience, and a strong desire to realize sales. The only defense against predatory and overly aggressive marketing is to understand and develop a strong specification foundation. The next protective barrier is to hold that specification. The following are suggestions for building a solid-state lighting foundation for any luminaire specification, with rationalization for each consideration in practical terms.

Be Aware: CCT from raw LED data/output frequently does not match luminaire CCT output
When specifying CCT for products, it is essential to demand that the stated CCT is luminaire output, not LED source output. The presence of lenses and diffusers will shift CCT of light sources within a luminaire by as much as 300CCT. Most white diffuser materials and diffuse optical reflective surfaces (powder coat, white optic films, etc) with cause a warming of CCT from the light source by 200 CCT at 3,500K, and as much as 450CCT at 5,000K. This means that a luminaire utilizing 4,000K LEDs, with a white diffuser, will generally deliver an output of 3,700 to 3,800CCT. This has an equally significant impact on S/P (Scotopic/Photopic) ratio, as the spectral power of the luminaire s changed from the source employed. The only determinant of actual delivered CCT from a luminaire is a proper LM-79 photometric test showing luminaire chromaticity output in CCT, CRI, and S/P ratio. Do not trust any manufacturer data founded on LED source data alone in any specification or submittal evaluation, as the data is essentially irrelevant. Don’t be surprised if in your quest for a 4000K LED luminaire, the best you will actually find is closer to 3800K. To create a specification that can be met will require inclusion of this, based on preliminary evaluation of products you intend to use. This may be problematic if products with different optical configurations generate different CCT values from the same core LED components. Unfortunately, not a great deal can be done about this, beyond being aware of it and including it in evaluation of products being specified.

When task performance is not a consideration, CCT is an acceptable minimum
Whether it is a residential space, a hospitality environment, or a general circulation area, qualified by spaces where general illumination levels are below 20Fc, selection of a CCT based on visual appearance or aesthetic preference is an acceptable choice. Whether the choice is of a low CCT source to create a warm ambiance, such as a restaurant, or home, or a high CCT choice to produce a cooler presence, care must be taken to include observed and documented preference of observers for warmer CCTs as illuminance levels are reduced. Further, it is likely that as illuminance levels are lowered and CCTs are warmed, the need for consideration of higher CRI/CQS color performance values must be considered to avoid undesirable results (see Color Performance below).

Where tasks and visual performance work is performed CCT +  S/P ratio specification is Required
There is a significant body of scientific, objective data, coupled with subject survey data to recommend that in productive environments, where task performance is the primary function of the lighting system, high S/P (Scotopic/Photopic) ratio light sources produce superior visual performance, and is preferred over sources with lower values. As a general rule, high CCT light sources, 4000K and above, out perform lower CCT sources. For details on specific concept, IES TM24-13 provides very clear guidance and recommendation, as well as the impact on illuminance requirements related. Based on this, any specification of CCT for task environments and commercial working spaces must include the associated specification of S/P ratio for the source. While there has not been a direct study that aligns illuminance levels and visual task demands with specific CCT or S/P ratio minimums, I personally recommend the following be considered as a viable approach, and is what I recommend to customers of solid-state lighting products:

  • For low level commerical space illuminance, ambient lighting in traffic or circulation areas, a CCT of >3,500K with an S/P ratio of 1.40 should be considered minimum
  • In moderate task environments general office lighting, and ambient lighting where supplemental task lighting is provided, or where visual tasks are primarily accomplished on illuminated displays, a CCT of between 4,000K and 5,000K with luminaires generating an S/P ratio of no less than 1.60.
  • For task level light sources, whether within a work space with ambient light, or as primary illuminance up to 100Fc, a CCT of >4,000K with an S/P ratio of no less than 1.60.
  • For task level light sources, where illuminance levels exceed 100Fc, a CCT of 5,000K or greater, with an S/P ratio of no less than 1.85. See IES TM-24-13 for more on this specifically.

Where high illuminance and color performance are both critical, CCT + S/P ratio + CRI/CQS specification is required
When illuminance levels are high, coupled to a demand or need for high color performance, the additional requirement of CRI or CQS is added to the above recommendations for CCT and S/P ratio inclusion. However, this combination raises some difficulty in solid-state lighting source selection, as high CRI products in general deliver poorer CRI results, making selection of higher CCT products a matter of finding an acceptable balance between CCT and color performance.

Color accuracy performance specification
While there is a significant debate raging about the viability of CRI as it applies to solid-state ligthing, to date there has been no significantly better alternative proposed to replace it. CQS, a well known proposal for replacement, has not been shown to represent a significant enough difference to establish itself as a superior standard. In fact, under sources other than CRI, CQS has proven to be as problematic and erroneous as CRI is under some forms of solid-state lighting. The extension of CRI to 15 standard colors, R9-14, represented as either CRIn or CRIe, produces a reasonable stop-gap solution to use of CRI with any source. This delivers and additional benefit of including specific performance minimums at specific colors, such as R9 (red saturated), when this is important to the intended application. While there are those who believe that the faults of CRI mean that it cannot be used at all in specification or selection of solid-sate lighting products, the suggestion that light sources be selected without any color standard is absurd. For the moment, use of CRIe is an acceptable, and applicable specification requirement. To address specific extended color metric values, I suggest the following:

  • In very low light, warm color applications, where sources are <3,500K CCT, selection of sources with an R9 value of >80 at >90CRI or greater is recommended. This will deliver observers the truest color perception in these environments
  • In retail environments where illuminance ranges between 30Fc and 70Fc, and light sources range between 3,500K and 4,000K, selection of light sources with >75 R9 though R14 is strongly suggested
  • In high illuminance task environments, select a light source with color characteristics that match the task in hand. Low color, high contrast (black and white) tasks, or where tasks do not demand high color differentiation performance can be specified using CRI alone. In applications where high color differentiation is a component of the task demand, the selection of CCT and CRIe, including values of R9-R14>70 will require balancing of CCT, S/P ratio, and CRI specifications to attain the best performance in all three metrics.
  • When in doubt, test products and specify the R values and Ra of the desired end products as the specified performance requirements.

Color consistency and appearance specification
Color consistency encompasses four dynamic layers:

  1. Within luminaires utilizing multiple LEDs in an array or linear row,  consistency between LEDs may be an important consideration. For example, in linear lighting applications where the emission from each LED, through discrete optics is visible on a wall surface,  variations between LEDs will be evident on the illuminated surface.
  2. Color over angle. This is most critical in applications where light emitted from a product strikes both vertical and horizontal surfaces. Inconsistencies in color are common in LED products, visible in light in the center of the light pattern generally being whiter or bluer than light at the edges of the beam pattern, which may be more yellow or magenta.
  3. Consistency between luminaires from one manufacturer’s products is crtitical. For example, recessed lighting products where visibility of the diffuser, lens or aperture will reveal differences from fixture to fixture, and on surfaces below and around the luminaire.
  4. Consistency of products from one manufacturer to another. This is a serious issue, where the products of several manufacturers are used together in one space. Failing to specify some standard color baseline and consistency requirement founded on an independent standard will result in color emitted from one manufacturer being visibly different to that of another. This cannot be used against any of the involved providers of product, or corrections required, unless a standard of color consistency from a know standard color point are specified.

Unlike conventional light sources, where specification of lamps from a single manufacturer could be used as a rough assumption of consistency of all light sources used in a space, or use of a sinlge lamp specification applied to numerous products from unique manufacturers, solid-state lighting products include LED light sources, frequently from numerous producers, including those used within a manufacturers product range. Unfortunately, at this time there are limited metrics available to include in specification to attain reasonable color consistency between products and within products. The following could be considered a minimal approach, founded on existing metrics available.

  1. Select the CCT based on available ANSI standard color specification per ANSI C78.377-2011. Use the Center Point for the selected CCT as the baseline for all color deviations applied to all manufacturers products, within fixtures or between fixtures.
  2. Specify a maximum  MacAdam ellipse value about the Center Point for theselectedCCT (without this requirement, differences between manufacturer’s products will not be controlled).
    • In spaces where color differences between LEDs or luminaires is not critical or of specific concern, specify a maximum deviation of 7 MacAdams ellipses (Energy Star CFL specification)
    • In spaces or applications where color differences between LEDs or luminaires is an important consideration, specify a maximum deviation of 4 MacAdams ellipsesper ANSI C78.376 (standard for fluorescent lamps)
    • In spaces or applications where color differences between LEDs within a luminaire, or between luminaires is a critical consideration, specify a deviation of 2 and no more than 3 steps.
  3. Specify a maximum variation in color over angle of no more than2MacAdams ellipse values at any angle throughout the beam pattern.
    1. For applications where variations in color within a luminaires beam pattern are critical, such as wall grazing, wall washing, or down-lighting that will also strike wall surfaces, include a requirement of a sample for review and verification that no significant variation in color exists that is objectionable in the intended use.
  4. To further insure that all colors are not presented as either green or magenta pink in comparison to one another, specify for each CCT a maximum dU’V’ value range of no more than .0002 to -.0002 from the Plankian locus at the Center Point for the CCT specified

The best source of actual color performance data from a luminaire is independent test lab results accomplished under LM-79 procedures, published as a chromaticity report.  However, this does not always provide adequate color-over-angle information required. Review of physical samples may be the only available option to insure satisfactory results and acceptance of a product submittal.

With these specification inclusions, and acceptance by those delivering products within them, corrections can be imposed to correct failures of color consistencies observed in application. Failure to include strong deviation and consistency specifications leaves interpretation of compliance and consistency in the hands of those providing the product, with little or no recourse for issues of between-manufacturer product differences that might arise in applications where luminaires from multiple manufacturers are included, or where multiple LED light sources from any one manufacturer exist.

Blue light content
Setting aside any academic discussion or blue light – there is no evidence that blue light from LED light sources presents any hazard to ocular or human health- assuming the brightness of individual LEDs is controlled and not aimed into the eyes of the observers or occupants of a space (which is poor practice with any light source, thus not specific to solid-state lighting). However, it is well documented that blue light in the wavelengths present in LED light, as it is in daylight, Metal Halide, Fluorescent, and some halogen sources, impacts melatonin suppression activity and sleep behaviors associated. For this purpose, consideration of blue light content from any light source, whether artificial (of any type) or natural, may be worthy of consideration in environments where disruption of sleep cycles is considered detrimental or undesirable. Examples of this could include dormitory room lighting, bedroom lighting in hospitality and residential environments, health car patient spaces, retirement and assisted care facilities. This effect will be most predominant where task lighting, such as reading lights, or light sources in direct view, will deliver sufficient retinal illuminance to create a physiological response. For this reason, in those spaces, selecting light sources of lower CCT (<3500) with high CRI (>90) may provide lower available blue light content than higher CCT/Low CRI light sources. If this is a consideration in the design of a specific application, review the Spectral Power Distribution specifications for the light sources under consideration and specify the maximum blue light as a percentage of total output, as a peak in the range between 430nm and 460nm. Otherwise, in high performance, high ambient and task level spaces, this is an unnecessary consideration. LED lighting does not present any greater impact, positive or negative, on melatonin suppression cycles than does daylight or other artificial light sources (save incandescent.) Further, consider that any impact light sources have on melatonin levels is temporary, lasting no more than 45 to 90 minutes from the removal of the source, or in total darkness. For this reason, there is no reason to consider this issue beyond spaces where occupants are expected to sleep within 90 minutes of exposure to any light source containing significant levels of blue light in the noted wavelength range. This means that regardless of inclusion of low blue light content sources for restful environments and sleeping spaces within a health care facility, hospitality, etc… application of this requirement in surrounding spaces, traffic areas, lobbies, exercise facilities, treatment rooms, or nurses stations is of little value, and may in fact produce undesirable results.

Rated lumen depreciated service life
While the application of lumen depreciation represented as L70 is widely used as a commercial standard for rated service life, the values provided by most manufacturers are misapplied and incorrectly stated. The correct current form of this rating standard has been established by IES TM-21, founded on data collected for the light source employed using IES LM-80 test standards, coupled to physical luminaire product test performance data. The procedure is relatively simple:

  • LM-80 testing of the LED indicates lumen depreciation of the LED source over time, at pre-set temperature levels, over a period >6,000 hours
  • TM-21 uses the data collected in LM-80 testing to mathematically project lumen depreciation beyond the actual test time
  • The manufacturer collects LED operating temperature from either in-house testing, or using tests completed in their UL investigation during listing processes
  • The manufacturer uses the LED operating temperature to factor, or interpolate, lumen depreciation of lumens based on mathematically factoring lumen depreciation for the specific temperature of the LED within the actual operating conditions of the luminaire involved

Because there exists a wide range of performance variables and factors of luminaire temperature, LED operation, etc.. the likelihood of  the result ending at 50,000 hours over abroad range of products is small. Further, TM-21 limits the projected lumen depreciation calculations to 5.5X test life for sample sets of 10-19 in LM-80 tests, and 6X test time for sample sets of >19 units. This means that the maximum lumen depreciation hours allowed to be reported under TM-21 for a sample set of 10-19 units, tested for 6,000 hours, is 33,000hrs. This applies regardless of whether the projected lumen depreciation using the calculations results in values greater than 70% at this point. For purposes of reporting, the lumen depreciated service life for the product in question will be stated as: >33,300L70 @ 6000hr LM80, or something similar. Based on this, specification of service life requirements of 50,000 or 70,000 hours precludes inclusion of any LED source that has not been tested for 9,090 hours (50,000hr) to 12,727 hrs (70,000hrs). In any case, the specification should include more specific requirements, within the bounds of TM-21, with which manufacturers can be held to reliably:

  • If the desire is for a source that delivers a lumen depreciation of a minimum of 70% of initial light to 50,000 hours:
    • Require LED sources to have been tested for a minimum of 8,333hrs (>19 sample size) or 9,090hrs (10-19 sample size) per LM-80 test procedures
  • If the desire is for a source that delivers a lumen depreciation of  a minimum of 70% of initial light to 70,000 hours
    • Require LED sources to have been tested for a minimum of 11,667hrs (>19 sample size) or 12,728hrs (10-19 sample size) per LM-80 test procedures

Ask to see the LM-80 data that the lumen depreciation values are based on. Many manufacturers are either fudging their numbers, using data from LEDs they are not actually using, or simply using unqualified depreciation data that does not include enough samples, or is not run long enough to determine realistic performance values. While there is no need to review every test in detail, look for the number of samples, and actual hours the test was completed over. Ask the manufacturer to provide a TM-21 test report using the Energy Star TM-21 calculator. This generates a report that can be compared to the LM-80 data, and will tell you whether the two are aligned, and whether the numbers provided are legitimate.  The report also shows whether the manufacturer actually bothered to create the interpolated depreciation input, which reflects the in-situ temperatures and values used to generate their advertised lumen depreciated service life claim.

A complete submittal should require a completed TM-21 report backed by the LM-80 data showing the LEDs employed. Without this, lumen depreciation service life claims can be anything the manufacturer wishes to publish, and may not reflect product performance at all.

Actual total luminaire service life specification and warranty demands
Lumen depreciation relates only the the depreciation of light output over time as it relates to the depreciation of the LED itself. This does not include any factoring of lumen loss from lens, diffuser, reflector surface degradation over time, dirt accumulation, or outright light loss (100%) due to failure of driver or power supply components. This compounds the issue of using lumen depreciation as a mode of service life determination, as the rated life of the LED, as set out by TM-21 and LM-80 data does not include any other component but the LED source itself. Aggravating this issue is that standard lumen depreciation values, for dirt, fixture age, etc.. used for conventional sources, assumes re-lamping and cleaning cycles of an average of 14,000 hours, so are not readily applicable to LED sources lasting three to seven times that. This will require re-thinking in the design of a lighting system around depreciation values compounded from fixture losses and LED lumen depreciation.

For supporting electronic components – power supplies and drivers, operated over extended periods of time, the issue of predicting failure mode become a consideration. The common standard for the failure mode of electronic components is often stated as MTBF, (mean time between failure). MTBF is a misleading indicator of reliability. Here is an example provided in an APC White Paper:

  • A population of 500,000 25 year old humans is used as the basis for evaluation over one year. (500,000 people years)
  • In that time 625 of the subjects died, or .125% per year
  • The MTBF is then calculated as 1/.00125 for a result of 800 years (7,008,000 hrs) MTBF

Obviously this is not realistic, but it demonstrates the failure of the metric for determining expectation of life for a product. MTBF is a reliability calculation, not a life prediction calculation, so application of this number to establish the life of a lighting product is a poor specification value. Further, just like in statements of L70 lumen depreciation values not founded on actual test data, most MTBF calculations are founded on pie-in-the-sky numbers common to a particular component in general, with no relationship to the actual driver or power supply being considered. Very few products sold into the lighting market are fully HALT tested to failure in statistically significant numbers. This means that most reliability values are calculated mathematically using discrete data for the included components.

If you reverse the math on MTBF, the oft quoted 100,000 hours MTBF indicates a failure rate of 8% per year (1/.08 = 12.5yrs or 109,500hrs). That would suggest an expectation that 8% of installed products will fail each year of operation. Since most failures of products occur at the early stage of life, followed by a trough of minimal failures, then escalating at end of life as product wears out, the 8% value is not constant. In fact, early and late failures may be several times this rate, while reliability during the middle phase is much greater.  That is hardly realization of anywhere near the suggestion that an SSL product installation will last 70,000 hrs, maintaining 70% of its initial light output.

Obviously, failure of a driver or power supply creates a 100% lumen loss. The question is then, how many hours in operation can one expect, or demand a solid-state lighting system deliver light, whether 70% of initial output or simply on-time before failure. Further, what can be expected as the in-operation failure rate, and what can be specified as a hard reliability value? The answer is not what anyone wants to hear. At this time, there is no established standard for determining the end of a products actual operating life, or its actual failure rate over a period of hours.

Aggravating the situation of MTBF value in determining actual life of a product, is the reality that few manufacturers actually perform life testing of their electronic components in the in-situ environments in which they are applied. A driver/power supply with a reliability rating at 25C will be derated significantly if it is exposed to a constant state of higher (say 50C) temperature operation, and an even more significant loss of reliability if operated at temperatures nearer its design maximum. LEDs lose lumens when exposed to heat, drivers and power supplies fail. If a lighting fixture is tested within the driver and LED manufacturers limits for temperature, the lumen loss from the LED will be included in the initial output value. TM-21 and the interpolated lumen output values will reflect the impact these elevated temperatures will have on lumen depreciation. LM-79 data will reflect the impact of the operating temperature on power consumption, and lumen output of the luminaire itself. However, none of this reflects or indicates the impact elevated operating temperatures have over time on the electronic components. In fact, lumen losses can be significantly accelerated as the driver/power supply components degrade under stress, as will power consumption of the products themselves.

All that set out, the next issue is the issue of lumen depreciation itself. As the installed products age in use, lumen depreciation will impart a loss that will need to be accounted for should any small number of products in a system fail and require repair or replacement. For example, if a population of products is operated for 10 hours per day for 30,000 hours before one fails, lumen loss for a system rated to deliver 70% for 50,000 hours will be roughly 18%. If the failed product is simply replaced, it will be brighter than the remaining products in service by 20% or more, due to lumen depreciation of the LEDs compounded by optic degradation, etc..  This issue exists to some degree with fluorescent and HID sources, but their service life is much shorter, the lumen depreciation over service life is much lower (fluorescent) and they rarely demand the entire luminaire to be replaced when the light source or ballast fails. For this reason, some consideration must be included to accommodate repair and replacement in such a way as to result in a satisfactory warranty remedy.

With no standard for reporting or indicating the actual reliable operating life of a solid-state lighting system of LED, optic, enclosure, driver and power supply, the relevance of L70 lumen depreciation rating for the LEDs utilized within a product seems almost irrelevant. This leads to the conclusion that the only specifiable demand is the requirement of a warranty that covers the following 3 components as they pertain to product life:

  1. Maintained total luminance (not LED loss, total fixture lumen maintenance) of no less than 70% over a specified number of operating hours (say 50,000 or 70,000)
  2. Full replacement and/or repair of failed products over a specific number of years, or operating hours, to return function to match surrounding products installed at the same time as the failed product.
  3. Availability of repairs or replacement for the balance of the rated lumen loss operating hours to maintain luminaire function until the LEDs employed have reached their LM-80/TM-21 predicted service life end.

This will be a highly contentious inclusion in any specification. First, the issue of “operating hours” over years creates reporting and accountability issues that few on either side of the argument will be willing to document satisfactorily. Second, replacement and/or repair components are likely to be difficult to deliver as time passes, as the technology simply moves too quickly for anyone to predict their ability to serve this requirement. Stocking power supplies for long periods of time in anticipation of replacement demand is not a practical solution. Power supplies utilize components, such as capacitors, that fail on the shelf over time, regardless of in-service use, so their is some likelihood that a replacement component will fail upon installation, especially when “operating hours” extends the amount of time significantly beyond shelf life – (50,000hrs is 14 years at 10 hours per day, for example).  Simply demanding 7 year warranties, or greater, will not resolve any of these issues, nor will it lead to manufacturers working toward greater reliability and longer service life solutions.

Pulling it all together
These factors, in total, bring together all of the factors involved in solid-state lighting product specification definition. Unlike conventional products that are a mix of hard goods (fixtures) and lamps, (which have limited variability for specification purposes), that have been stabilized over years of application, solid-state lighting presents a plethora of variables that must be addressed. How deeply one dives into development of a detailed specification is dependent on three major factors:

  1. The specifiers knowledge of the variables themselves and how they impact a lighting systems performance as a whole
  2. The tuning of inclusions in the specification aligned with specific needs on a project or within spaces of a project
  3. The willingness to stand behind the specification and demand it be met

Writing a specification around unfamiliar variables is a useless exercise, as it will only lead to conflicts within the specification that will be used by those more expert to break it. Writing overly-demanding specifications for spaces or projects that do not require every factor be perfect will raise costs unnecessarily, while failing to include important specification components in critical projects will lead to issues that are difficult if not impossible to resolve. Writing a great specification that protects the ultimate customer from field issues and unresolved failures is useless if there is not willingness to stand behind it and demand it be met. This is the art and science of specsmanship, and applies to a much higher degree in solid-state lighting than any other source before it.

If you would like to discuss this in detail, or would like help in creating a more precise boilerplate specification for your applications, let me know. With the right foundation, one can create a balanced end product to suit specific requirements, within the comfort level necessary to make it work. In some cases, this may include gaining a greater understanding of some of the details, and incorporation of evaluation tools for product selection that were unnecessary for conventional product selection.




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s