Cree does it again-160 lumens per watt @350 mA

[tstartrekdude]

Placing XP-G die in a different package wouldn't increase the efficiency at 350mA to 160lm/W IMHO. Must be a different die.

I was speaking more to size than the die its self, yes they are making a bit of a leap forward in efficiency, still they state 2 amps as a very high drive current, so i can't really see it being bigger than 2mm^2

 

[jirik_cz]

Maybe the maximum current will be higher...

 

[Gryloc]

I just wanted to post to share my thoughts. I would not doubt if the die is from 4-6mm^2 just because it would take quite a breakthrough to do all this in 1mm^2 or probably 2mm^2. So, I did what I did previously when they announced the XP-G: I tried to make it relatable to current technology. I did this by scaling the performance of existing LED emitters to see if it is similar to the new technology.

Therefore I took the sample measurements of jtr1962's XP-G R5-bin and scaled them to different equivalent die sizes. I scaled the dies to sizes ranging from 3mm^2 to 6mm^2 (and I tossed in a 1mm^2 scaling for the heck of it). One thing that I did not do is take the lumen output of my simulated dies at 350mA or 2000mA and scale them up to match the announced lumen output levels from the press release.

First off, I want to apologize for using a free file hosting site. After 10 downloads each they will expire but I will try to re-host it and edit this post if they do expire. I forgot that the scaled tables are on "Sheet 2" of the excel sheet. (links removed - see post #61 for links to the updated file)

To my eyes, from this scaling/extrapolation, I think that the die is either 4mm^2 or 5mm^2 (like what WeLight said). I can imagine that just like when the XP-G was developed, they tossed in a few enhancements to the technology to make it more efficient than just making a 2mm^2 die with the same surface brightness of a XP-E R2-bin die (which scaling the XP-E up was very close to the performance of the XP-G). So I can imagine that some new technologies are being used in the making of the XM to make it more efficient at the lower current levels. However, I believe that the droop reduction is mostly due to the scaling of the die size. I did not do any of those droop calculations like what jtr1962 does, though. I may play with the numbers later to further scale up/down the charts to match the performance levels of the XM press release. I did this with the simulated XP-G charts from months ago and my simulations were extremely similar to the XP-G R5-bin sample data when jtr1962 did his testing.

I hope this stirs the conversation more...

Cheers,
-Tony

EDIT - I removed the links above because I had quite a few formulation errors in them and I was not happy with that. Eck! Please discard your older file and refer to the newer post for the most updated file. Thanks!

 

[saabluster]

I just wanted to post to share my thoughts. I would not doubt if the die is from 4-6mm^2 just because it would take quite a breakthrough to do all this in 1mm^2 or probably 2mm^2. So, I did what I did previously when they announced the XP-G: I tried to make it relatable to current technology. I did this by scaling the performance of existing LED emitters to see if it is similar to the new technology.

Therefore I took the sample measurements of jtr1962's XP-G R5-bin and scaled them to different equivalent die sizes. I scaled the dies to sizes ranging from 3mm^2 to 6mm^2 (and I tossed in a 1mm^2 scaling for the heck of it). One thing that I did not do is take the lumen output of my simulated dies at 350mA or 2000mA and scale them up to match the announced lumen output levels from the press release.

First off, I want to apologize for using a free file hosting site. After 10 downloads each they will expire but I will try to re-host it and edit this post if they do expire. I forgot that the scaled tables are on "Sheet 2" of the excel sheet. (Office 07 excel sheet) (Office 97-03 excel sheet)

To my eyes, from this scaling/extrapolation, I think that the die is either 4mm^2 or 5mm^2 (like what WeLight said). I can imagine that just like when the XP-G was developed, they tossed in a few enhancements to the technology to make it more efficient than just making a 2mm^2 die with the same surface brightness of a XP-E R2-bin die (which scaling the XP-E up was very close to the performance of the XP-G). So I can imagine that some new technologies are being used in the making of the XM to make it more efficient at the lower current levels. However, I believe that the droop reduction is mostly due to the scaling of the die size. I did not do any of those droop calculations like what jtr1962 does, though. I may play with the numbers later to further scale up/down the charts to match the performance levels of the XM press release. I did this with the simulated XP-G charts from months ago and my simulations were extremely similar to the XP-G R5-bin sample data when jtr1962 did his testing.

I hope this stirs the conversation more...

Cheers,
-Tony

Incredible work! I think Welight was right. Your data seems to back that up. I don't know if that was a slip on his part or not but I am glad to see Cree has better thermal characteristics than the same sized Phlatlight. Means we should be able to run the Cree harder than the SST-50.

 

[Tally-ho]

So I can imagine that some new technologies are being used in the making of the XM to make it more efficient at the lower current levels.

I found this news in dealextreme flashlight forum:

Green LEDs for Efficient Lighting (link)
Solar-cell manufacturing techniques could yield LEDs that require 20 percent less energy.

...
LEDs, devices that emit photons when an electrical charge is applied to them, are more efficient and last longer than incandescent lightbulbs. By varying the composition of the semiconductor LEDs, materials scientists can coax the devices into emitting different colors. At the minimum, producing white light requires combining red, blue, and green, but so far, only red- and blue-light-emitting diodes are well developed. To produce green light, LED manufacturers typically apply one or more phosphor materials to blue LEDs. The phospors convert high energy blue spectrum light into lower-energy light through a process that reduces overall luminosity by approximately 20 percent.

To eliminate this loss of efficiency, researchers have tried to develop efficient green LEDs that don't require phosphors. But a major stumbling block is that the different known semiconductor materials that can be combined to emit green light, typically indium and gallium nitride, have different-sized crystal lattice structures. For semiconductors to work efficiently, each layer of the device has to have a similarly sized lattice structure as the layer above or below it.

To get around the lattice-size mismatch, NREL researchers used a fabrication method that they had previously developed for building highly efficient multi-junction solar cells. Their method relies on using additional layers of other semiconducting materials with intermediate-sized lattice structures that bridge the gap between the disparate-sized semiconductors. "If you try to do it in one shot, the whole thing will be defective," says Angelo Mascarenhas, team leader for solid state spectroscopy in the Center for Basic Sciences at NREL. "You have to grow a sequence of layers in a step-wise fashion."
...

 

[saabluster]

LEDs, devices that emit photons when an electrical charge is applied to them, are more efficient and last longer than incandescent lightbulbs. By varying the composition of the semiconductor LEDs, materials scientists can coax the devices into emitting different colors. At the minimum, producing white light requires combining red, blue, and green, but so far, only red- and blue-light-emitting diodes are well developed. To produce green light, LED manufacturers typically apply one or more phosphor materials to blue LEDs. The phospors convert high energy blue spectrum light into lower-energy light through a process that reduces overall luminosity by approximately 20 percent.

To eliminate this loss of efficiency, researchers have tried to develop efficient green LEDs that don't require phosphors. But a major stumbling block is that the different known semiconductor materials that can be combined to emit green light, typically indium and gallium nitride, have different-sized crystal lattice structures. For semiconductors to work efficiently, each layer of the device has to have a similarly sized lattice structure as the layer above or below it.

http://www.technologyreview.com/energy/25028/?a=f

Thanks for the info but it does not belong in this thread. Try making a new thread for that.:thumbsup:

 

[Tally-ho]

Thanks for the info but it does not belong in this thread.

Read it again and you will probably understand why I put this here. I do not pretend that CREE use this to improve efficiency of is new LED, but it could be a clue.

 

[blasterman]

Diatribe from the press release:

The LED also delivers 750 lumens at 2 A, which is equivalent to the light output of a 60 W incandescent light bulb at less than 7 watts.

Notice we're always comparing ugly cool white emitters (with the same CRI as a circa 1978 K-mart brand fluorecent tube) to incan light bulbs. Maybe Cree should realize not everybody who reads their press releases is a moron.

Another option is to put 5 of those LEDs in a bulb and drive them at 5 watts total for 800 lumens

I'd rather just use a single 800lumen LED that's available now and not obsess if it's made by Cree or not.

Question: As I understand the technological constraints of LED technology, current density causes a huge problem with quantum efficiency and the larger die is just a natural progression to solve this?

 

[znomit]

Read it again and you will probably understand why I put this here. I do not pretend that CREE use this to improve efficiency of is new LED, but it could be a clue.

Read the earlier posts by myself and Gryloc. Theres no huge breakthrough here, just using an XPG S2 die thats twice the size.

 

[TorchBoy]

To my eyes, from this scaling/extrapolation, I think that the die is either 4mm^2 or 5mm^2 (like what WeLight said).

Incredible work! I think Welight was right.

That may have been what he meant but what he actually said was 5mm square, not 5mm^2. It's very hard to believe it would be 25mm^2.

 

[saabluster]

That may have been what he meant but what he actually said was 5mm square, not 5mm^2. It's very hard to believe it would be 25mm^2.

I know but I think we can use reason and come to the right conclusion here.

 

[Odysseus]

I have to agree about the Neutral White. Haven't seen XP-Gs in Neutral or Warm White. Want to see these, too!:thumbsup:

 

[znomit]

I know but I think we can use reason and come to the right conclusion here.

Imagine what current a 5x5mm die could hold....

But assuming the die is actually a reasonable 5mm^2... isn't 2A a little low for a max current (though perhaps reasonable for a fixed lighting application with passive cooling)? The XPG has a current density of 750mA/mm^2 so a 5 sq mm die should handle over 3A.
1000lm at 3A maybe?

 

[saabluster]

Imagine what current a 5x5mm die could hold....

But assuming the die is actually a reasonable 5mm^2... isn't 2A a little low for a max current (though perhaps reasonable for a fixed lighting application with passive cooling)? The XPG has a current density of 750mA/mm^2 so a 5 sq mm die should handle over 3A.
1000lm at 3A maybe?

Who ever said that was the max spec? Cree is conservative with things like this. Especially this early in the game. Remember when they announced the XP-G? They said nothing about an R5 bin or 1.5A capability. All that came later. They are just teasing us with that 2A data point. They don't want too much info available until the product is on the market. It excites me that they seem to have done some work on the thermal side of things. Now if we could just get them to bring out high CRI versions of these new LEDs.

 

[TorchBoy]

As saabluster says, even if it is the max spec now, the XR-E and XP-R went from 700 mA to 1 A after release, while the XP-G went from 1 A to 1.5 A, again retroactively. But it would depend more on how well the XM package conducts heat away from the die.

 

[John_Galt]

I wonder how well it can handle being over-driven. Considering that a 5mm^2 die would be edging into Luminus' market share, I think Cree is probably going for smaller than that. 3mm^2, anyone?

We should take bets on surface area... lol

Mentioning Luminus again, what is the output at 2A? Probably significantly less, due to the low efficiency... So the Cree will definitely win the efficiency front...

 

[TorchBoy]

Considering that a 5mm^2 die would be edging into Luminus' market share, ...

Why wouldn't they? :shrug:

 

[BentHeadTX]

I'd rather just use a single 800lumen LED that's available now and not obsess if it's made by Cree or not.

Question: As I understand the technological constraints of LED technology, current density causes a huge problem with quantum efficiency and the larger die is just a natural progression to solve this?

From what I gather, a regular screw in type light bulb can remove 5 watts of heat without any exotic cooling systems. Sure, you can have 800 lumens from a single LED RIGHT NOW but not without heat pipes, copper bases and current limiting with heat sensors. Take six XP-G R5 bin LEDs and drive them at one watt each for 800 lumens. Don't forget the copper and heat pipes to dump the heat so it will stay alive screwed into a socket. The LEDs themselves will produce around 4 watts of heat when driven at 6.4 watts, throw in the heat from the AC/DC converter regulator and it can be done.

The COST is very high though. The best way to make LEDs live in light bulbs is a large surface area to easily pull the die heat away. Maybe in another year Cree will have a 6 sq. millimeter die that pumps 850 lumens at 4 watts. At the 212 lumen per watt level, it will produce 1.5 watts of heat which is very easily and cheaply dealt with.

Now that incandescent bulbs are being legislated out of business in the next few years, the big money is in replacement bulbs. BILLIONS of light bulbs will migrate over to LED bulbs in the next 5 years--cha ching!

Just as most people won't pay more than $20 for a flashlight, most people won't pay more than $20 for a light bulb. One big LED die that can pump 850 lumens without frying itself because it can produce less than 4 watts of heat doing so...that is the holy grail. The warm tints and assorted whining? The warm always follows the cool tints so all is good.

 

[luckybucket]

ugly[/B] cool white emitters (with the same CRI as a circa 1978 K-mart brand fluorecent tube) to incan light bulbs.

That reminds me of Joe Versus the Volcanoe. I have always hated flourescents because of their crappy light characteristics. I'll be switching to leds soon after high cri P7's arrive.

 

[Gryloc]

Crap... Double post. Sorry.