Light Thoughts

LED lighting offers claims of very high efficiency. Recently I’ve seen claims of over 100 lumens per watt and efficiency of four to five times that of incandescent. First, a refresher on efficiency (and efficacy) for those that need it. The following are link-backs to the previous posts I made regarding efficiency, and they should open up in a new window so you can just close it when you’re done reviewing to get back here.

• How Do I Know What’s Efficient?
• Efficiency Comparison Pitfalls: Age
• Efficiency Comparison Pitfalls: Systems

The problem with trusting efficiency claims from LED manufacturers is based on the complexity of the systems. The initial efficiency measurement is made with a controlled junction temperature of 25˚C and lasts for only a millisecond. These measurements are made of the LEDs themselves and before the LEDs are built into any sort of fixture.

LEDs are sensitive to current. As current increases, the efficiency decreases. Most LEDs used for lighting applications run at around 350 milliamps, or 0.350 amps. Like a fluorescent lamp requires a ballast to control voltage, an LED needs a driver to control the current. And just as a fluorescent ballast decreases the efficiency of a fluorescent lamp, the driver decreases the efficiency of the LED module. At present, the rule of thumb is that you can expect a loss of 10 to 15% efficiency due to the driver. Remember, the LED is an electronic component and the power supplied is connected by an electric component. The driver is the link between them.

LEDs are also sensitive to heat. As the temperature rises above or drops below the testing temperature of 25˚C (which is about 77˚F) the efficiency drops. LEDs are commonly described as not producing much heat, but that is only partially true. The current passing from the circuit board to the diode creates heat, so the heat that is generated comes out the back of the module, not the front. You can touch the front of an LED and it is cool to the touch, but the module itself will be mounted on some sort of heat sink that will be hot. When the initial testing is done it only last a fraction of a second, but in continuous use the efficiency of that heat sink comes into play. Again, the rule of thumb at present is that you can expect a loss of 8 to 10% for thermal management.

Playing even more into that loss of efficiency is sticking an LED module into some sort of fixture housing. You can’t just stick an LED module on your ceiling or wall as a light fixture. (In order to do that your wall and ceilings would have to be made out of circuit boards, so it would look like living inside a computer, which I suppose some people might thing was cool.) The LED module has to go into something, and at present the LED manufacturers and fixture manufacturers are different companies. A fixture manufacturer buys an LED module from someone else and then has to figure out how to deal with the heat and everything else. While the best fixture design possible might maintain that 8 to 10% loss in efficiency due to heat, a poor fixture design can increase that loss even more.

LEDs are also directional sources. If you take a regular incandescent light bulb and stick it in a bare socket, the light travels in all directions and lights up the whole room. If you take a halogen PAR lamp and stick it in a base socket, it lights up mostly in one direction like a cone of light. The LED is similar to the PAR lamp although it typically has a tighter cone of light. To use an LED to light up a room there has to be some method of optically controlling the light distribution. At present, the rule of thumb is that you can expect another 10 to 15% decrease in efficiency due to whatever method is used for optical control.

Aside: I’ve used a halogen PAR lamp as an example of a directional light source since most people are familiar with that type of bulb. Keep in mind that a PAR lamp is a combination of a light source an optical control system. The filament inside the lamp is omni-directional, just like a regular light bulb, but the glass and metal envelope surrounding it control the light to make it directional. The final result is less efficient than if we just uncovered the bare filament inside the bulb, but since we’ve packaged the whole thing as a lamp we ignore that loss and just use the efficiency of the entire unit for calculations. Keep in mind we’ve had a century of experience now at controlling omni-directional light, which is what we get from all incandescent and fluorescent sources. LEDs are the first fundamentally directional sources we’ve developed so we are still fairly new at utilizing the light generation.

So if we add up the efficiency losses from the driver, heat, and optical design, we’ve already lost 28 to 40% of the LEDs efficiency just by taking the thing out of the laboratory and trying to stick in into a lighting fixture (or flashlight, or whatever).

There are other losses in efficiency that are harder to illustrate. LEDs typically have a very high color temperature. Decreasing that color temperature to something that would be more appropriate to general lighting decreases the efficiency. LEDs emit light in specific regions of the color spectrum based on the materials used, which is similar to the issues regarding fluorescent lighting. When efforts are made to increase the color rendition, similar to increasing the CRI of a fluorescent lamp, the efficiency decreases.

The take-away here is that you can’t just look at the efficiency of the LED module and make an assumption about the efficiency of the LED light. The only way to determine the efficiency is to look at the efficacy of the entire system after it is designed and potentially installed. That why if you review my postings about the new lighting requirements for California they specify that the efficiency of LED lighting has be done based on the entire LED module/driver system, not just the lamping. The efficiency of the lamp all that is required for incandescent or fluorescent lighting options.