Help Support Candle Power Flashlight Forum
jtr1962 said:
As you surmised it depends upon how full the spectrum is. For something like sunlight I think it would only be in the 200 lm/W area. However, you could leave out some of the extremes at either side of the spectrum, get a CRI of about 98, and arrive at an efficiency of about 300 lm/W. If you're willing to tolerate a CRI of about 80 (OK for streetlighting), you can go as high as 400 lm/W. The luminous efficacy of the spectrum of a typical blue plus YAG phosphor white LED is about 330 lm/W, meaning that Nichia's latest accomplishment really only manages to convert about 30% of the input power into light. That's still not bad, and it's about as good as we've been able to produce white light using other technologies. The 150 lm/W which they hope to arrive at next year will put us into uncharted territory. We've never had a white light source offering 150 lm/W efficiency. Some fluorescents and the larger stadium HIDs give in the 110 lm/W area but that's about it. Moreover, what I love about LEDs is that they're scalable. Fluorescents and HIDs scale poorly. A 5W HID only offers about 50 lm/W efficiency. 1 or 2 watt fluorescents are in the 30 to 40 lm/W area at best. Today's LEDs offer something which has literally never existed until now-a sub 1 watt light source offering efficiencies far greater than the 10 to 15 lm/W of incandescents. Better yet, the forthcoming 100 lm/W LEDs will offer HID-like efficiencies yet scale to the milliwatt range. You can't help but love that!monkeyboy said:
I measured the efficiency of some U-bin CS samples at an average of 75 lm/W. Based on the measured mcd though they looked to be right at the upper edge of the U-bin. Typical U-bin would likely be in the 60 to 65 lm/W area. As for V-bin, I haven't obtained any to date but based on the relative intensities I'd say probably 80 lm/W minimum, and 85 lm/W average. The upcoming Nichia 85 lm/W LEDs slated to be released in April will likely mean that V-bins will become more easily obtainable, and maybe there will even be a rare W-bin (assuming that Nichia adds a new bin instead of revising the existing bins). I also really wish Nichia would offer LEDs for sale by bin. I'm sure most people would be willing to pay a small premium to obtain a U-bin as opposed to a T now, and a V-bin as opposed to a U after April.jsr said:
(but many won't like it because it's still too blue). 
They did say we'll have production 5mm LEDs which deliver 100 lm/W at 20 mA at normal operating temperatures late this year, and 85 lm/W ones in April. And I'm reasonably sure we'll have 100 lm/W LEDs available on eBay by the fall, if not sooner. You can buy ~80 lm/W LEDs right now by ordering the Jeled 50Ks. If you run them at 5 mA you can even get efficiencies in the 100 to 110 lm/W area. None of this is experimental. These LEDs are available now or will be in the stated time frame. What is experimental is that Nichia managed to hit 113 lm/W a month or two ago, presumably at 20 mA. No doubt we'll have production LEDs with that efficiency or better sometime in 2007.offroadcmpr said:
That's because of the tapered shape Nichia uses. Note that the light falling outside the 60° cone you mentioned can still be used if you have a small reflector. Also, when using these LEDs as area lights in things like solar garden lights virtually all their light is used, including the light coming out the back.Pinter said:
There are probably better shapes then even Nichia's tapered one, but they might not be practical to manufacture. You could likely project 90% of the chip's light into a narrow forward cone with the proper optics.
