An interesting read...

http://gadgetwise.blogs.nytimes.com/2009/09/24/is-this-the-lightbulb-of-the-future/

With millions in subsidies, outlawing the competition, and a lot of good luck, they will get the price of a single light bulb down to $20?

Even at todays' rates, $19.45 buys a lot of electricity.

Thanks for the link. Interesting subject, nice to see the gov has $10 mil. for the prize!

Wonder how much the 'samples' sent to DOE cost to produce? And, how much will the street price be, IF this is the winner.

$25 for a 25k/hr bulb that uses 10 watts that saves 50 watts sure don't sound like a great ROI; I'll crunch the numbers later.

How's this sound....
$25 cost of LED
10 watts vs. 60 watts = 50 watt savings
25,000 hours life

Savings of 1,250,000 watts over lifetime

1250 KWHr @ .13 KWHr equals savings of $162.50 or $ 6.50 year
Approx 4 year payback on LED energy efficiency only.

Sound right??

Nope! They only give out half the light claimed by the makers, so that's 8 years. But they dim exponentially to 50% of the original output and the plastic lenses go yellow - make that 14 years. In the meantime, the price of electricity doubled because of lack of baseload and cost of all those stupid windmills that never generated enough revenue to pay the capital costs of building them. Make that 28 years.
By then, all your luminaires will be obsolete!

Once the Vietnamese start banging these lamps out for a buck each they will make sense.

If you burn them for 8 hours a day that would be 8.5 years.

Perhaps a step in the right direction?

All technology goes through a bit of growing pains. Initial costs, design misfires, etc. Remember, it took quite some time for compact fluorescents to make it to the mainstream and become cost effective.

We'll see how this evolves over time...

If these are dimmable, and actually last 25 years? I'd be happy to throw away my old CFLs, just for the sheer convenience factor. I think the 10W number is generous, too; given the efficiency of LEDs, it will probably draw far less than that.

The $5 1W LED light I put in my daughter's bedroom as a night-lite paid for itself in 6 months over the 15W bulb she was using before.

Their point of introduction will be in refrigerated spaces.

Their energy efficiency impact is leveraged in the cold.

BTW, typical fluorescents stink in commercial cold spaces: too many suffer from condensation via the raceway.

A canned NTL solution with LED's on the cold side and a low voltage DC quick-connect to the power supply at the warm side would be a run-away success.

I think you are right about LEDs in coolers. The outrageous cost of cold start ballasts will push this along.

Walmart and other grocery chains have used LEDs as lighting in new refrigerators for well over a year now...

Biggest issue is reliability of the led driver circuit.
These are usually SMPS and will not last the time.

Led's have claims of 100k hours + but when driven hard by the PSU the colour will fade over time and brigtness is reduced.

For cold environments they are excellent provided they are fed from a low voltage supply and the SMPS is not required.
Of course the unit has to be properly sealed agains water ingress, otherwise the led terminals will corrode and an ealry faillure is iminent.

I have built quite a few LED night lights around my house and they run at 10 - 20 mA current at mains voltages with a simple series resistor to control the current.
The reliability is excellent.

I don't see where Switched Mode Power Supplies make much sense for LEDs.

Their limited power draw would seem to indicate that a simple low pass filter with a hefty cap would be sufficient. A tiny transformer, embedded in epoxy, would be able to kick out the joules. If this power supply is properly configured to tolerate plenty of heat... you'd be good to go.

Since a LED drops 2 - 2.5v across its junction I would think you could string 55 or so in series with a small value resistor to scrub off the rest and limit the current although I bet there is a chip that would do a better job of current regulation for a nickel a unit or less in the quantity.
I am not sure what the new white LEDs need to get to max illumination but I am guessing it is in the 20ma range.
(~24 watts)

LEDs should be operated at a constant current, not at a constant voltage. The IV curve is exponential, so a resistor, especially off an unregulated source, is a poor choice.

I just used Supertex's CL2N8-G as a 20mA constant current source in some backlit signs I just made. For higher current, you can use something like an LM317 but power dissipation will start to add up.

For higher power (350mA to 1A LEDs), I have used switching designs.
/mike