Specifying LEDs in lighting design
Specifying engineers and lighting designers need to understand more about LEDs and the ways information is presented so we can provide clients with the best lighting options.
There is a great deal of hype today about light-emitting diodes (LEDs) as a viable “near-perfect” source of light. LEDs are generally expected to last far longer than any conventional light source and be more energy-efficient. As a professional specifer, I hear the hype weekly, if not daily, and have a few findings and opinions that might help other specifiers understand what is going on, and where to look for reliable, current information.
Last year we relocated our own offices and have some personal experiences to share as a result. The lighting field is changing so fast that anything written is outdated before it is published, so please consult the references provided in this article to do your own investigation.
LED luminaires are built using single or multiple LEDs. Sometimes they are all on one assembly; other times individual chips are aggregated into larger assemblies and then used in a luminaire. This year I have specified luminaires with 1 to 170 individual LEDs. Performance of these LEDs has varied with Correlated Color Temperature (CCT) from warm at 2700 K to cool at 6500 K. Some dim, some don’t. Some are indoors, while others are outdoors. Understanding the lighting situation and the luminaires available is important from the lighting designer’s point of view.
Those of us old enough to remember the mass adoption of LEDs as indicator lights in the 1970s will recall the excitement when we saw blue LEDs. They were different, interesting, and far more expensive than the common red, amber, and green LEDs. We were quick to adopt these LEDs in exit signs, aisle lights, and traffic semaphores, for example. Little did we know that the blue LEDs were the innovation that would eventually lead to white LEDs and make LEDs feasible for general interior lighting. And the first “white” LED luminaires created the color white by mixing red, green, and blue LEDs to “make” a version of white light. This system is still used to get white lighting from LEDs, although it is not the most energy-efficient process, as we will discuss later.
Red, green, and amber LEDs have a relatively long life, between 70,000 and 120,000 hours. In these applications, LED life is measured in mean time between failure (MTBF) while maintaining usable light output. While measured precisely in the lab, the resultant light and life in the “real world” was not much of a problem as most of these applications could add LEDs or adjust the current limiting resistors to achieve the required systemic light output. This worked fine for most LEDs used in indication situations.
However, this “bench rating” of LED output was problematic for white LEDs. The mounting and operational environment of LEDs for general lighting purposes was physically not as accommodating for this new application. The operating environment for LEDs in general lighting service was far from the ideal temperature on the development bench.
As a result, in 2006 the U.S. Department of Energy stated that current industry claims of efficacy (remember claims of 150 lumens/Watt?) were not reasonably attainable in the field and could not be used as the basis of rating LED performance in the lighting business. At the same time, the government said that the lighting industry had to develop testing procedures designed to guide everyone to a fair and useful methodology. As a result, specifiers should use only valid test claims from manufacturers that state specifically that the data was obtained by following one or both of the Illuminating Engineering Society (IES) standards, LM-79 and LM-80, created to address the needs of LED illumination manufacturers. There are specific testing and rating guidelines and, as this is written, more documents are under development.
The U.S. Dept. of Energy has established the Caliper Program, a testing program designed to assist users in evaluating LED technology in lighting applications. The program’s website contains interesting test reports and technical information on LEDs. The Caliper Program is also the sponsor of the "Lighting Facts" product label, which is a reliable badge of technical honor.
LM-80 establishes the method for calculating the rated life of an LED: determine the light output at a specific point in the life of the LED to identify and standardize the acknowledged “end of (useful) life.” This is known as L-70, or the point where the lumen output of the LED is at 70% of the initial light output. This end-of-life determination is necessary because LEDs have the same end-of-life property as mercury vapor lamps. That is, they don’t die; they simply continue to fade out. I tell my students that this reduces maintenance on many a barn light across this country, because the lamp does not fail per se, but it is little more than a green glow over the barn door after 40 years.
LEDs have this same problem. The limit of useful lighting has to be determined, and L-70 is the agreed specification. Specifiers need to understand the meaning of L-70 and base their project calculations on that understanding. I have seen manufacturers use other values, such as L-80 or L-50 for specific messages—that is, mean lighting level for interior luminaires for the L-80, and a claimed useful life for L-50. Unless product comparisons are made using the same reporting basis, however, the differing numbers could lead to a misunderstanding of performance.
While on the matter of specifications and calculations, the light output for all LED luminaires is measured by using “absolute” photometry, tested per LM-79. This is because the photometric performance of an LED luminaire is completely dependent on the composition of the specific LED engine. Most other luminaires are tested and legitimately report using “relative” photometry. This is useful in facilitating changing lamps in the same luminaire; for example, a compact fluorescent (CF) downlight can be simply recalculated changing the compact fluorescent lamp lumens with a high degree of accuracy. LEDs are not this flexible, so a photometric test must be run for each lamp assembly it will hold.
Individual performance reports
LEDs get some of their optical advantage by being natively directional. Therefore, each individual LED has a lens, reflector, prism, or other individual optical control device. This directionality is used in downlights (street and area lights, for example) to affect the overall luminaire performance. As a result, LEDs require clusters of optical controllers, not one reflector or lens over the whole assembly. If you look closely at LED luminaires, you will see the individual lenses or prisms. This works well, but we must remember to get the fixture-specific photometric file, and not “edit” the one we have. This applies to a previously interchangeable metric such as changing lamps to change color temperature, for as we will see, many LEDs have different light outputs that are variable dependent on color.
Remember to find the photometric performance file for the exact luminaire (or at least specific LED assembly) that you intend to use as your basis of design, and be careful in evaluating substitutes for compliance with your important project criteria.
There are a few LED assemblies that offer “lamp” interchangeability, but these are just coming on line. Today it is not easy to “change the lamp” as we are used to with other forms of lighting. I am confident that as the LED matures as a light source, it will restore some of this design and specification freedom. Stick with individual performance reports, using absolute photometry, and surprises and disappointments will be minimized.
How “white” LED light is made is interesting. We previously discussed the mixing of colored LEDs. RGB is the most basic technique, with amber sometimes added (RGBA), or even up to seven colors (ROYGBIV; look familiar?). The most common approach with individual LEDs is to use a blue LED and insert a yellow phosphor filter inside the package to create white light. The third method of creating white light using LEDs is to use blue or violet LEDs to excite a phosphor that is a part of the luminaire but separate from the LED that emits the white light. Each of these has advantages and disadvantages.
The advantage of mixing colored LEDs to achieve white light is the color flexibility and color effects that it allows as well as mixing to create white light. This is very effective in display and entertainment applications.
The yellow phosphor on a blue LED device is by far the most common white LED used today. Since there are many variables in the manufacture of these LEDs, they are individually rated for their performance by the LED manufacturer. They are sorted into “bins” following ANSI standard C78.377-2008 for determining the CCT of LEDs as devices. The light engine manufacturer can use closely “binned” LEDs and have more consistent color—at a higher cost—or accept (or use) a wider selection of LEDs from more bins and reduce the cost, balancing this with performance. Since it is more expensive to produce warmer LEDs, most applications that do not require warmer light (parking lots and roadways, for example) allow the use of visibly cooler LEDs to achieve lower product cost.
There is a human factor for this “bluer” light, however. As we age, the physiology of our eyes begins to scatter the blue wavelengths of light. In extreme cases this makes blue light appear glary. I have had clients select the orange light of high-pressure sodium lamps specifically because of their perceived comparative comfort, without glare. The optimum solution would be to use warmer LEDs in these applications, which will be more practical as the cost and performance of warmer LEDs draws closer to matching the performance of cooler LEDs. There is research that identifies an increase in human health issues associated with blue light in the sub-500 nm range (National Institutes of Health: Meeting Report: The Role of Environmental Lighting and Circadian Disruption in Cancer and Other Diseases-2007). The light from cool LEDs falls in this band. Specifiers might want to watch for developments in this area. For now, avoid using blue LEDs in night lights.
The third method of creating white light indirectly from LEDs using the separate phosphor technique has the advantage of more consistent light color creation that more closely follows conventional specification. The science of using phosphors to create visible light from nearly invisible energy is well known. Using this method is creative, effective, and comfortable for lamp manufacturers. There are inherent systemic inefficiencies in this method and some packaging and optical differences that need to be addressed. I have found the color advantages and the less “technical” appearance of this method to be worth the effort when considering white light in residential and many commercial applications. This method also requires continuous evaluation because it too might change with each product improvement.
There is a limited color-mixing technique used to create variable color temperature in the “white” range. This technique uses extremely warm “white” LEDs and extremely “cool” white LEDs and a dimming system to mix them to achieve almost continuous white light with a changeable CCT and variable overall intensity, near full-range dimming. This is useful where the room needs to dramatically change character. Examples include restaurants facilitating the breakfast rush as well as nightly dining or a retail paint sales department that benefits from making the whole area a “color box.”
The other metric used to evaluate light is known as the Color Rendering Index (CRI). Many LEDs have difficulty generating the red light wavelengths necessary to produce light equal to today’s tungsten-halogen sources. LED technology will undoubtedly continue to address this issue, which is one of the advantages of the improved phosphor method. The LED industry, particularly those from outside the lighting industry, have tried to raise the shortcomings of the CRI metric itself, saying it measures the wrong thing. The initial CRI metric was based on measured color performance on largely pastel colors, but now there is an expanded CRI color field that includes four more saturated colors. The current technical reference for color rendering that mentions performance on the “R9” color is directly referring to the light source’s ability to generate and therefore render red light for lighting the red in objects. Some manufacturers select LEDs from adjacent bins, favoring the warmer colors to emphasize the red spectrum. This represents care and a concern for a high level of performance. Other manufacturers allow wider tolerance of LED color, perhaps alternating warmer and cooler LEDs in adjacent positions in the luminaire. In my opinion, this technique was acceptable, even creative, when the cost and availability of tighter performance LEDs were more difficult to achieve, but this is neither needed nor acceptable today. Specifiers must still be sure the manufacturer’s tolerance for color variability will not be seen in the application. Coves and indirect lighting situations come to mind, where there is a chance to see a considerable number of lights at once. Care must be taken to ensure these light sources match as expected when installed.
LED efficacy and droop
The efficacy of these cooler chips is higher than the warmer colors, so there really is more light (lumens)/Watt of electricity consumed. This improved efficacy at the blue color can be used to compare many LED lighting proposals, comparing metal halide to LED for example. Remember, because there is a difference in the light output or the electrical wattage consumed, even changing by color, the full picture must be considered when comparing possible lighting systems. There is a misconception in many quarters that light sources can be compared and selected based on CRI alone. In fact, CRI is only appropriate as a comparison metric between light sources of the same CCT. It will help you keep manufacturers’ claims and your requirements clear in your mind.
Based on today’s LED technology, there are reasons to think that we have nearly reached the technical limits of the amount of useful light that can be created by any single LED chip. There is an excellent discussion on LED “droop” in the August 2009 issue of IEEE Spectrum magazine. Given this limitation, we must look to the pure research labs to resolve this problem. Meanwhile, we are solving our need for more quality light from LEDs by getting larger and larger luminaires.
LED power supply
The next issue to consider is the power supply or LED “driver.” The specifer rarely creates a specification for the power supply; it is selected by the luminaire manufacturer. The specifer should be certain the LED array and the power supply are suitable for the project and compatible with the project lighting control system. Until very recently, there has been little improvement on power supplies for LED luminaires. Since the LED array itself was created by electronic engineers, the power supply was developed alongside the LED, and frankly, it was an easy undertaking for these engineers. However, this process caused a delay in the development of more universal or standardized LED arrays and power supplies. This is changing. Today more LED luminaires are being developed using standardized drivers. In fixed light scenarios, where dimming is not required, improvements to these drivers are leading to reduced line Wattage. In time, more efficient power supplies will become common.
In three-color mixing systems, each of the colors has to be dimmed to mix colors, requiring variable individual power supplies. More line Watts are consumed to drive the LED as compared to the lowest approach, fixed individual power supplies. Therefore, lower line Watts are possible using single LEDs in general applications.
The LED power supply that dims can have any of the three common control techniques (forward or reverse line voltage, phase control, or 0 to 10 Vdc control) depending on the particular driver. Specifiers must pay special attention to these details as some manufacturers have recently adopted third-party LED drivers and changed existing products without much notice.
Today’s ceilings and light poles can handle larger luminaires. Luminaires with larger masses of LEDs require better thermal management as a part of the luminaire design, whether the luminaire is in a plenum or on an outdoor pole. Very few conditions will diminish the performance of lamps in general, and LEDs in particular, more than heat will. Once the issue of recognizing the plenum as the customary location for LEDs in interior lighting situations was resolved, luminaire manufacturers addressed the critical need for thermal management at the LED. As designs evolve, watch for thermal management to become a center of improvement. Be very aware of your intended ambient temperature at the location the fixture is installed.
When LEDs are dimmed, there is not the customary color shift toward warm as seen with tungsten-halogen dimming. In fact, the opposite may be true depending on the amount of dimming and method used to generate the red light energy. This cool shift may or may not be very pronounced.
The last issue specifiers need to be aware of is emergency operation of LED luminaires. Until now, there was little standardization of the power supply. In my own offices, we circuited together the emergency LED luminaires as planned. We ran them together through their control groups and attached them to a consolidated battery inverter. There are a few commercial battery inverters available similar to the ones we are used to seeing in the CF world. While this field will standardize in the future, care is particularly necessary today in selecting the method of powering emergency LED luminaires.
Specifiers must be very careful in selecting and approving LED light sources from a color and CRI perspective. The pressure will be present to select the cooler colors, probably more cool than the specifier is accustomed to. If high color reproduction is needed, insist on LEDs with sufficient red light energy. Be aware of energy claims and the specification of the power supply (check things like harmonics). If dimming is desired, compatibility with the control system must be coordinated. The cost of the LEDs is up to 40% of the actual cost of the luminaire, so selecting luminaires that tightly select the LEDs they use is generally worth it. Ask about the manufacturer’s binning specification. Careful manufacturers will know (and be proud of) their tight binning.
Some semiconductor companies in the LED lighting business are not as attuned to all the human needs of lighting and are learning about the need for improved CRI, for example. Yet these are the very companies that will overcome the light output or droop problems. Meanwhile, the standardization of the LED driver features will lead to improved electrical efficiency and advances in dimming and staged switching. Because this is truly a technology in its infancy, and manufacturers are “improving” their products constantly, specifiers need to exercise extra care in the products they consider for their projects.
In summary, specifiers need to understand more about LEDs and the ways information is presented for our consideration. Part of this is due to the way various manufacturers report their performance, and part of this is due to the rapid changes in this exciting field. Specifiers will always do the best they can for their clients, but two challenges remain: separating the truth from fiction and knowing the important criteria and features for each project.
- Good, LC, IALD, FIES, USITT, LEED AP, has a bachelor’s degree in lighting and theatre design from the North Carolina School of the Arts and has more than 30 years of experience in lighting and theater design. He is a principal for Salt Lake City-based Spectrum Engineers where he designs high-performance lighting systems featuring sustainable solutions. Good is a member of the International Assn. of Lighting Designers (IALD) and the U.S. Institute of Theatre Technology (USITT).