Phosphor conversion of photons in LEDs & photon recycling efficiency

[Genzod]

Re: Phosphor conversion of photons in LEDs

I know him personally. He used a cool-white XP-G2 and de-domed it using the classic way (imersion in nitro paint thinner or similar). This removes the silicone lens on top of the LED, but does not (to my knowledge) alter the silicon-phosphor-mix underneath. But you can ask him yourself, he is also a member here on the forum.

Thanks, Driver. I got a reply from him today. He used a chemical dedome with no sealer. That means the index of refraction is 1.0 and 1.8 and the critical angle is 34 degrees. Since the difference between the virtual critical angle of 48 (that yields 55% efficiency in lumen gain) and the actual critical angle is 14 degrees, I think it implies the exposed phosphor is rough, changing incident angles of photon impacts and facilitating better absorption to capture more of the reflected light.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Conceptual error: I re-examined my calculations for photons approaching the die at shallow angles.

Snells law can't describe an angle of approach that prevents photons from entering a slower medium (air to phosphor) like it can in reverse (phosphor to air) to describe TIR.

If photons in air coming from 70 degrees (from vertical axis) impact the phosphor surface, refraction into the die takes place 31 degrees from the vertical axis. Even 90 degrees (horizontal to die surface) has a refractive angle of 34 degrees from normal, as weird as that seems to my intuition. So there is always some refraction taking place from all angles of approach coming from the collar to the die.

What I need to understand now is: to what extent photons are also reflecting off the phosphor at shallow angles of approach. I've seen a % reflection vs. angle of incidence profile for air to glass. Does anyone know where i might find such a profile for air to bare YAG phosphor?

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

- reflection as you mentioned (what I know: at 90° angle the phosphor reflects ~3% at the phosphor-air-junction)

Now that I have a better grip on reflection/refraction with Snell's law (considering photons as a particle only without equal attention to its dual wave nature had thrown me for a conceptual loop), I reconsidered what you said here and it makes a lot more sense now, and is very helpful.

I found this equation for normal reflectance from this source:

I don't know if this form of equation can be applied generally to the indices of different mediums. Pressing carelessly ahead, using 1.8 for phosphor and 1 for air and substituting, I get 8% reflection for head on air to phosphor. Do you have a reference for the 3% you mentioned?

For glass, head on, reflection is about 4% with 96% transmittance. When the incident angle approaches about 60 degrees, the efficiency improves (I think close to 0% reflection), then starts dropping again toward 4% reflection as angle continues to become more shallow (as when the aspheric lens is twisted closer to the LED for flood). Here is the profile of reflectance vs. incident angle for glass with more detail beyond 70 degrees in the original plot of this I saw elsewhere. There might be some analogy here with air and phosphor. If I could just find a similar plotted profile for the optical efficiency of phosphor.

To throw another wrench into my noobular mindset, I'm wondering if bare air to phosphor reflections of light rays returning from the reflector are Lambertian (2nd type) and possibly dominated by yellow green wavelengths?

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

I don't know if this form of equation can be applied generally to the indices of different mediums. Pressing carelessly ahead, using 1.8 for phosphor and 1 for air and substituting, I get 8% reflection for head on air to phosphor. Do you have a reference for the 3% you mentioned?

Yes, here. It's basically just the fresnel equations.

You need to take into acount that no LED hast just phosphor. It's always a phosphor silicone mix which mostly contains silicone. Thus you need to use the refractive index of silicone for 450nm: ~1.41
If you want to be really precise you need to take into account the temperature related shift of the refractive index of silicone and also the temperature related shift of the dominant wavelength of the blue led die.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Yes, here. It's basically just the fresnel equations.

You need to take into account that no LED has just phosphor. It's always a phosphor silicone mix which mostly contains silicone. Thus you need to use the refractive index of silicone for 450nm: 1,41

Okay, I just used the refraction index of 1.8 for a phosphor type LED from Dr. Jones CPF/BLF. I see using 1.41 for silicone you get about 3%.

I also see in the article reference you provided that the reflection for a phosphor LED type surface is specular not Lambertian. Thank you.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

I researched this stuff during the past year for my laser phosphor project. I summarized my findings including the sources here (all in German).

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

I researched this stuff during the past year for my laser phosphor project. I summarized my findings including the sources here (all in German).

What an amazing study! Ausgezeichnet! I had read an article from Dec 2017 where a university was studying laser based LEDs. They said the anticipated results would improve efficacy by 10x over the typical YAG phosphor LED. I was contemplating getting a blue laser and targeting a cooled LED to see how it might improve upon the wavien collar, and you saved me a little work! I was discussing this with Degarb. I think if they are able to provide LEDs with 10x luminance over what we have now in the XPG2, the wavien collar will die an untimely death. Even if laser LEDs cost 10x that of a $3 XPG2, that's still 3-4x less than a wavien collar that only gets you 2.2x.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

The LED with the highest luminance (Osram Black Flat HWQP) currently does up to 260cd/mm^2. Osram already offers a laser phosphor module with 3700cd/mm^2. That is in the realm of the best short-arc bulbs (up to 6000cd/mm^2). This module is much bigger though than a XP-G2 based flashlight.

At the present time laser phosphor modules are much more expensive compared to LEDs. If you want 1000 lumens (to get a somewhat large hotspot) you will need to pay at least 100-150$. You need the expensive diode, the phosphor target and a special laser driver (more precisely regulated). You also need to think about safety if you are using separate components.

A nice alternative might be the Soraa SLD. It's a smd module with laser and phosphor that doesn't emit any laser radiation.

The Wavien collar is already "dead". The company is gone. Another company owns the patent, but is not selling it separately and not fully using its potential in the one light they offer.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

The LED with the highest luminance (Osram Black Flat HWQP) currently does up to 260cd/mm^2. Osram already offers a laser phosphor module with 3700cd/mm^2. That is in the realm of the best short-arc bulbs (up to 6000cd/mm^2). This module is much bigger though than a XP-G2 based flashlight.

The Wavien collar takes up a bit of space itself, and then you have to create more available length in the light cavity to meet a 60 degree beam that ordinarily would be 90-120 degrees. With the wavien collar in the way, the ability to zoom to flood is lost to a great degree. I haven't seen any laser modules as of yet, but I would imagine that the option of constructing the laser to come from under the phosphor would leave the space in front free of obstruction.

I'm glad to hear Osram is offering an LD module! Could you link me to a sales page for the one you mentioned? EDIT: Okay, called Phaser 3000 and Phaser 500. Got it.

EDIT: With 3700 cd/mm^2 @40 watts, I get about 2730 cd/mm^2 @18.63w for a maxed out Flat Black for the comparison. If the 260 luminance figure you provided is anywhere close to 795 lm @ 4.6A and 4.05V, then that's roughly 10x more intensity for the same power supplied.

Size and weight for the product is 90mmx220mmx175mm and 3.9kg, not exactly EDC yet.

At the present time laser phosphor modules are much more expensive compared to LEDs. If you want 1000 lumens (to get a somewhat large hotspot) you will need to pay at least 100-150$. You need the expensive diode, the phosphor Target and a special laser driver (more precisely regulated). You also need to think about safety if you are using separate components.

So it is nearly breaking even with the old price of the collar, and it is certainly cheaper than having one made for you by a machine shop or a molding and coating manufacturer. At least now you can buy a module, which is something you can't do with the present Waiven patent owners.

The Wavien collar is already "dead". The company is gone. Another company own the patent, but not selling it separately and not fully using its potential in the one light they offer.

As long as the patent is still out there and has legal bite, the collar technology has value and the idea is still alive even though the manufacturing company is dead. I meant that the legal threat of the patent will become irrelevant when the technology loses its relevance in lieu of the laser module revolution. I would expect prices for laser modules to eventually come down just like anything else. I remember when it cost several thousands of dollars for a 40" 1080p LED TV. We waited for the technology to become cheap and bought ours for $500.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

The Wavien collar takes up a bit of space itself, and then you have to create more available length in the light cavity to meet a 60 degree beam that ordinarily would be 90-120 degrees. With the wavien collar in the way, the ability to zoom to flood is lost to a great degree. I haven't seen any laser modules as of yet, but I would imagine that the option of constructing the laser to come from under the phosphor would leave the space in front free of obstruction.

I'm glad to hear Osram is offering an LD module! Could you link me to a sales page for the one you mentioned? EDIT: Okay, called Phaser 3000 and Phaser 500. Got it.

EDIT: With 3700 cd/mm^2 @40 watts, I get about 2730 cd/mm^2 @18.63w for a maxed out Flat Black for the comparison. If the 260 luminance figure you provided is anywhere close to 795 lm @ 4.6A and 4.05V, then that's roughly 10x more intensity for the same power supplied.

Size and weight for the product is 90mmx220mmx175mm and 3.9kg, not exactly EDC yet.



So it is nearly breaking even with the old price of the collar, and it is certainly cheaper than having one made for you by a machine shop or a molding and coating manufacturer. At least now you can buy a module, which is something you can't do with the present Waiven patent owners.



As long as the patent is still out there and has legal bite, the collar technology has value and the idea is still alive even though the manufacturing company is dead. I meant that the legal threat of the patent will become irrelevant when the technology loses its relevance in lieu of the laser module revolution. I would expect prices for laser modules to eventually come down just like anything else. I remember when it cost several thousands of dollars for a 40" 1080p LED TV. We waited for the technology to become cheap and bought ours for $500.

Yes, the collar is not really a very practical thing. It was also overly expensive considering what it does. They could have made it out of aluminium and offered it for a quarter of the price.

The lasers are already cheap if you are ok with a using multiple diodes in a single light (a laser bank where all the beams are combined using optics). Only the very powerful multi-watt diodes from Osram and Nichia are actually expensive.

The Osram Black Flat's luminance has been measured by multiple people and several record braking throwers have been built with it. The highest value ever measured (that I have seen) is 260cd/mm^2 at maximum output (4.5-5A). In the light linked in my sig we got up to around 240-250cd/mm^2.

EDIT: your PM box is full!

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Yes, the collar is not really a very practical thing. It was also overly expensive considering what it does. They could have made it out of aluminium and offered it for a quarter of the price.

The lasers are already cheap if you are ok with a using multiple diodes in a single light (a laser bank where all the beams are combined using optics). Only the very powerful muliti-watt diodes from Osram and Nichia are actually expensive.

The Osram Black Flat's luminance has been measured by multiple people and several record braking throwers have been built with it. The highest value ever measured (that I have seen) is 260cd/mm^2 at maximum output (4.5-5A). In the light linked in my we got up to around 240-250cd/mm^2.

EDIT: your PM box is full!

Entschuldigen Sie, sehr geeherter Herr! Es it jetzt leer.

A machined and polished aluminum block (6061T) will get equal to or better than the 99% reflectivity of the collar across the spectrum?

 

[Enderman]

Re: Phosphor conversion of photons in LEDs

A machined and polished aluminum block (6061T) will get equal to or better than the 99% reflectivity of the collar across the spectrum?

No, only precision dielectric coated glass or silver coated diamond-turned reflectors can get anywhere near 99%.
Polished aluminum in the best case is 95%.
Polishing would also decrease the surface accuracy, and normal machining will not give an accurate surface finish.
You need to electroform the reflector or have it diamond turned.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

No, only precision dielectric coated glass or silver coated diamond-turned reflectors can get anywhere near 99%.
Polished aluminum in the best case is 95%.
Polishing would also decrease the surface accuracy, and normal machining will not give an accurate surface finish.
You need to electroform the reflector or have it diamond turned.

Enderman! I've followed your two light cannon builds with much interest! Thank you for contributing here. You lumen-wizards will make a photon-professor out of me yet!

Out of curiosity if you know (I imagine you would), which would be cheaper to do--contract for one electroformed reflector to 99% or silver coat one diamond turned reflector?

 

[Enderman]

Re: Phosphor conversion of photons in LEDs

Enderman! I've followed your two light cannon builds with much interest! Thank you for contributing here. You lumen-wizards will make a photon-professor out of me yet!

Out of curiosity if you know (I imagine you would), which would be cheaper to do--contract for one electroformed reflector to 99% or silver coat one diamond turned reflector?

Hahah thanks

Diamond turning is very expensive and usually only done for stuff like making mandrels.
Electroforming is a much more budget option, but only if the electroforming company has an already made mandrel for the reflector size you need.
Making a custom mandrel will cost like 5-10k, so you're limited by the stock options (with the option of having the reflectors cut to smaller sizes too if you need)
Unfortunately neither optiforms not phoenix have spherical reflectors of the size or angle we need, I may try contacting phoenix to see how expensive it would be for them to make a custom spherical one.

For the electroformed optics the silver coating is best with ~98% reflectivity but is very delicate, so unless you have it in a completely sealed enclosure it needs to have a protective coating on it making it even more expensive.
Aluminum coating is ~92% reflectivity and more durable, also cheaper.
Rhodium coating is what people use for short arc lamps that emit UV and need a durable reflector, but the rhodium is less than 85%.

If you get glass reflectors then you can have stuff like cold mirror or dielectric coatings which are 95-99% but glass is also more expensive and delicate.
The reason a cold mirror is ideal for collars is because all the infrared low wavelength light emitted is not reflected back, and this reduces the amount of heat increase due to the collar.
This low wavelength light does not increase the brightness at all and just heats the LED up needlessly, and more heat means lower output.

There is, as you noticed, also light that leaks through the cold mirror, which could theoretically still be used to increase intensity.
Leds however don't output that much infrared energy so I'm curious if using a solid metal collar reflecting all energy back would actually increase or decrease performance.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

My comment concerning the price of aluminium vs glass was based on production of larger numbers of aluminium reflectors. I don't think electroforming is needed for such a collar.

The glass variant from Wavien with dichroic coating leads to highest perfromance, yes. But 5% more output for 4x the price is rarely needed. I think they should have sold both variants.

The light loss in the Wavien reflector should not be of concern. It has a reflectivity of around 95% if it has a standard cold mirror coating. 5% of the output of a XP-G2 is around 50lm. Thats very easy to see, but it doesn't make much of a difference.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Saabluster said an aluminum core reflector would work, but also added a painted reflector would not. I'm thinking in terms of measured albedo, and I know for white acrylic paint, the albedo is 80%. That's not far removed from 85-95% for coated reflectors. My point isn't to use white paint, but simply say, does the argument about efficiency really boil down to what Saabluster said, yes or no, all or nothing?

If you take the Wavien collar Sven_m used in his experiment, the gain mathematically described is (0.75*0.4+0.25)/0.25=2.2. That means 40% of the wasted light (light predestined to not make it directly through a lens) is reflected back to the die and is redirected out the opening toward a lens.

Assuming the reflector's reflectance efficiency is 99%, we can rearrange the equation as [0.75*(0.4040)*0.99+0.25]/0.25=2.2 to say the same thing, but identify where the reflection efficiency plays its role.

Now if you have a glass or PMMA bubble with a 60 degree beam opening, and you paint the outside with acrylic paint so the interior surface looks smooth, the equation becomes [0.75*(0.4040)*0.8+0.25]/0.25=1.97 That's only a 10% loss in performance over the wavien collar.

So how is it reflector paint won't work, when white paint only has a 10% performance loss?

Keep in mind this isn't an assertion! It's a construct designed to give us a medium to dissect and analyze so we can bring out the actual truth. I think I might understand why it won't work, but I'd like to bounce it off everyone here to get a better understanding of the dynamics of the light in the LED and collar.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

It would be extremely helpful in dissecting the above argument if someone has a link to anyone who has actually experimented with an aluminum core reflector and obtained gain results by experiment.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

Saabluster said an aluminum core reflector would work, but also added a painted reflector would not. I'm thinking in terms of measured albedo, and I know for white acrylic paint, the albedo is 80%. That's not far removed from 85-95% for coated reflectors. My point isn't to use white paint, but simply say, does the argument about efficiency really boil down to what Saabluster said, yes or no, all or nothing?

If you take the Wavien collar Sven_m used in his experiment, the gain mathematically described is (0.75*0.4+0.25)/0.25=2.2. That means 40% of the wasted light (light predestined to not make it directly through a lens) is reflected back to the die and is redirected out the opening toward a lens.

Assuming the reflector's reflectance efficiency is 99%, we can rearrange the equation as [0.75*(0.4040)*0.99+0.25]/0.25=2.2 to say the same thing, but identify where the reflection efficiency plays its role.

Now if you have a glass or PMMA bubble with a 60 degree beam opening, and you paint the outside with acrylic paint so the interior surface looks smooth, the equation becomes [0.75*(0.4040)*0.8+0.25]/0.25=1.97 That's only a 10% loss in performance over the wavien collar.

So how is it reflector paint won't work, when white paint only has a 10% performance loss?

Keep in mind this isn't an assertion! It's a construct designed to give us a medium to dissect and analyze so we can bring out the actual truth. I think I might understand why it won't work, but I'd like to bounce it off everyone here to get a better understanding of the dynamics of the light in the LED and collar.

The equation does not make any sense to me!

The LED is de-domed, so it is a lambertion emitter. 75% of the total amount of light that exits this LED hits the collar. The remaining 25% goes through the hole in the collar. The blue content of this 75% enters the phosphor again (minus reflection discussed above) and is converted and exits the phosphor in a different angle. This leads to an increase of 120% of the 25% of the light which I already mentioned. The gain of 120% will vary from LED to LED, also depending on drive current, heat, tint etc. It has no direct mathematical relation to the amount of angular light that hits the collar.

Where do you get your 40% figure from?

Any standard reflector of the same shape will work. The performance will only vary slightly depending on the reflectivity. The larger the size difference bwteen reflecotr and LED diameter becomes, the more precise the reflector needs to be.

Photon in the German forum tried selfmade stainless steel collars a few years ago. They worked, but not very good. Why? Because stainless steel only reflects around 30% of 450nm light. Low gain wa salso caused by the larger hole on the top, by the fact that he polished them instead of putting on a real coating, and imperfect shape, small collar size, sub-optimal led tint, maybe imperfect focus etc..

Concerning paint:
Easy test for you: take a focussed flashlight and shine it at the kind of surface you are describing (in the dark). Does the reflected "beam" throw as far as the flashlight does by itsself? You can be more precise by using a lux meter. You will have the same losses with a spherical shaped surface. I can't in any way imagine how white paint would ever work!
Reflective silver-colored paint would certainly work, but it will not be as smooth as a real reflector. This reduces the effective surface area so not all of the available light will actually hit the LED. In addtion to this the (probably lower) reflectivity needs to be accounted for.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

The equation does not make any sense to me!

Where do you get your 40% figure from?

40% of the light hitting the collar is added to the opening. 40% * 75% = 30%

This is the 30% of total light you need to make the added 120% you mentioned, as 30% / 25% =120%

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Concerning paint:
Easy test for you: take a focused flashlight and shine it at the kind of surface you are describing (in the dark). Does the reflected "beam" throw as far as the flashlight does by itself? You can be more precise by using a lux meter. You will have the same losses with a spherical shaped surface. I can't in any way imagine how white paint would ever work!
Reflective silver-colored paint would certainly work, but it will not be as smooth as a real reflector. This reduces the effective surface area so not all of the available light will actually hit the LED. In addtion to this the (probably lower) reflectivity needs to be accounted for.

If the albedo is 80%, it's obvious throw would decrease. I imagine someone has already taken a lux meter to acrylic paint and measured 80% reflectance. But the point was never to use white paint, just illustrate numbers.