Skip navigation
Electronic Design

Beyond The $10 Million Light Bulb

How many EEs does it take to screw in a light bulb? Maybe we should ask why EEs should care.

Signed into law in January, the Energy Independence and Security Act of 2007 directs the U.S. Department of Energy (DOE) to establish the “Bright Tomorrow Lighting Prizes” (L Prize) competition. This contest is designed to spur the development of ultra-efficient, solid-state lighting products to replace the common light bulb.

Specifically, the DOE hopes to replace the 60-W incandescent lamp and the PAR 38 halogen lamp. It also calls for a “21st Century Lamp” that delivers more than 150 lm/W. The competition will award significant cash prizes, plus opportunities for federal purchasing agreements, utility programs, and other incentives for winning products (see the table).

Prizes, which aren’t funded yet, will total as much as $10 million for the 60-W incandescent replacement lamp and $5 million for the PAR 38 halogen incandescent replacement lamp. Specs and prizes for the 21st Century Lamp will be determined later. Program details are available at

Whether or not Congress funds the contest, the fact that light bulbs snuck into a bill that primarily addresses fueleconomy standards for cars and trucks is significant for design engineers. In fact, there’s more important news.

Energy Star criteria established last September added quality specifications to simple energy efficiency for lighting. The criteria are based on a proposed American National Standards Institute (ANSI) chromaticity standard, a now-approved “Luminaire Efficacy” standard, and a “Lumen Maintenance” standard.

These criteria require indoor fixtures to have a minimum color rendering index (CRI) of 75, zero off-state power draw for the fixture, and a power factor no worse than 0.7 for residential use and no worse than 0.9 in commercial use. Lumen maintenance must be better than 25,000 hours for indoor use or 35,000 hours for outdoor use. And, luminaire efficacy must be 20 to 35 lm/W, with the prospect of higher efficacy requirements looking forward.

Most EEs can run a string of LEDs from a dc source. For more subtle issues like dimming and strobing, a plethora of ICs hails from Analog Devices, Linear Technology, Maxim Integrated Products, National Semiconductor, ON Semiconductor, Texas Instruments, and others. (If you don’t want white light, see “Designing Multichannel HBLED Systems,”) Yet when it comes to lighting, we tend to be less well versed in the nuances of color selection than we’d like to be. What, for example, is that “ color rendering index” in the L Prize table?

Part of this explanation comes from a long talk with Mark McClear, director of business development at Cree Inc., and part is abstracted from an article called “The Color White” in Architectural Lighting by James R. Benya (

As McClear explains it, there are two kinds of “white” LEDs. The ones used for “mood” lighting and in some LCD backlighting devices are triads of red, blue, and green LEDs. Those used for general illumination have only a single diode. Blue light from its junction excites a phosphor system on the inside of the glass bubble over the junction, causing the phosphor to emit polychromatic light across the visible light spectrum, though not uniformly.

A “cool” (6000 K) white LED emits a combined spectrum like the blue curve in Figure 1. A pronounced blue peak develops at about 460 nm from the LED, along with the spectral response of the phosphor for the other wavelengths. Different phosphor systems produce different relative amounts of green (555 nm) and red (620 nm) as well as the other colors.

The red curve in the figure shows a much stronger response toward the red end of the spectrum. You’d call that a “warm” white LED. Warm white LEDs closely match the colors typically demanded by indoor lighting situations. On the other hand, “cool” white LEDs are about 30% more efficient and lend themselves to outdoor applications like street lights.

The human eye is extraordinarily sensitive, so small process variations in chip wavelength make a big difference. So do phosphor thickness, concentration, and composition, as well as deposition. In any batch of LEDs, there’s enough variation in chip wavelength and phosphor system behavior to have a noticeable effect. This leads LED makers to bin for color, and understanding that kind of binning is the trickiest part of designing lighting systems with white LEDs.

The international standard for describing the color of light is the CIE Chromaticity chart (Fig. 2). On that chart, it is possible to draw a curve called the Planckian black-body locus (BBL) that describes the color of light emitted by a theoretical non-oxidizing object as it’s heated. At about 1000 K, the object will radiate a dull reddish light. As the temperature increases, the object’s light becomes warmish white, then coolish white and then blue above 10,000 K. A measure called the correlated color temperature (CCT) defines specific points along the BBL.

In the real world, only the sun and incandescent light bulbs closely match the BBL, and such bulbs can only do so until their filament melts. To describe the light from other sources, such as high-intensity discharge (HID) lamps, fluorescents, and LEDs, we have the CRI. In this 0-to-100 scale, 100 represents color rendering equivalent to a black body. Not even sunlight is 100 CRI all the time, as clouds and even window glass act as filters. (CCT, however, remains high.)

As a practical matter, people who design lighting fixtures insist on a CRI of at least 80, and this has been the basic challenge for LED manufacturers. The L Prize, of course makes it even tougher.

Contiune to page 2

Much of the drive behind this sort of engineering came out of fluorescent lighting in the 1950s. The science tends to be a little touchy-feely, though, with a lot of studies dealing with “warm” (reddish) and “cold” (bluish) light sources and diurnal rhythms and melatonin production.

Whether the results are hard-edged enough or not, two things emerge: good correlation exists among individuals in groups that share a common culture, and people who design lighting systems for offices, restaurants, hotels, and schools take it quite seriously. As a result, LED makers and designers pay attention. Indeed, touchy-feely notions sometimes turn out to have solid roots.

“From the 1960s to around 1990, an unexplained phenomenon called ‘visual clarity’ was a common discussion topic among lighting practitioners. When viewed under the light of high CCT, high CRI lamps, many tasks appeared easier to see, yet the cause could not be explained by the vision scientists of the era. Starting in the late 1980s, important new work by Dr. Sam Berman and his colleagues at Lawrence Berkeley National Laboratory finally provided an explanation,” Benya wrote in “The Color White.”

“An abundance of shorter wavelength blue light causes the pupil of the human eye to contract, and through increased depth of field and reduced visual noise, visual performance is enhanced. Further experiments by Berman, Navaab, and others have also demonstrated that high CCT light appears brighter than lower CCT light at the same footcandle levels,” he continued.

“Based on these findings, large-scale experiments retrofitting complete buildings with high CCT lamps, but lowerthan- normal footcandle levels, have been conducted by Pacific Gas and Electric with some success. \[Although\] whether this is an acceptable practice is still being debated among vision scientists,” Banya wrote.

McClear also starts with the chromaticity chart of Figure 2, adding some notations of his own. Down in the left corner, the blue LED emits photons with a wavelength of approximately 460 nm. To produce white light, McClear draws a heavy black line across the color model, through the locus of white light, to find the equivalent wavelength of the phosphor system that will balance the blue of the diode. To indicate the uncertainty of the blue LED wavelength, it’s a fat line. McClear also draws several dotted paths to indicate the uncertainty of the phosphor wavelength. All of the variations in chip-process parameters and variations in phosphor chemistry contribute to variations in color produced by the LED.

Figure 3 is the same color model, but now we’re tightly zoomed into the white area. The magenta line is the black line McClear drew on the color space between the LED emitter wavelength and the phosphor wavelength in Figure 2. This figure also includes the BBL, which would have been hard to show at the scale of the previous figure. The colors around the edges of the “white” area bleed in from the more saturated areas of the color model. If you get far enough off in LED or phosphor wavelength, the nominally white LED light output takes on tinges of those colors.

The boxes around the BBL represent possible bins of LEDs. In fact, these particular boxes represent Cree’s bins. Other LED makers will bin differently. Cree’s engineers work with customers to help them determine which bin or bins fit their needs. The lighting system designer provides the basic input and makes a judgment based on the criteria Benya wrote about.

Gas stations, restrooms, and hotel lobbies all demand different kinds of “white” light, and it matters whether you’re in Trondheim or Tahiti. What is critical is that “care be taken to match the bins above and below the black-body locus. Equal numbers of LEDs from the bins above and below the BBL will color mix in most applications to produce a white chromaticity that appears to be right on the BBL. To help the mixing, diffusing films can also be added,” says McClear.

In other words, it takes a lot of work to get the CRI as close to 100 as possible. That’s why it’s not going to be a walk in the park to meet the 90+ CRI requirement to qualify for an L Prize.

But if the L Prize is still a chimera, Energy Star is real. It affects how many governments around the world manage their procurement and how many ordinary people make buying decisions. Under the new standards, quality of light matters a lot.

Three brand-new or ready-to-bereleased standards matter. The ANSI chromaticity standard was approved in March. The IESNA (Illuminating Engineering Society of North America) LM79 luminaire efficacy standard was approved in May. And, the LM80 lumen maintenance (how long the LEDs last) standard is still in draft form. Power factor, also an issue, is part of the ac adapter design.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.