Experiments in LED encapsulation

[Th232]

Just a bit curious, what's the spatial distribution now like? If it's not too much hassle I wouldn't mind seeing some pics of how it's distributed, even if it's just holding up a piece of paper on the axis of the LED or some similar method.

 

[Illum]

I'm inclined to ask, why the encapsulation of the dome only and not the entire LED?
I'd figure that if this is a mean to provide protection for the LED the weak link still exists from lateral shear, where impact or trauma from an angle may still take out the LED from the PCB substrate [NOT the star] below.

I like the dome encapsulation though, definitely one of those things to have for fixed lighting projects, especially if theres some way to create an increase in die projection:twothumbs

 

[saabluster]

Just a bit curious, what's the spatial distribution now like? If it's not too much hassle I wouldn't mind seeing some pics of how it's distributed, even if it's just holding up a piece of paper on the axis of the LED or some similar method.

I'll try to get some pictures up for you as soon as I can.

I'm inclined to ask, why the encapsulation of the dome only and not the entire LED?
I'd figure that if this is a mean to provide protection for the LED the weak link still exists from lateral shear, where impact or trauma from an angle may still take out the LED from the PCB substrate [NOT the star] below.

I like the dome encapsulation though, definitely one of those things to have for fixed lighting projects, especially if theres some way to create an increase in die projection:twothumbs

Because the dome is where the light comes out. Not really sure what you are getting at there.

This system is no different than the one Cree uses and has the same failure modes. It should should survive high G event just fine. It would be slightly more prone to having the dome pop off if bumped as the lens no longer has the ridge of the ring to help prevent lateral motion. My point in doing this experiment was not to create a more robust package but to create better package optics for our specific application. I think I have done that.

 

[clint357]

Way to go! I was actually thinking of doing something like this to my Bridgelux since it is such a massive led as far as emitter area. The bridgelux is cheap, easily driven, and has no problem putting out over 3,000 lumens....but all that light is a real trouble to put into a beam without an aspheric condenser right on the dies. Again, great job and congrats.

 

[Inliner]

This was just a proof of concept.

And a darn good one:rock:

I'd love to get my hands on a few if you crank some out for sale.

 

[bshanahan14rulz]

do you think regular norland gel (or whatever it is kids call it these days) that you can buy in the MP will work? I might be getting some for some trits and figure as long as I have it, I could use it to thinly protect my xr-e.

I'm hoping for crazy throw, even in an OP reflector, since preliminary tests (Me holding emitter on reflector with low current) seemed to show that I could still achieve focus *eventually*

 

[saabluster]

And a darn good one:rock:

I'd love to get my hands on a few if you crank some out for sale.

Thanks. We will see. I still have a bit of testing to do.

do you think regular norland gel (or whatever it is kids call it these days) that you can buy in the MP will work? I might be getting some for some trits and figure as long as I have it, I could use it to thinly protect my xr-e.

I'm hoping for crazy throw, even in an OP reflector, since preliminary tests (Me holding emitter on reflector with low current) seemed to show that I could still achieve focus *eventually*

I believe the Norland people use for trits is an epoxy and most definitely will not work.

 

[bshanahan14rulz]

aww, k. Well, there will still be a glass window between the outside world and my LED.

I've always thought that even though it is said to reduce output, it would be worth it to remove the lens just for the higher amount of light per area of light emission (even if overall light emission is lower)

Anyways, I'm gonna sit back and watch the show

 

[SFG2Lman]

excellent work once again...your dedication to every foot of throw is unbelievable...That must be what makes your lights so perfect.

 

[Gunner12]

That reminds me of another idea I had and still want to try out. The rebels would be ideal for it as well. My idea is to take 5 dies and position them so as to form a box. One would be flat as a normal LED. The other 4 are raised up to form the sides of the box with the top open for light to escape. The idea is that the four LEDs will help add some surface brightness to the base one which is the one the lens would be primarily focused on. You should be able to use an efficient cool white LED and have the resulting tint output to be warmer than it would otherwise. The increase in brightness would be minimal I'm sure but I am looking for any way to get ahead. It is just an idea I have had in my head and I have not had a chance to test it out as yet.

I've been following this thread a bit and as I read the above, an idea popped into my head. You have probably though of it before but here it is.

The yellow parts would be the die of the LED. The blue stuff can be whatever is used to connect the pieces of metal in grey. There will be 5 pieces of metal for the heatsink. The base one would have the LED on a pedestal with the substrate cut away as much as possible. The electricity would be conducted through the bottom of the LED. The other dies would have a LED at the top left corner of a rectangular piece of Al or Cu with the substrates cut aways as much as possible and electricity also conducted through the bottom. Then the rectangle pieces with an LED at a corner will be fit around the base piece with the LED on a pedestal. Does that make sense?

I don't know how the bottom contacts/thing would work though.

Great job with the experiment!

Would this work?

A highly polished reflector (maybe of Al) with a silicon encapsulate LED or one you modified yourself pointed towards the reflector (like the pelican recoil system), filled up to the brim with the same silicon encapsulate used for the LED to reduce the apparent die size to the actual die and also prevent oxidation of the reflector which should result in more throw. The wires and heatsink bars to the LED will block some light though. The LED can sit at the bottom of the reflector like normal, but that increases the rays of light not hitting the escape surfaces at perpendicular angles. Maybe a well designed lens for use in the silicon encapsulate can be suspended at the right place to redirect the non perpendicular light rays to perpendicular. Am I making any sense?

Either way, looking forward to more results!

 

[Illum]

Because the dome is where the light comes out. Not really sure what you are getting at there.

This system is no different than the one Cree uses and has the same failure modes. It should should survive high G event just fine. It would be slightly more prone to having the dome pop off if bumped as the lens no longer has the ridge of the ring to help prevent lateral motion. My point in doing this experiment was not to create a more robust package but to create better package optics for our specific application. I think I have done that.

yeah, I misunderstood your intentions until I reread it a couple times and got my browser to load the pictures, I thought the encapsulation was only for the purpose of protecting the LED, I did not expect to change the output, and therefore collimation of the LED:wave:

So in essence you are turning XR-Es to be able to be used in conjunction with XP-E reflectors?

 

[Gryloc]

saabluster,

Thanks for the reply. Yup, I guess I didn't read enough of your past posts before posting my questions.

I was wondering why you must remove the dome or else the re-encapsulation will not work. I thought that the optical grade silicone gel you have would have a similar index of refraction as the silicone domes that is over top the K2, Rebel, and the XP-line. Is it a matter of adhesion (so if bumped, air bubbles causing internal reflections would appear)? I was hoping for non-DEFT applications, the apparent die size (like of the XP-G) could be shrunken to work better with smaller reflectors (with only a small decrease in overall output). I was hoping for an emitter that would yield similar results as your oil and water experiments, but without the potential for leakage. Leaky flashlights cause me to get funny looks from non-flashaholics (I use buttered grits as fuel lol).

If the silicone would not work well in contact with with the original silicone gel, could another optically clear and thermally stable material be placed between the original dome and the added silicone gel? I was thinking about something with similar properties to the "binder gel" that sits between the phosphor impregnated silicone and the glass dome of the XR-E, like oil or some other high boiling temperature liquid or gel. I understand what you are doing. I see why it would be easier to remove the dome to guarantee a smaller apparent die size (and why the XR-E is favorable for removal of the dome). I worked on the XR-E as deep as the gooey binder gel above the die's phosphor layer. I did not want to strip the rest of the gel all the way to the phosphor layer in case I would break a bond wire (plus I had no need to go further in my test).

So, I guess that I am not asking you to try this (I got overly excited before). However, is what I am asking even possible?

Oh, and could you tell us where you got your optical grade silicone gel? I would eventually like to invest in some of the gel to try the same. It is awesome to see a fellow nerd use the stuff to make a standard emitter work better in their application. Thanks.

-Tony

 

[bshanahan14rulz]

So in essence you are turning XR-Es to be able to be used in conjunction with XP-E reflectors?

I think he is going for a more optically usable source, since he is a maker of an aspheric thrower. This would decrease the size of the source while only slightly decreasing the total lumens output.

I've always said that there is something to be had for an LED w/o a dome. Sure, it helps light extraction, whatever that is, but it also changes the way the light is emitted. With light coming from a smaller area, there is less wiggle room for light to spread, so you hopefully end up with a more collimated beam. It's what I'm hoping for, at least. I'm not doing anything near as fancy as SaabLuster, but it would be cool because when I'm done, I'll have a high-output drop-in, a warm-tinted, better-throwing drop-in (beautiful beam, no cree rings, I might add), and a throwing drop-in with smo reflector

 

[mudman cj]

Sweet! Much higher surface brightness with only a 5% loss in output means much greater throw. It's hard to believe that you also get a desirable tint shift to boot. This is a very exciting development indeed. :twothumbs

 

[saabluster]

I've been following this thread a bit and as I read the above, an idea popped into my head. You have probably though of it before but here it is.

The yellow parts would be the die of the LED. The blue stuff can be whatever is used to connect the pieces of metal in grey. There will be 5 pieces of metal for the heatsink. The base one would have the LED on a pedestal with the substrate cut away as much as possible. The electricity would be conducted through the bottom of the LED. The other dies would have a LED at the top left corner of a rectangular piece of Al or Cu with the substrates cut aways as much as possible and electricity also conducted through the bottom. Then the rectangle pieces with an LED at a corner will be fit around the base piece with the LED on a pedestal. Does that make sense?

I don't know how the bottom contacts/thing would work though.

Great job with the experiment!

Would this work?

A highly polished reflector (maybe of Al) with a silicon encapsulate LED or one you modified yourself pointed towards the reflector (like the pelican recoil system), filled up to the brim with the same silicon encapsulate used for the LED to reduce the apparent die size to the actual die and also prevent oxidation of the reflector which should result in more throw. The wires and heatsink bars to the LED will block some light though. The LED can sit at the bottom of the reflector like normal, but that increases the rays of light not hitting the escape surfaces at perpendicular angles. Maybe a well designed lens for use in the silicon encapsulate can be suspended at the right place to redirect the non perpendicular light rays to perpendicular. Am I making any sense?

Either way, looking forward to more results!

Your drawing is essentially what I was talking about. No need for the electrical complications though. Thanks for the input. I shouldn't have gotten off topic myself but let's go ahead a keep the rest of the thread on the encapsulation theme.

Filling a reflector with optical grade silicone would be expensive. You would need to first remove the dome and gel(if an XR-E) so you can find and set the new focal point. Otherwise you will end up with an LED out of focus and no way to move it. I have some cheaper truly clear silicone that might work but it is not purpose made for that. I'll do that some time in the future.

 

[saabluster]

saabluster,



I was wondering why you must remove the dome or else the re-encapsulation will not work. I thought that the optical grade silicone gel you have would have a similar index of refraction as the silicone domes that is over top the K2, Rebel, and the XP-line. Is it a matter of adhesion (so if bumped, air bubbles causing internal reflections would appear)? I was hoping for non-DEFT applications, the apparent die size (like of the XP-G) could be shrunken to work better with smaller reflectors (with only a small decrease in overall output).

It's not that re-encapsulating the whole package wouldn't have created a similar result but it seems mechanically more cumbersome and prone to failure to put my lens on top of the other lens. I also wanted to avoid any potential cure problems with the gel. There is also the issue of the index of refraction. Pulling off just the dome does not leave a perfect surface. Yes by putting a different optically clear silicone over it will make it appear like the surface roughness is gone but it is in fact not. Not unless you use a material with the exact index of refraction. Since Cree does not publish this information and I don't have the money to have the gel tested I figured it best to get rid of it. But who knows it may be even better with it. Only one way to know for sure.

I was hoping for an emitter that would yield similar results as your oil and water experiments, but without the potential for leakage. Leaky flashlights cause me to get funny looks from non-flashaholics (I use buttered grits as fuel lol).

I remember the thread you are alluding to but I can't take credit for the oil/water test. Someone else did that.

If the silicone would not work well in contact with with the original silicone gel, could another optically clear and thermally stable material be placed between the original dome and the added silicone gel?

Forgive me but I'm having a hard time understanding what you mean here.

 

[saabluster]

I've always said that there is something to be had for an LED w/o a dome.

I believe this is why Phlatlight has the name that it does. You can order their LEDs without domes.:naughty:

Sure, it helps light extraction, whatever that is,

That term is referring to getting all the lumens that are actually created out of the package. A lot of light dies before it ever gets to the surface of the die. Still more dies as it travels through the various encapsulation layers. So methods that help set free that light are helping extraction. Note extraction does not imply the creation of light only the transference of it.

 

[saabluster]

So I thought I'd post some shots of the beam to show the change created by my own encapsulation. I tested 6 lights with my light meter to find one that had similar output as the modded XR-E. I used several different positions and methods to try and measure the outputs of the various lights to get a sense of the total relative light output. These include wall/ceiling bounce and various diffusers. As far as raw output is concerned the modded Cree held its own fairly well with the consensus showing it to be about mid pack of the 6 lights. It is important to note that I do not know where this particular LED would have landed prior to being modified but, that said, they were all R2 WH so they couldn't have strayed too far from one another. So the raw output tests helped me to find a light which I knew was very close in output so that the deciding factor in throw was not that one LED was brighter than the other but due to apparent surface brightness. Despite these measures my tests can hardly be considered as lab accurate.

And now the pictures.

All pictures start out with the normal XR-E followed by the modified one.

The first four shots are of a completely bare LED.(no optic or reflector) Keep in mind the light meter says the total output is on par with one another. Since the spacial distribution is different than the normal one it appears in these shots that it is just less bright but it is just an illusion. The normal XR-E has a sharp peak in its output in the very center with intensity decreasing and tint getting less blue as the angle increases. The modified one does not have a blue center portion. I am going to post some more pictures at another time that will show the spacial distribution better.

These are shots against my wall next to a candle holder to help with scale. Again, normal XR-E first.

Just in the interests of full disclosure both of these lights had the small optic(for the long range tests only) that goes just over the LED on the DEFT. There can be variance in the performance of these, lens to lens, so I decided to use one and swap it back and forth. They both also used the same main lens to get rid of that variable. I also used the same batteries for both lights which was the same used to figure the raw output numbers.

There are a few things to note between these next two pictures. First the beam from the normal XR-E appears gray and without color while the modified one is rich in color. This really is how it appears in person. Second it is quite obvious that the second one is much more intense than the first. Third, you will notice the Cree rings on steroids in the second shot. I plan to eliminate those entirely in the next test subject.:naughty:

 

[Th232]

Very nice! Amazing how much of a difference there is...

Can't wait to see what happens to your next vict... err... "test subject".

 

[dom]

Great performance with the new emitter!
You'd be over the moon with that ,i bet

A bit of black nailpolish on the inside of the
ring to get rid of the Cree rings?