LED Predictions From 15 Years Ago

iamlucky13

Flashlight Enthusiast
Joined
Oct 11, 2016
Messages
1,139
Searching for old performance specs for the sake of a discussion elsewhere, I came across the following:
http://www.theledlight.com.cn/306-knowledge-ssl-led-roadmap-recommendations.html

It details a development roadmap published by Cree sometime around 2002. I knew it would be interesting to see how the predictions compared to reality, and was intrigued to note they actually came fairly close.

For example, regarding efficacy:

2002 Actual - 25 lm/W in 2002 (not much better than good halogen lights, and terrible compared to fluorescents)
2007 Target - 75 lm/W (not quite sure if they hit that)
2012 Target - 150 lm/W (I think the best emitters were just about there under ideal conditions)
2020 Target - 200 lm/W

Cost per 1000 lumens (a little bit more light than a 60W bulb):

2002 Actual - $200
2007 Target - $20
2012 Target - Less than $5
2020 Target - Less than $2 (I've already seen 800 lumen bulbs for $2)

Practically speaking, it seems they've been running just a little behind the targets, but they're pulling ahead on the cost per lumen figure. And these targets were for just the emitters, not complete lamps.

Hope you enjoy the look back and comparison to how far things have come.
 

degarb

Flashlight Enthusiast
Joined
Oct 27, 2007
Messages
2,036
Location
Akron, Ohio
I thought Cree hit 80 lpw in 2006. But got my first in 2008. Luxeon rebel 80 and 100, I mostly used in 2008 and 2009. The lpw rating dropped when everything went ansi, around 2010, yet the lights actually became brighter. ... My 2005, Luxeon 1 lights, measure now about 29 lumens at 1 watt. However, remember seeing press release in 2003 or 2004 that they hit 40 lpw.... I never could stand the rings of the xre. So, I preferred the rebel, until 2011, when I tried the xpe. However, i tried the xpg r3 within 3 months of the xpe, and never looked back. I recall the xml 156 lpw, in 2011ish. Stagnation since, other than color, minor bumps, increase in lpw of small die, huge current leds. But mostly, Chinese knock off, lowering of lpw, for sake of cost... The 202 lpw neutral xpl v6 was first to hit 202 lpw, probably 2 years ago, though hard to find them even today. The obsession today, seems to be getting the deep blood red rendering (at any low lpw or reduction in rendering of other colors) , cost-which is ok with me as long as it doesn't delay an acceptable 240 to 300 lpw led that Cree hit in lab years and years ago. Since Cree "reorganized", I have not noticed their old innovation and leadership. Just stock market babble. The problem is that most executives believe that stealing ideas is better than creating new ideas: It is cheaper and have a bigger pool to choose from. A real innovation killer.
 
Last edited:

ssanasisredna

Enlightened
Joined
Oct 19, 2016
Messages
457
Searching for old performance specs for the sake of a discussion elsewhere, I came across the following:
http://www.theledlight.com.cn/306-knowledge-ssl-led-roadmap-recommendations.html

It details a development roadmap published by Cree sometime around 2002. I knew it would be interesting to see how the predictions compared to reality, and was intrigued to note they actually came fairly close.

For example, regarding efficacy:

2002 Actual - 25 lm/W in 2002 (not much better than good halogen lights, and terrible compared to fluorescents)
2007 Target - 75 lm/W (not quite sure if they hit that)
2012 Target - 150 lm/W (I think the best emitters were just about there under ideal conditions)
2020 Target - 200 lm/W

Cost per 1000 lumens (a little bit more light than a 60W bulb):

2002 Actual - $200
2007 Target - $20
2012 Target - Less than $5
2020 Target - Less than $2 (I've already seen 800 lumen bulbs for $2)

Practically speaking, it seems they've been running just a little behind the targets, but they're pulling ahead on the cost per lumen figure. And these targets were for just the emitters, not complete lamps.

Hope you enjoy the look back and comparison to how far things have come.

They were really off w.r.t. the precipitous price drops.... same as with solar.

At the emitter side, 2000-3000 lumens/$ is possible in high volume (and maybe even more). At the emitter side, 200 lumens/watt is now possible in production at a reasonable cost. 1000 lumens/$ and 170+ lumens/watt is quite doable ..... maybe not in flashlight and similar emitters, but they are not where the volume is for LEDs.
 

iamlucky13

Flashlight Enthusiast
Joined
Oct 11, 2016
Messages
1,139
They probably looked at Haitz's Law and wrote the roadmap around that. https://en.wikipedia.org/wiki/Haitz's_law

Maybe. The timeline seems fairly close between Haitz publishing his forecast while at Agilent in 2000 according to Wikipedia and Cree publishing theirs in 2002. It could have been their own work, or just repeating what Haitz said I suppose.

They were really off w.r.t. the precipitous price drops.... same as with solar.

At the emitter side, 2000-3000 lumens/$ is possible in high volume (and maybe even more). At the emitter side, 200 lumens/watt is now possible in production at a reasonable cost. 1000 lumens/$ and 170+ lumens/watt is quite doable ..... maybe not in flashlight and similar emitters, but they are not where the volume is for LEDs.

Good point about emitter specs, rather than lamp specs. I wouldn't say way off, though. $2 per 1000 lumens forecast for a few more years is closer to the current point than the 10 year ago point for the forecast of $20 per 1000 lumens.

Most notably, since they were no doubt trying to make this forecast to decide on the business case for further investments in improving LED's, it seems significant that they were able to make improvements faster than their forecast.
 

degarb

Flashlight Enthusiast
Joined
Oct 27, 2007
Messages
2,036
Location
Akron, Ohio
I think they mistakenly used the lesser than 80 cri for their last two led cri predictions. Surely, they weren't expecting the cri to drop from 2007.
 

Hugh Johnson

Enlightened
Joined
Jan 15, 2017
Messages
334
This makes me wonder what we'll have decades from now. Maybe 10x the efficiency with a new technology. We'll have 10,000 lumen thermally efficient pocketable lights with a couple of hours of run time on turbo and a single battery.
 

iamlucky13

Flashlight Enthusiast
Joined
Oct 11, 2016
Messages
1,139
I think they mistakenly used the lesser than 80 cri for their last two led cri predictions. Surely, they weren't expecting the cri to drop from 2007.

They must have just used a less than sign mistakenly instead of a greater than sign.

This makes me wonder what we'll have decades from now. Maybe 10x the efficiency with a new technology. We'll have 10,000 lumen thermally efficient pocketable lights with a couple of hours of run time on turbo and a single battery.

10x efficiency isn't possible. For a green light at the eye's maximum sensitivity, the maximum possible efficacy is 683 lumens/W. That's 100% efficiency.

For white light, the best possible efficacy is lower, depending on the exact characteristics of the light. For a high CRI light, somewhere around 300 lumens/Watt is possible. That's about double what some of the most efficient flashlights currently achieve in their lower modes.

As efficacy improves towards that maximum, however, the heat production declines significantly, which makes it easier to maintain high lumen outputs without overheating.
 
Last edited:

Hugh Johnson

Enlightened
Joined
Jan 15, 2017
Messages
334
Interesting to know, iamlucky13.
Heat is a big factor for me. I'd love to run 1,000+ lumens in a pocketable 18650 light with a roughly 25mm head.
 

degarb

Flashlight Enthusiast
Joined
Oct 27, 2007
Messages
2,036
Location
Akron, Ohio
For a green light at the eye's maximum sensitivity, the maximum possible efficacy is 683 lumens/W. That's 100% efficiency.


So, what is the lpw of the most efficient Green LED in 2017? Who makes it and what fl incorporate it?
 

The_Driver

Flashlight Enthusiast
Joined
Jan 6, 2010
Messages
1,177
Location
Germany
So, what is the lpw of the most efficient Green LED in 2017? Who makes it and what fl incorporate it?


Green LEDs are actually less efficient compared to white LEDs. Check out the XP-E2 for actual numbers. The best green XP-E2 bin (Q2) only puts out 67% of the lumens of a standard cool-white Xp-E2 in the highest bin (R4). The XP-E2 Torch has the U5 Bin which is even more efficient. If you consider how much of an efficiency advantage the green LED has because of it's wavelength, you will realize that it's really not that great.

Blue and red LEDs are much better when you consider their real efficiency in Watt of output per Watt of input. White LEDs are better than blue LEDs in terms of lm/W because they convert the blue light into mostly yellow-green light with a very high efficiency (around 95%) and the human eye is, as has been mentioned here, much more sensitive to these wavelengths. Red LEDs benefit from their much lower Vf.
 
Last edited:

ssanasisredna

Enlightened
Joined
Oct 19, 2016
Messages
457
White LEDs are better than blue LEDs in terms of lm/W because they convert the blue light into mostly yellow-green light with a very high efficiency (around 95%)

Way less than 95% efficiency. The phosphors are lucky to have a quantum efficiency of 1, meaning 1 photon in, one photon out, but since the wavelength is much longer of the outgoing photon, there is a loss of energy. Practical phosphors can max out with a QE in the high 0.90's, about 0.97.

You would be lucky to get 70-75% efficiency energy conversion efficiency with reasonable CRI for a white LED phosphor.
 
Last edited:

The_Driver

Flashlight Enthusiast
Joined
Jan 6, 2010
Messages
1,177
Location
Germany
I was talking about lumens per watt though, so a unit that has been adjusted for the human eyes sensitivity to different wavelengths. Converting blue light (the eye has a low sensitivity to this) to green light with a high efficiency should result in higher lumens/watt because the eye is so much more sensitive to green light.
 

ssanasisredna

Enlightened
Joined
Oct 19, 2016
Messages
457
I was talking about lumens per watt though, so a unit that has been adjusted for the human eyes sensitivity to different wavelengths. Converting blue light (the eye has a low sensitivity to this) to green light with a high efficiency should result in higher lumens/watt because the eye is so much more sensitive to green light.

I think you were quoting quantum efficiency (95%), which is not the same as power efficiency and is also completely at odds with Lumens/Watt.
 

degarb

Flashlight Enthusiast
Joined
Oct 27, 2007
Messages
2,036
Location
Akron, Ohio
https://www.digikey.com/en/articles...t-addresses-lower-efficacy-of-warm-white-leds


Too busy to read carefully, but hopefully relevant.

It also explains that phosphor conversion was so understood and widely used in color crt tvs by 1993, that even if Nakamura was first to phosphorize his blue (probably done on earlier blues), it would not be notable.

Observed for hundreds of years, mid 19th century studied in more depth by scientists.
 
Last edited:
Top