Die luminance, emittance and advanced die heatsinking

[Ra]

Re: Die luminance (surface brightness, important for throw)

BTW Have you been reading my threads?:

No,, But I did just now... I'm doing the same thing with the XP-G R5 right now.. Two are finished.. Still 4 to go..


Regards,

Ra.

 

[bshanahan14rulz]

Re: Die luminance (surface brightness, important for throw)


Cree uses SiC substrates in xr-e, much like what was described in Anthony's link.
Also, fans of How It's Made might like this clip, showing how they do binocular lenses. At one point, the head lens grinder shows how he uses optical mounting to secure a prism to be ground. http://science.discovery.com/videos/how-do-they-do-it-binocular-lenses.html
I like seeing the big on-the-edgers exchanging ideas so enthusiastically too. Brain Storms for the win!
So yeah...

 

[Ra]

Re: Die luminance (surface brightness, important for throw)

Cree uses SiC substrates in xr-e, much like what was described in Anthony's link.
Also, fans of How It's Made might like this clip, showing how they do binocular lenses. At one point, the head lens grinder shows how he uses optical mounting to secure a prism to be ground. http://science.discovery.com/videos/how-do-they-do-it-binocular-lenses.html
I like seeing the big on-the-edgers exchanging ideas so enthusiastically too. Brain Storms for the win!
So yeah...

SiC is a big advantage.. Do they use the same in the XP-G and XP-E led's ??


Ra.

 

[Dr.Jones]

I added some die emittance data to first post.

 

[Ray_of_Light]

Take a look at this PPT presentation:

http://www.group4labs.com/ppt/G4L_THREADS-Presentation_v13.ppt

it goes somewhat into detail - with comparisons about SiC vs CVD substrates.

I personally believe that the next generation of power LED will be based the use of InGaN semiconductor on CVD substrates. Technology will bring the costs down, as usual.

Regards

Anthony

 

[saabluster]

Nice work. I would say however that I've never seen any evidence that the XP-E has the ez900. As far as I know it is only the XR-E. Also the XP-E comes in R3 as the top bin.

As to the ez900 vs ez1000 it should be noted that the ez900 die cannot produce the same amount of output at 1A as the ez1000 version even if they are binned the same. This from a Cree internal document. I'm not sure I can post it as it says confidential although I don't see what's so secret about the info contained within. Anyway thought you might like to know.

 

[Dr.Jones]

Thanks for that info. I updated the list and added XP-C.
XP-E and XP-C seem to be quite interesting choices...

As to the ez900 vs ez1000 it should be noted that the ez900 die cannot produce the same amount of output at 1A as the ez1000 version even if they are binned the same. This from a Cree internal document. I'm not sure I can post it as it says confidential although I don't see what's so secret about the info contained within. Anyway thought you might like to know.

Hm, I see. That means the flux-vs-current graph is inaccurate for EZ900, which sounds plausible. Also thermal sag might be worse. Did they give any estimated values on that?

 

[saabluster]

Thanks for that info. I updated the list and added XP-C.
XP-E and XP-C seem to be quite interesting choices...


Hm, I see. That means the flux-vs-current graph is inaccurate for EZ900, which sounds plausible. Also thermal sag might be worse. Did they give any estimated values on that?

It's not much at 1A. They don't have the data spelled out, just a graph. I'd say the ez900 is down about 6 lumens at 1A to the ez1000. However for people who overdrive it should be noted that that gap gets bigger and bigger as the current rises. Both dies are at the same levels until about 450mA when the ez900 starts to tail off a bit. Again nothing major at normal drive levels but there is a difference.

 

[Al Combs]

Thanks for pulling all this data together for us. Could I ask where you got the thermal resistance figures for the SST-50 and the SST-90? The Luminus doc's say the SST-50 has a thermal resistance of 2.45 °C/W and the SST-90 is 0.64 °C/W.

 

[Walterk]

I am surprised to find the SST90 this high in the ranking. Nice though!
And it explains the performorance of the FlyDragon.

 

[Dr.Jones]

Thanks for pulling all this data together for us. Could I ask where you got the thermal resistance figures for the SST-50 and the SST-90? The Luminus doc's say the SST-50 has a thermal resistance of 2.45 °C/W and the SST-90 is 0.64 °C/W.

I took the junction-to-heatsink value, page 12. Might not be the best idea, since the values for the CREEs are junction-to-solder-point. I'll change that later.

 

[Ra]

So to rebump this thread, I did some testing (and still do..)

As there is much discussion about the best setup for throw (and lumens output), I decided to put some things to the test.
The results are reason to start a new thread (in a few weeks..or months..) about the various collimator setups, and their effect on throw and lumens output (torchlumens).

But the thing I want to share right now: In my quest for bigger glass based TIR optics, I just made a 35mm diameter TIR. Below is the 35mm test setup with a Cree XR-E R2 (EZ1000)

And the next picture immidiately shows why so many people prefer the TIR over the aspheric lens:

Compare that to the spot of the aspheric setup: (note that differences in spot brightness are due to different camera settings!)

Now the purpose of these test setups: The big question that is asked many times: Which is better for throw, aspheric lens or TIR optic ?

I carefully created two very identical setups, one with the 51mm aspherical lens, and one with the 35mm TIR. In each setup, I used the same led, provided exactly the same current, and measured exactly at the same distance, using exactly the same lux meter (as I only have one..)
For correcting the difference in size, I recalculated the data from the asphere to the level of the 35mm TIR, and to be sure, I also used an exact diaphragm on the asphere to bring the effective apperture down to that of the 35mm TIR..

(Please note that the aspheric lens and TIR-optic I used during these tests are both uncoated.)


Here are the results: (3.6 mtr measuring distance, 750mA to led)

51mm Aspheric lens without diaphragm : 4,620 Lux (59,875 lux@1m)

35mm TIR optic : 2,020 Lux (26,179 lux@1m)

Aspheric lens 35mm diaphragm : 2,182 Lux (28,279 lux@1m)


So the aspheric lens wins !!!!???? Not so fast guyzz !! :

My 35mm TIR has a 10.5mm diameter 'refraction chamber' that does not play along until an extra collimation lens is placed, and I've not placed it yet..
So that means the overall effective surface of my 35 mm TIR is not 962 square mm as with the 35mm aspheric setup, but more like 875 square mm !!

This means that when the TIR is finished (with central lens) it will give 2,219 lux.. (28,758 lux@1m)

Conclusion: (surface recalculation included)

Aspheric lens 51mm : 29.31 lux/mm2 (@1m)

35mm TIR : 29.91 lux/mm2 (@1m)

35mm asphere (diaphragm) : 29.40 lux/mm2 (@1m)


As these results are only 2% apart, I hope you'll agree that these differences are not big enough to declare a winner! So, aspheric lenses (of decent quallity of cource) give the same throw as high quality TIR optics with the same diameter.

But almost everyone knows that TIR optics give a much nicer beam with more usefull sidespill (more lumens) and no extra colors.

The main reason why so many use aspheric lenses is that they are cheap and commonly aviable, the aviabillity of TIR optics are limited by size and material (30mm tops, acrylic based, damaged easily).
Certainly the glass based TIR optics cannot be found anywhere (correct me if I'm wrong..) I think that right now, I'm the only one that makes them..

I'm working on a 65mm diameter high quatily TIR optic right now. It will be finished next week.. I hope...

Edit: Pushing the limits, with the XR-E R2 at 2amp, I managed to reach 45,000 cp with my 35mm TIR. When recalculating, that would give like 140,000 cp with the 65mm TIR ! (DEFT killer?)

And here is a picture of the illumination of the internal surface of the 35mm TIR optic:

Now some claim that throw with a aspheric lens can be improved by adding a pre-collimator lens close to the led.. This is not true !! In a high quality setup, it only improves the overall lumens output.
When a pre-collimator lens gives better throw with an aspheric lens, the only possible conclusion would be that the aspheric lens has imperfections that are overcome by the pre-collimator lens.

In fact: Compared to a single high quality aspheric lens, a pre-collimator setup has (somewhat) less throw, caused by the reflection and absorbtion losses within the pre-collimator!


Regards,

Ra.

 

[Al Combs]

Wow, a 65 mm TIR.:bow: Would love to hear some details on how you make glass TIR's. If it's not proprietary info of course. I ground and polished a few telescope mirrors years ago. Can't begin to imagine how those techniques would apply to a parabola that steep. Or whatever type of conic section it is.

 

[Ra]

Wow, a 65 mm TIR.:bow: Would love to hear some details on how you make glass TIR's. If it's not proprietary info of course. I ground and polished a few telescope mirrors years ago. Can't begin to imagine how those techniques would apply to a parabola that steep. Or whatever type of conic section it is.

Hi Al,

Especially for you, I made picture of the not yet finished 67 (!) mm TIR optic:
The refraction chamber is the hardest part, but it's finished now (after two day's..)

As you propably can see, alot of glass still needs to be removed! A dutch sculpterer once said: "The shape of the sculpture is already there, I only need to remove the material around it.."

I cannot explain how to make them without making this the longest post ever! Sorry about that.. It mostly is work by hand.


Regards,

Ra.

 

[Al Combs]

Nice. Thanks for the look see. Also I never realized Michelangelo was Dutch.

Hey if you put a CSM-360 in that TIR optic, would there be a donut? Can't wait to see what you do with that.

 

[ma_sha1]

Woe, 67mm TIR, that's too cool!

 

[EasyB]

I'll chime in with respect to throw and simple lenses. A lens just projects an image of the LED far away. If you want the image to be very far away, you put the LED very close to the focal point of the lens. The size of the image can be calculated from the fact that the image magnification is equal to i/o where i is the image distance and o is the distance from the LED to the lens. So approximately, we see that the size of the image of the LED is i/f times the LED size. So the longer the focal length of your lens, the tighter your beam will be.

Of course, if you have a long focal length lens your LED is very far from the lens and you are losing a lot of light to the sides. This is why you would want a larger lens and/or possibly a collection lens close to the LED like in my SST-50 spotlight. The effects of the collection lens I must admit I understood better when I was designing my spotlight, but it is a compromise between light collection and enlarging the effective LED size.

This little thing:
http://www.shermanlab.com/science/physics/optics/ThinLens.php

helped me visualize what's going on better when your LED is within the focal length of the lens, which is the case for the collection lens in my spotlight.

 

[Ra]

Nice. Thanks for the look see. Also I never realized Michelangelo was Dutch.

Hey if you put a CSM-360 in that TIR optic, would there be a donut? Can't wait to see what you do with that.

Maybe Michelangelo knew the dutch sculpturer...


CSM-360 ? :huh: Never saw that before ! Thanks..

The TIR optic acts like a lens with a very short focal length: Very narrow focus area (compared to size), so with the 36 square millimeter of the 360, I think you indeed need at least a 50mm TIR for the CSM-360 to grab it's full potential.
Fortunately, as you can see in the pictures above, the TIR does not give the annoying die image of the led.. But how it reacts to the CSM-360 setup, I don't know yet.


Do you happen to know where I can get a 360 ? (and are they expensive??)


The 67mm TIR is for my new LedBlaster, with one Cree XR-E R2 pushed to the limits: max 2.3 Amp heatsinked on CVD Diamond ! (CVD Diamond has more than 5 times the heat conductivity of copper)
With the diamond setup, the led runs much cooler than ever possible with a copper heatsink, so there is a gain in efficiency (which already is very poor at those currents)


Regards,

Ra.

 

[Ra]

And....

As for advanced heatsinking: Sanding down the base of the Cree led's significantly improves heatsink effectivity! No mater if you use aluminium, copper, silver or CVD diamond for heatsinking, the ceramic base of the Cree has a thermal conductivity of around 30 W/mK.
Aluminium has a thermal conductivity of 250, copper 400, silver 429 and CVD diamond aprox 2000 !
So the closer you can get the die to one of those materials, the better your heatsink will work!

I already sanded down almost all of my led's.. Here's an example of a sanded down XP-G R5: (small led closest to the (BIG!) XR-E)

The remaining ceramic base is so thin that it becomes transparent..



Regards,

Ra.

 

[Al Combs]

The TIR optic acts like a lens with a very short focal length: Very narrow focus area (compared to size), so with the 36 square millimeter of the 360, I think you indeed need at least a 50mm TIR for the CSM-360 to grab it's full potential. Fortunately, as you can see in the pictures above, the TIR does not give the annoying die image of the led..

Do you have a 16 mm refraction chamber hole? That's the size of the CSM-360's dome.

But how it reacts to the CSM-360 setup, I don't know yet.

Does that mean what I think it does?:naughty:

Do you happen to know where I can get a 360 ? (and are they expensive??)

In the United States I would buy a CSM-360 from Avnet. They have a top Bin GV100 for $157.07. Unfortunately I've seen from several people in Europe they want a $75 shipping fee. That would pretty much make them a last resort for you. Pepko has a thread on a CBM-360 MagLite. He said in post #16 he bought it, "through local (czech) distributor EBV". They are listed as the European dealer at the bottom of the Luminus CSM-360 link as well. The CBM-360 is the one without the lambertian dome and slightly less powerful. The CBM-360 has a top bin WX and a 4,300 to 5,100 lumen output @ 6.3 amps. The CSM-360 has a top bin WV and a 3,600 to 4,300 lumen output @ 3.15 amps. Pixel interpolation from their, "Relative Luminous Flux vs. Forward Current" graph on page 9 of the specs gives a 193.2% boost @ 6.3 amps for a total output of 6,955.2 to 8,307.6 lumens.:devil: