Slashdot Mirror
Lighting The Future: Lasers And (Wild) LEDs
Effugas writes: "Thank Memepool for pointing everyone towards The LED Museum, an absolute geek haven if there ever was one. The Museum archives, tests, and describes the multitude of LED types out there, from Ultraviolet to Turquoise to Infrared--and, most amazingly, true full color LEDs! Apparently, I wasn't the only tech drooling over the possibilities that these devices could be put to use towards: Color Kinetics threw 75 of 'em into a spotlight can and added a surprisingly versatile digital controller with external interfaces. The result: A 16.7 Million Color Programmable Spotlight, for $500US." This is an incredible site; I look forward to when the full color LEDs (and that spotlight - Yow!) are a lot cheaper. And unaccountably, Effugas has linked that "A" to something awful.
mindpixel writes: "Sandia National Laboratories is reporting the demonstration of the first UV microcavity laser which could be used to replace gas-filled fluorescent tubes in home and commercial lighting applications. Successful penetration of the lighting market by this and LED technology by 2025 "should translate [globally] into cost savings of $100 billion a year, power generation capacity reductions of 120 gigawatts, and carbon emission reductions of approximately 350 million tons per year (assuming that all the savings come from coal-fired plants)."
"
I rigged up a color-programmable fishbowl full of Christmas lights (RGB) through three dimmer switches for the same effect, much cheaper... I have noticed, however that there is a bit of a heat problem. I wonder whether LEDs would be any better in that regard.
Incandescent bulbs (Christmas lights) use a tungsten filament heated to white hot inside a vacuum.
A natural by-product of this is heat.
(Actually, truth be told, a natural by-product of an incandescent bulb is *light*; most of the energy is wasted as heat.)
A Light Emitting Diode is based on the concept that when some semiconductor PN junctions (ie. diodes) are forward-biased, they convert the energy lost to their forward voltage drop into light.
LED light actually comes from the semiconductor junction itself; it's highly efficient (85% or more), highly color-stable (the color emitted depends on the doping of the junction), has none of the thermal inertia issues associated with tungsten filaments (ie., look at a car with tungsten tail lights and an LED third brake light), and is virtually impervious to mechanical shock.
The LED is the way of the future. Your fishbowl would benefit.
I'm not sure about stagelights, though. I used to work in the sound/lighting/professional video field; I don't know how I'll like stagelights that don't feel *warm* when they shine on you!
Although, I'd never have to dig out the asbestos gloves and climb a ladder to change a hot quartz bulb during the 10-minute intermission or fart around swapping gels...
Hey, is there a Linux version of the Color Kinetics software out there? I'm just wondering, when Windows blue-screens, do their cans change to blue in sympathy?
Either way, these guys are set to give Intellibeam, RoboScan, etc. a run for their money. Most of the times I ever used those, it wasn't for the tacky little gobos or the fact that they'd follow a target: it was because they changed color *quickly*, certainly faster than conventional cans with a gel reel setup.
Fire and Meat. Yummy.
I wonder if you arranged 800x600 of these things on a grid (assuming the price drops significantly) we could actually see LED projectors in the future! Sounds like a promising alternative to LCD, huh? Then maybe I can watch projected movies on my big loft wall without spending US $10,000.
Uhhh... No. If you made an array of LEDs and wanted to project them, you'd still need a lens.
To do it more reasonably, just make the 800x600 grid direct-viewable in whatever size that you want.
Problem 1: Average current to each LED is, let's say, 15mA. 800x600x0.015A = 7,200 amps. Evidently, this is going to have to be multiplexed somehow. (Oh, yeah, that's 7,200A *per color*, since I didn't multiply that number by three for each of the primary colors. (I assume you want each pixel to be three color.)
Assuming the average forward voltage drop per LED is 2.0V, that's a total consumption of 14.4kW. With the cost of a kWh of electricity where I am hovering around $0.06, watching the X-Files would cost me $8.64 in electricity. Per color! ("Maw, we's is goan' hafta winna lottery before we is can afford ta turn on th' whole TV set... 'ntil then, it's Jerry Springer 'n green only.") And that's just for the LEDs, not the support electronics.
Problem 2: Let's say these are three-color T-1 3/4 LEDs. And let's say that you've somehow figured out a way of wiring them to each other that doesn't occupy any display real estate.
A T-1 3/4 LED is about 1/4" in diameter. If you have 800 of them in a row, that's gonna be 200 inches long. 16.66 feet. 600 LEDs tall: 150 inches; 12.5 feet. A 16.6 x 12.5 foot screen. Or, using Pythagorean Theorum to figure out how they'd advertise it if it were a TV set, 20.78 feet diagonally. 249.36 Viewable Inches! On Sale Now! (Got space for it?)
LEDs vary from lot to lot, factory to factory. Blue on one side of your display with be different from the blue on the other side. At the same time, it'll be difficult to get your driver system to be adjusted flat across the entire color spectrum. You will have color purity issues. Look at a big LED display board in a public place now; few of them even approach the scope and resolution that you're talking about. And they're all spotty. There's a gorgeous example on the Paramount Theater at Toronto's Richmond and John Streets.
Finally, 800x600 is 480,000 LEDs. Each one is gonna cost you at least a buck (three color T-1 3/4), even in quantity. Each one is going to take you at least 30 seconds to solder into place (realistically, doing it by hand). Got 240,000 minutes (that's 4,000 hours; 166 days)? And we haven't even looked at the support electronics.
Just wait another 20 years or so until a stadium somewhere is being demolished, and scoop the JumboTron out of it. They're not LED-based, they don't have the same resolution, but they're pre-built and they look really good.
Fire and Meat. Yummy.