Cree 231 lumens per watt !

Perhaps not yet in production, but that is amazing for the bar to be raised by that increment.
http://technews.tmcnet.com/news/2011/05/09/5496319.htm

Just to be clear, is the article stating:

-it is a single die?
-it is around 4500K?
-its forward voltage is about 2.86?
-it is being powered at only 350mA?

2.86 volts * .350mA = 1.001 watts

Holy cow!

That's nuts!

When can I buy the warm white and high CRI version? :twothumbs

Amazing, and honestly not a whole lot of room for improvement at this point ( at least for standard blue plus phosphor LEDs ). I never thought we would reach this level by 2011. I had figured by 2015, or more realistically 2020, you might finally hit close to 250 lm/W in the lab.

I would certainly like to find more detail. Like how are they getting the increase? Better out-coupling with the phosphor layer or have they developed their own high RI encapsulant or is it the die itself? Does "standard room tempurature testing" imply 25C instead of the 85C being adopted by most producers? If they can translate those gains into product in the near future then that would be incredible! I would like to see a 6500 K four die myself.

This seems very promising, 350ma at room temp. The future looks bright.

Harold_B said:

Does "standard room tempurature testing" imply 25C instead of the 85C being adopted by most producers? If they can translate those gains into product in the near future then that would be incredible! I would like to see a 6500 K four die myself.

Click to expand...

I believe by standard room temperature testing they might be running the LED on a large enough heat sink so the base plate is more or less at room temperature. Obviously the die will be somewhat warmer. I also think as LEDs become more efficient the 85°C being adopted by some producers will be less and less relevant. An LED only heats up because it's less than 100% efficient. At 231 lumens per watt, I'd estimate you're getting at least two-thirds light and only one-third heat. A 100-watt replacement using such an LED would only use about about 7.5 watts (plus ballast losses which might be another 10%), and produce around 2.5 watts heat. Difficult to imagine any part of the lamp, even the LED die itself, reaching 85°C even in a totally-enclosed fixture.

I hope they write a paper about this like last year.

The blue base diode needs to have at least 85% wall-plug efficiency for something like that. Incredible.

This also changes the cooling paradigma: More heat is generated due to stokes losses in the phosphor compared to the the die itself... with obviously worse thermal pathways.

The need to test at 85C is always relevant (in my opinion) since there are a lot of areas in the world where ambient is greater than 25C. I'm thinking about applications like street lights in arid locations not just a replacement light bulb.

IMSabbel said:

I hope they write a paper about this like last year.

The blue base diode needs to have at least 85% wall-plug efficiency for something like that. Incredible.

This also changes the cooling paradigma: More heat is generated due to stokes losses in the phosphor compared to the the die itself... with obviously worse thermal pathways.

Click to expand...

That actually makes a good case for remote phosphor which spreads the heat over a relatively large surface area. This makes cooling the die almost trivial. If we use my earlier hypothetical 100 watt replacement for an example, you have ~1.4 watts Stokes losses (dissipated easily over the globe by using remote phosphor), and a mere ~1.1 watts at the LED die. Even with a small heat sink, the die will practically be at ambient ( at most 10°C above).

LEDAdd1ct said:

Just to be clear, is the article stating:

-it is a single die?
-it is around 4500K?
-its forward voltage is about 2.86?
-it is being powered at only 350mA?

2.86 volts * .350mA = 1.001 watts

Holy cow!

That's nuts!

When can I buy the warm white and high CRI version? :twothumbs

Click to expand...

what is even more nuts is with these they could make direct drive 2AA alkaline battery lights that put out over 231 lumens that would stomp current 2AA boosted lights in stores. The days of the 2D cheap plastic flashlights could be back again (with LEDs in them instead of incan bulbs). Essentially when these efficient emitters become mainstream the days of cheap incan flashlights could be numbered.

There of course remains the minor challenge of surface brightness. General lighting applications are not very sensitive to this, but flashlights certainly are. At low current densities, surface brightness is going to be low regardless of the technology.

HarryN said:

There of course remains the minor challenge of surface brightness. General lighting applications are not very sensitive to this, but flashlights certainly are. At low current densities, surface brightness is going to be low regardless of the technology.

Click to expand...

Lumens per mm^2 is lumens per mm^2 regardless of the current density. If they did this on a 1mm^2 die, then the surface brightness of this at 350mA would be similar to an XR-E at 1A, regardless of current density.

That being said, bring on the highly efficient dies in good themal packaging so we can drive the **** out of them.

I was poking around on a circuit board today , using a D11 as it was a bit dark in there and i noticed for the first time actually , that when i was looking at the bands to read the values of resistors (old style) those colours did not show up right and had to swap to a incan .

Just a thought on these Leds and colour rendition?

Paul

HooNz said:

Just a thought on these Leds and colour rendition?
Paul

Click to expand...

Google high CRI Paul.
Norm

HooNz said:

I was poking around on a circuit board today , using a D11 as it was a bit dark in there and i noticed for the first time actually , that when i was looking at the bands to read the values of resistors (old style) those colours did not show up right and had to swap to a incan .

Just a thought on these Leds and colour rendition?

Paul

Click to expand...

Dont be so picky and expect a lab experiment to have a 105 CRI, 2700k CCT, with the die at 100C? :nana:

On a serious note though, Its all a just progress. They release an LED (like the XM-L since that just came out) but only cool white is available, they come up with a happy phosphor receipe for neutral and warm whites, (like where we are with the XM-Ls now) and then eventually high CRI whites.

But XP-Gs with high CRIs are available, but manufacturers dont put those in flashlights, or neutral or warm LEDs since that doesnt sell like CREE R5 OVER9000LUMENS. But as you noted, CRI is definitely vital in a lot of applications.

HooNz said:

I was poking around on a circuit board today , using a D11 as it was a bit dark in there and i noticed for the first time actually , that when i was looking at the bands to read the values of resistors (old style) those colours did not show up right and had to swap to a incan .

Just a thought on these Leds and colour rendition?

Paul

Click to expand...

Honestly, I find cool white LEDs make it MUCH easier to read the white lettering on black background present on most integrated circuits. As for resistors, incans might offer superior ability to distinguish between red and orange bands (and sometimes they pick colors which are really close), but it comes at the expense of being able to distinguish violet and blue from each other, or even both from black. For resistors then I think something like a 4000-5000K high-CRI LED might be ideal. Nowadays though I mostly use surface mount resistors which have the value numerically marked on them. Much easier than the color band system.

I want these LEDs... now... Then I'm going to make bulbs for the entire house and save a fortune on my lighting bill!!!
WOW... 231 Lumens per watt... I'm salivating over here... XM-Ls are going to be obsolete soon! Unless these new LEDs
are like 60mm in diameter... There's not a lot of data there... Just some Cree representative that COULD be talking out
his arse.

Shao

4500K?!

That's basically daylight color. Provided of course they are an even white and not rose, purple, or green.

This is a fantastic achivement by Cree, and of course, not unexpected. LEDs seem to follow Moore's Law as well as their transistor-based silicon cousins. I am surprised, though, that this was achieved using a 4500K LED rather than a less useful, less pleasing 6000K LED.

jellydonut said:

I am surprised, though, that this was achieved using a 4500K LED rather than a less useful, less pleasing 6000K LED.

Click to expand...

I had the same exact thought!


If we assumed for the sake of play that Cree used the same letter/number efficiency scheme they use now, like R2, R3, R4, R5, S2, etc., where would 231 lumen per watt LEDs sit?

If we use the XR-e or XP-G binning at 350mA, it would be a U2 bin.
edit: This differs from the XM-L bins because those are done at 700mA, this is a ~7 bin or so jump from the current production items, so if we jump that far in the XML range it would be the equivalent of a V3 bin.