Luminous Efficiency of LEDs vs. CFL

There is a lot of misinformation about the efficiency of LED lighting. Much of the information out there is not about what LEDs can do today, but rather what they are working toward. In the end, what we all care about, when trying to save energy, is Luminous Efficacy, which is the efficiency of the conversion of electricity into usable light. It's specified in Lumens per Watt.

A good quality fluorescent bulb with an electronic ballast provides on the order of 100 lumens per watt. If it's a dimmable fluorescent, that number can go down quite a bit when dimmed to low levels (down, you ask? Yes, there is a fixed amount of power to run the ballast that doesn't change with dim level, which isn't converted into light). A compact fluorescent is usually in the 50 to 80 lumens per watt, I understand. Only the most exotic prototypes in a lab environment an beat this with LED sources. Most of them are well below what a compact fluorescent can do.

Of course we all want a better light source, so enormous amounts of money (both private and public) is being poured into LED research. I'm sure, in time, the efficacy will improve. I found this nice article about the fundamental sources of heat produced in LEDs, which gives a good outline of where the waste energy is going, but it stops short of giving efficacy numbers. The biggest challenge today, I think, is thermal management. If the LEDs are run at too high a temperature, they will not live up to their promise of long life. CFLs suffer the same problem, and in the wrong kind of fixture they only last a few months instead of the many years they are capable of.

The best way to save energy, now and always, is to turn off lighting that is not needed. While I still train my children to turn off the lights when they are not in the room, they still forget often enough, and the motion detectors I've set up take care of the rest.

CFL power quality

Long-term (assuming that CFL bulbs become a substantial percentage of the load on our power grid) the cheap power supplies that are built in most compact fluorescent bulbs will create power quality problems. See this article for some details.

While I agree that it's a problem, having designed power supplies like them, I would assert that it's already been a problem historically, and perhaps is presently, too. The harmonic content generated when a rectifier sits directly on the AC power line, connected to capacitors on the DC side, is pretty nasty. Perhaps counter-intuitively for those of you with an electronics background, the noise becomes worse if you use more capacitance after the rectifier. Think of it this way: a larger capacitor doesn't droop as much between power cycles, so the recharge occurs over a smaller piece of the power line cycle, making a shorter (but higher-amplitude) spike of current. During the remaining power cycle, little or no current is drawn.

While this doesn't represent the traditional inductive or capacitive load (a simple phase lead or lag of current) it's a power quality problem. Capacitor banks on the utility distribution lines will not solve the problem. In the end, the power supplies at the end of the service line must be improved.

As an example of this problem, see this image below of the AC waveform at my house. Compared to a pure sine wave, the peak is significantly clipped. I believe it's due to non-power-factor-corrected power supplies that are in use throughout my neighborhood (and I think the same is true of most parts of the power grid).