LED Droop IEEE Article

[Genes]

Over the weekend I received my copy of the IEEE Spectrum. This month there is a nice article on LED droop at high power levels. For those of you who don't get this publication, here is a link to the article.

http://spectrum.ieee.org/semiconductors/optoelectronics/the-leds-dark-secret

Peace,

Gene

 

[bshanahan14rulz]

printed for a good read later. Thanks!

 

[jtr1962]

Great article. I stumbled upon it about a week ago. Minor factual error in the article:

The current state-of-the-art white LED pumps out around 250 lm/W, and there's no reason why that figure won't reach 300 lm/W.

If only that were true.

 

[cleonard]

Great article. I stumbled upon it about a week ago. Minor factual error in the article:

The current state-of-the-art white LED pumps out around 250 lm/W, and there's no reason why that figure won't reach 300 lm/W.

If only that were true.

If you keep reading they go on to state this is only at low current levels. They say that at the higher currents the efficiency drops to the approximately 100 lm/W that we are all used to seeing.

 

[jtr1962]

If you keep reading they go on to state this is only at low current levels. They say that at the higher currents the efficiency drops to the approximately 100 lm/W that we are all used to seeing.

The best production LEDs right now are only managing around 150 lm/W at milliamp current levels. I've read about some laboratory prototypes from Nichia which managed around 190 lm/W at 3 or 4 mA, but nothing approaching 250 lm/W so far.

 

[saabluster]

The best production LEDs right now are only managing around 150 lm/W at milliamp current levels. I've read about some laboratory prototypes from Nichia which managed around 190 lm/W at 3 or 4 mA, but nothing approaching 250 lm/W so far.

He did state "state of the art" not production LEDs. He may know what is capable in the lab right now but has not been released formally. It is not beyond possibility but I too thought that was awfully high.

 

[jtr1962]

He did state "state of the art" not production LEDs. He may know what is capable in the lab right now but has not been released formally. It is not beyond possibility but I too thought that was awfully high.

It seems the author of the article responded to my comments about the 250 lm/W efficiency figure:

"The 250 lm/W figure was given by Nichia at this year's Photonic West show in January. It's for a lab device."

All I can say is that this is amazing news. Maximum theoretical for blue plus phosphor whites is a bit under 270 lm/W, and that assumes a 100% efficient blue LED. So 250 lm/W implies they have blues with 93% efficiency. If we could make comparable reds and greens, when we could have 370 lm/W RGB whites (and at 93% efficiency heat is a non-issue, even for sources generating a few thousand lumens). The fact that they've reached 250 lm/W means there are no major barriers to doubling LED efficiency yet again. We just need to put the lab work on the production line (admittedly not an easy process).

 

[saabluster]

It seems the author of the article responded to my comments about the 250 lm/W efficiency figure:

"The 250 lm/W figure was given by Nichia at this year's Photonic West show in January. It's for a lab device."

All I can say is that this is amazing news. Maximum theoretical for blue plus phosphor whites is a bit under 270 lm/W, and that assumes a 100% efficient blue LED. So 250 lm/W implies they have blues with 93% efficiency. If we could make comparable reds and greens, when we could have 370 lm/W RGB whites (and at 93% efficiency heat is a non-issue, even for sources generating a few thousand lumens). The fact that they've reached 250 lm/W means there are no major barriers to doubling LED efficiency yet again. We just need to put the lab work on the production line (admittedly not an easy process).

Thanks for the update. That is tremendously exciting news! I wish they would offer some for sale on the market even if the price was astronomically high. I would buy one.

 

[lolzertank]

If anyone (waves their magic wand and) gets their hands on one and it turns out to be unbelievably expensive/hard to get, can we have a passaround? We can make the rules so that you can only keep it for an hour, which should be more than enough to blind yourself.

 

[jtr1962]

More about this development here. You need to sign up to view the entire article but here are the more relevant parts:

Researchers at Nichia have grabbed the attention of their LED industry peers by cranking up the efficacy of their white emitters to 249 lm/W at 20 mA.

However, the results reported at Photonics West on January 29 show that this falls to 145 lm/W at 350 mA, the drive current often used to compare the output of today's high brightness LEDs.

This decline in efficacy with current is incredibly steep, and Nichia's Takashi Mukai told delegates that it might be due to a mistake in the process.

Mukai initially introduced the result by stating that the theoretical efficacy limit for white LEDs is 263 lm/W. Yet, when questioned by the audience he explained that Nichia has pushed this ceiling to over 300 lm/W thanks to the introduction of a new phosphor technology.

 

[defloyd77]

249 lm/W at 20 mA, how bright would this be? That would make one heck of a Fenix E01 type light upgrade!

 

[PhantomPhoton]

249 lm/W at 20 mA, how bright would this be? That would make one heck of a Fenix E01 type light upgrade!

All depends on the Vf of the leds. If we were to throw out conservative guess of ~3V @ 20mA (.06W) then we get almost 15 lumens as a guess.

 

[defloyd77]

All depends on the Vf of the leds. If we were to throw out conservative guess of ~3V @ 20mA (.06W) then we get almost 15 lumens as a guess.

Ok thanks, that's what I assumed it'd be around.