If we’ve got LED fxtures that use ~15% of the energy of incandescent, consider that at 100 watt bulb will cost ~$100 PER YEAR to operate if on 24/7 (at 12 cents per kWh). Even if you spend $100 on a high end LED fixture, you get full ROI in a little over a year – yes, if that is in a location where the light is on only 6 hours a day, then it takes 4 years to recoup, but still, that is pretty impressive.

Worth it? Absolutely. It helps to run the numbers on the energy cost. Consider an industrial application like an airport bathroom, with 24 light fixtures (yes, I counted), all on 24/7/365. A large airport may have 100 bathrooms. If they can cut their lighting costs by 85%, or even 50% on conversion from fluor. to LED, the savings add up very quickly.

In this discussion of lighting quality, usually couched in terms of CRI, people have offered incandescent lighting as a standard, but few (none?) have mentioned sunlight. Some one hundred million years of evolution have adapted our vision to sunlight, with moonlight running a close second for naturalness. Humans have evolved with firelight for perhaps a couple million years.

Fluorescent lighting and LED lighting (which itself has aspects of fluorescence in the case of white LEDs) lack fidelity to the standard of sunlight in many, perhaps unlimited, ways. In fact, as the Wikipedia article on Color Rendering Index mentions, it is difficult to even compare such light to the black-body radiation spectrum of sunlight or incandescent light, so CRI values have less meaning than you might believe.

Our eyes have four color receptors (not neglecting the highly sensitive rods, you see), and any lighting system attempting to provide satisfaction must at the very least emit light that excites those receptors similarly to sunlight. The comments here on blue-yellow “white” LEDs reflect our reaction to this artificial spectrum, which has a single band to cover red and green alike.

Perhaps future lighting systems will be adjustable, producing light that mimics sunlight, moonlight, firelight, according to the user’s desires.

However, placing the emphasis on efficiency (“green lighting”?) will work at cross purposes to producing light of high quality. And be wary of any attempt (including the CRI) to reduce the measurement of lighting quality to a single number. Caveat emptor.

Real high-def LED Screens are made by Daktronics and others but are huge, for
stadiums and concert venues. They are amazingly bright and clear, but we will have to wait a while for a home-sized true LED tv.

As for quality LED lighting, there are some dynamite products from Cree-LLF,
including a Par 38 that provides halogen-quality light. (92 CRI)

Philips Color Kinetics range of products are also super high brightness and quality, white, and RGB, as are the products from Renaissance Lighting.

Prices are high, but are coming down almost weekly, and these product do live up to their claims of life and color and lumen maintenance.

I plan to replace my track lighting with LED lights simply because they look *better* than the traditional incandescent flood lights I’m using now (and way, way better than CF lights). That they use a fraction of the power is merely a plus.

They are a bit too expense right now, but the price will inevitably come down.

Can you comment on specific brands? What do you think of the architectural lighting developed by Cree? They claim to be paying a lot of attention to things like how the heat gets rejected from the fixture and so forth, and they also claim to be having a lot of success in installations like airports, exhibit halls, and auditoriums.

If you can comment on that or any other brand, please do!

And the bit with nystagmus and the PWM, someone’s just got to do a better job. You can chop LEDs plenty fast, faster than the 120Hz that we can easily see. For example, a “BuckPuck” driver supports a pulse rate of up to 10kHz, and the driver (slower than the LED) has a rise/fall time of 25uS or less. It would be a bit of design work, but in an array, you also ought to be able to chop sets of LEDs at different phases, which I think would fool your wiggling eye. Of course, the whole point of cheap snake oil, is to not bother with the design work.

If CRI is important, avoid the “brightest” LEDs, which will inevitably be “cool white”, and emit the gentle light of a welding arc. I’m using CREE XRE “Neutral White” in my kitchen, and they’re pretty good, though not perfect (there is a slight green, not much — there’s a picture on my blog, search for “undercabinet”). I bought more, later, and the second batch seemed noticeably greener. If I had it do over again, I would mix it up to include some warms and a few cools, as insurance against an overly green batch of “neutral” LEDs.

And 42 lumen sounds about right for the tech we have now. Luxeon and Cree both make some fairly high powered leds that are pushing 200 lumen from a single unit. However you’ll still be burning 5-6 watts to do that. (You are starting to see these get popular in the flashlight market as well.) And there’s another issue, heat.

Anyone familar with processor cooling in a pc should understand the issue with a die shrink and cooling. If that led is tiny, and it has to dump ~5 watt of heat, then it’s gonna get hot, and if not cooled well, really hot. I retro fitted an old 6V lantern flashlight to a Luxeon K2 led, and was suprised by how much heatsinking the led assembly needed in order to maintain brightness without exceeding it’s temperature limits.

I think the high power market will drive along the efficiency of the smaller units, and if things like spray/paintable semiconductors work out then led’s would be an easy choice. Until then, they have their market. Replacing 100W (equivalent) CFL’s is just not one of them.

jwb: I’m not going to get too into back-patting because my information on which companies are the good ones comes very second-hand. But it is my understanding that Cree are part of the good guys on this–producing quality products and being reliably honest about them.

Of course many applications use lights more powerful than 20W in the first place.

LEDs are different; the basic mechanism produces a range of frequencies grouped around a center frequency, and, if you use those same phosphors to convert blue/violet into white, you get a multiplied range. My understanding, from reading wikipedia (and my son had a bike light that did this) that are some that just do “blue+yellow”, and those are crap. Good ones don’t do that. Unfortunately, your only hope for getting a “good” LED is to buy a from a reputable supplier, or get your hands on the data sheet for the LED you are buying (or both). Even with good ones, there seems to be a blue spike at the underlying LED color.

Examples below:

HALOGEN DIFFRACTION

CFL DIFFRACTION

LED DIFFRACTION

Have your friend consider ceramic metal halide. There are new MR-16 sized CMH lamps that are perfect for retail environments. Long life, great color and high efficacy (lumens per watt). The only issue is the 5-7 minute warmup time to stabilize the salts in the morning but they are what I recommend for retail establishments. Everything the LED promises but cannot yet deliver.

The main issue is going to be purchasing new fixtures, there are no retrofit kits due to the need for a ballast for the lamp.

Maggie, great article, nice to see rationality and facts brought into the hype of LEDs. They’re just not there yet everyone!

There aren’t that many ways to dissipate heat. A chunk of metal can increase dissipation area, but alas you are still ultimately screwed if the heat sink is in an enclosure. Forced air is messy and complicated, solid state cooling just moves the problem a few cm away, water and VCC are just not an option for commodity products like light bulbs unless someone can produce MEMs size cooling devices that are incredibly cheap <$!1.00US

Heat dissipation is the limiting factor in all transistor based devices.

OK, there’s hype. But what are the real world (architectural etc) applications where good LEDs are possible?

For example, a friend manages a medium retail supermarket, using a lot of MR16 halogen spots for product lighting. I’m telling him, (based on theory) LED substitutes would save him a lot.
He needs long life (at least 2 years plus), good light (not flat white but with a little bit of warmth), and a purchase figure that I can rationalise into calculating a saving for him, (say something like 4x the cost of a standard MR16 halogen but with 6x the life & 1/4 the running cost).

Right now, all we’ve got are product manufacturers claims. And that’s too risky to take an (expensive) punt on.

An LED doesn’t have this problem, and so it can produce useful amounts of light for longer.

When incandescent bulbs become banned, I will be buying a hundred or so and hoarding them for the rest of my life. I don’t accept anything less than a CRI of 99.

More than a few mentioned the greenish cast (C)FLs give to a scene.

FWIW, CFLs and FLs give things a greenish tint because they produce light by a two-step process: ionizing Hg (mercury) vapor which produces blue-purple light and UV (ultraviolet); the phosphor inner coating of the (C)FL absorbs the blue-purple & UV and it fluoresces, re-emitting visible bright white light. An FL w/o the inner coating is aka a “black light”.

As it happens, the spectrum of ionized Hg vapor also has two strong green lines, quite close together. These green emissions pass through the inner coating with some attenuation, but are still strong enough to distort the color “picture” we see.

I have purchased 45 LED bulbs and have had mixed reliability.
The good news – some are very reliable. I have five LED bulbs outside that have run dusk to dawn for two years with no problems.
The bad news – some bulbs are VERY unreliable. VERY high failure rates.
I purchased 12 of one type LED bulb and 7 have failed (8.5W product 47856 from LEDLight.com). To make matters worse they are refusing to replace them now.
Beware of LEDLight.com. This company is selling products that they know are defective. No support for failed LED bulbs. These bulbs are very expensive ($20 – $105) and in some cases last only two or three weeks. They refuse to replace defective bulbs. LEDLight.com is selling known defective products and has bad customer service.

That’s an interesting question. My suggestion is to look to the Internet, specialized lighting stores, or talk to lighting designers.

As several people have said, companies like Cree seem to be doing a good job (And there are others that do as well! I just haven’t done enough research to be comfortable giving you a list.).

Good stuff is out there–especially for exterior/parking/street lighting–but it won’t be cheap and it’s probably not something you can find at any of the places you usually shop.

They’re doing some awesome things with parking garages and street lighting, using LED’s ability switch on and off without a warm-up time.

Think: Parking garages that are set to motion sensors. You save a ton of energy and money when the lights don’t have to be one full-bore 24/7.

Also: Street lights that can be programmed to blink in front of a specific house so that ambulance crews can find it faster at night.

Very, very neat. Ann Arbor, Michigan is in the process of going LED for their street and public lighting.