The Tuning Process of Reflective Apertures



Sharpie, a RA does not try to focus stray light to a direction. It simply reflects the emitted light directly back where it came from.)

Reflector. Light from point A goes to point B depending on angle. (A spherical parabolic or concave or any shape can do this with differing results and efficiencies

RA. Light from point A is reflected back to point A. (A flat mirror can do this with a laser, but it’s a different story with an LED because it emits light in a wide pattern.)

I haven’t followed this whole thread so my observation here may be irrelevant, but I think one thing you may be missing is that one of the primary functions of wavien collars and the like is to use that “misdirected” light to excite underutilized phosphors and thereby create more output from the emitter than it would normally provide.

Sharpie, you bring up some interesting points, and to optimize an LED/RA system lots of these points should be considered. But it is an observed fact that dedoming (some) LEDs significantly increases the intensity of the flashlight beam, and use of a hemispherical RA also increases the beam intensity. I don’t think people here have all the answers, but let’s start with the facts, that these modifications do work to increase beam intensity. They don’t necessarily maximize the total flux output, but beam intensity is the main concern for some flashlights.


de doming, light recylcling, pre collimator lenses and multiple lenses system really works

Saabluster 1,3mcd monster here

Gaston 1,3 mcd monster here

Mem 1,5mcd here

Ervin Anastasi 1mcd here

They are good for breaking candela records but usability?

When someone makes this with all characteristic of good modded zoomie I’ll buy it :slight_smile:

I guess this sort of conversation could go on forever; me telling you we have all observed these modifications increasing beam intensity, and you saying it just can’t be?

It sounds like you have not read this thread. Also this thread has a good measurement of the luminance increase upon dedoming.

BLF is fun and educational place.

Thanks LI, those are the over-the-top examples I knew existed but didn’t know where to find. :wink:

I’ve done one at 665Kcd before and it was pretty impressive to me. It’s funny to see someone explain why it won’t work, when you’ve built em and know it does.

So, they collimated the light with an aspheric condenser, then reflected it back to the LED with a flat mirror, and didn’t get results like MEM has claimed are possible from a totally different arrangement. So… this is relevant, how? :person_facepalming:

Sharpie your own “proof” that it cannot be done actually proves that it can :stuck_out_tongue:
First they used a flat mirror which would be of very limited value
From your link

Secondly I think that the graph shows a dramatic increase in output considering the flat mirror.
Again from your link

While the improvement is small in the 596nm range it is gigantic at 570/580 and 600/610nm and we see all of those frequencies :laughing:
Also that paper is from 8 years ago, led technology has moved forward a lot since then, I could not even find any info on the led they used it is that out of date.

Cheers David

oh yeah, I can only see those fairies when I am on those little green pills the purple pills cause the dragons to scare them away :cry:

The laser headlamps are, as far as I have found, not legal in the United States.

Not gigantic, are you looking at the same graph as me, 38% average increase or are you looking at the red line that shows the measured increase with no correction for the test method ? look at the dotted blue line that shows the increase with the correction!
When testing using this method you DO need correction for all the bits the light has to go through before it gets back to the led.

Why would they put money into developing “light recycling” when led technology was/is moving so fast,

From this giant, yes I found an image of the led used in that paper :smiley: to the current crop of leds and that alone makes that paper outdated and irreverent as newer leds use different technology to the one tested.

And on that note I am done as you can never be proven wrong in your mind.

Cheers David

“I have no interest in projecting a tiny spot onto a watertower…” lol

Yeah, that’s what I always thought about the tight pencil beam throwers… they reach further than I can see anyway, right? But for the process of reaching new heights, I have become interested in just what actually IS possible…

The beam profile from my Z1 with the SBT-70 is impressive, but not in the max thrower sense. It’s a narrow cone of light that does reach fairly far but it grows larger out there as well, the neat thing is how little spill it makes, as in almost none at all. So a subject can be illuminated without causing much distraction to others nearby. I keep coming back to the Theater Spotlight when I have to describe it, just a nice round circle of light.

Which brings me back to how far can a LED be projected? That square of light on a water tower, won’t be tiny by any measure when the right distance is employed, but just being able to illuminate something that far away is the key. Searching on open water, shining across a valley from one mountain to another, or even just being in big space in the country… a mile eats up virtually every light out there, but to be able to see that mile, or more, could be lifesaving in some instances.

Maybe some can’t understand this concept, but we have places here in Texas that you feel like you can see into next week, and it would take days upon days to walk out to the distance you can see. Up in the Panhandle, it boggles the mind… the country is ruggedly rough, travel of any kind is almost impossible and would take great lengths to traverse. The visual impact, the possibility to search at length, intrigues me. Google Earth search for Canadian, Texas. You might start understanding, even though pictures or satellite images simply can’t convey the sense of standing in it, looking out….

Laser flashlight is something that could attract attention although we already have them(laser genetics) green :confounded: diffused laser lights. I had them and they were not any better than UF T20(10:money_mouth_face: flashlight. I also had version that refused to work on - temperatures :slight_smile:

So how they can achieve right (CW or NW) tint projection with lasers? All that would be interesting to see…

It would be nice to see a flashlight that can light up water tower mile away and have flood mode of 1000+ lumens. And trust me plenty of people would want to light something at that distances.

And my vote and bet is that future light will be zoomie… Laser or LED? I don’t care if it will work but I want all (throw + flood).

Wait, I missed this part. I must confess I didn’t read the whole paper through. A monochromatic LED without phosphor would make this paper even less relevant to us. It appears that they were only trying to get the surface of the LED to act as a simple mirror. Light recycling as we are discussing is more than that. The phosphor is very much involved. Also, it doesn’t take anything special to return a perfect image of the die to the LED. The hemispherical mirror does that already, if it is focused properly. But, as you’ve already noted, the image would be inverted. Since we’re talking about white light, not a theater projector image, inversion doesn’t hurt anything at all. But, for such things as RGBW emitters, where the color needs to hit back on its own square die, not on the opposing corner, Wavien had claimed to have a solution. It was a collar that had two reflections (one internal) instead one, so the image was basically re-inverted before it went back to the die. That’s the one that wouldn’t have been a hemisphere but some other shape.

Guys, I’m not really all that knowledgeable. I just know how to read. Everything I’ve said on the subject of light recycling is info which is available in the wild. Having read some of Wavien’s marketing material, and MEM’s threads on the subject, as well as various build threads which utilized RA collars, it all makes perfect sense to me. Of course, one would need to understand some basics about light, LEDs, optics and mirrors in general. All of that is search-able as well. Almost everything I know that is relevant to this discussion, I’ve learned since being a member of BLF.

“At least take an interest, and please discuss.”

That would be all well and good in a thread you started about laser projection, but of course this is MEM’s thread about the Tuning Process of Reflective Apertures which of course, incites interest in the application of refining the output of an LED and is why most of us are here.

12V car laser headlamps are in their infancy, developing drivers is something perhaps YOU could be working on so we could get a head start on development, perhaps you could start a thread on that… :wink:

Quite a lot of audacity, don’t you think? To hijack another’s thread on the premise that you decide what is relevant? Rudeness, incarnate.

Chalk me up my first rude point, but I for one am very surprised you’re still here.

Whether the admin see’s fit to allow you to continue the turmoil, I no longer wish to see you or your comments. Goodbye Sharpie.

Seriously? :person_facepalming:

Not that I am defending you but threads often goes off topic on BLF. This is interesting technology and please open up your own thread. Hope you could demonstrate that…

MEM already demonstrated his RA.

Good lord, 44 new replies!

There seems to be only one person trying to tell everyone here this doesn’t work or make sense, or that it’s a bad idea because the best approach is to simply design optics correctly, am I reading that right? Somewhere, Dr. Jones is laughing at all of us. :smiley:

Digging through the internet attempting to find other methods of recycling tests is a rather mundane way to disprove something when the data is not even getting close to duplicating the arrangement scenarios discussed herein. You’ll see that I don’t need to disprove my methods, simply because I cannot. I would have to outright lie to say they aren’t more effective than correctly designed optics used alone—to a great degree.

To start right there on the objective end of the system—the optical end, yes. Designing the optic system ahead of a reflective aperture is certainly how you improve the entire system. Yet they are still independent systems as much as a battery is not an LED. I have some lens sets which correct chromatic aberration as well as spherical aberration extremely well, and they are some of the best corrective arrangements of optics I have used when placed in front of an LED for finite-infinite conjugate projection (cost for those is about $1,500 for 75mm sets from optic suppliers if you can find them cheap, so a bit pricey). They certainly do a job like few other lenses can, and I have not found aspherics alone that can perform as well. I’ll come back to that and explain why I even mention it.

Here are some basics which will help tie together the whole matter. A dedomed Cree emitter as most will have witnessed in very many standard aspheric lens arrangements, will typically have pinkish-red, “purple”, and deep blue fringing extending outward around the projected die image, respectively (there is no purple light in an LED though, technically it is red partially overlapping blue, to clarify). You can often see that the die image has various degrees of “whites” and typically will see a greener white towards the outside edges within the projected die image. Again, this is chromatic aberration, and it is caused by the broadband spectrum of colors in a white LED passing through a typical lens. An aspheric lens, even a very good aspheric lens, is only considered to be so good because of its ability to collimate narrow bands of wavelengths without presenting much spherical aberration at all. More of the lens is able to display a small incoming source image (AKA an LED) across its entire surface area when viewed head on within the focal range, so more of the lens area can actually be used in an aspheric lens to pack light into the main image beam. Spherical aberration reduction is why common aspheric lenses are used in projecting flashlights, period. The below photo depicts what aspheric lenses do not solve, which is chromatic aberration:

All of the excess light beyond the die image above which is red to blue, is effectively useless towards maximum intensity, as it floats outside of the primary die image. That is a good chunk of available kcd and throw which could be boosted if that excess light could be packed inside of the square rather than leaving as non-collimated blur.

Spherical lenses have a problem. They have a single radius across their surface. This means that the focal length will remain the same when starting at the center of the lens and traveling towards the outside of the lens, but the rays getting to those places travel different distances. When there is a small image in proportion to the lens diameter, you can picture a triangle of light coming from the LED with two equal side lengths extending to the imparting lens face outsides. Those longest legs of the triangle travel to side left and side right of the lens. However, the center rays from the LED going to lens center would split the triangle into two right-angle triangles, and becomes known as the adjacent side. The adjacent side is always shorter than the hypotenuse in a right angle triangle. Lol. To say that much more simply though, the rays going to lens center are always traveling less distance than rays traveling to a larger diameter of the lens from an LED. So the LED can never “please” all areas of the spherical lens.

The aspherical lens attempts to correct this problem, by using a calculated profile which generates a changing radius across the lens surface. Light rays which travel a further distance from the LED to the outside of the lens meet the lens where it has a longer focal length. Rays which travel a shorter distance meet the lens where it has a shorter focal length. If the asphere is designed and then produced optimally, all areas of the lens will be met by light rays at their focal length, and the entire lens will contribute to collimation of the image. Often an asphere can not fully achieve this, and if held above some graph paper the lines might be distorted at one area more than they are across other areas. Ideally the lines should remain straight like they exist in the real grid.

All that I have explained has so far described primarily the problems that arise from the areas of the lens being at different distances from the source of light. However, the other problem I briefly brought up with the photo was chromatic aberration. This is something which a single lens cannot solve for a white LED, because unlike spherical aberration, chromatic aberration arises from various colors also having different focal lengths. This is the effect that creates rainbows in the sky or through a prism. Purple light refracts much more sharply than red, and the order of this is based on the wavelength, in the order: purple/blue/green/yellow/orange/red. In a white LED, of course you have a lump on the right side of those colors, and a spike of deep blue contained which make up the white light seen. If you focus the lens on the color blue, yellow to red will be much out of focus, and vice versa. Even though the colors come from nearly the exact same place barely as thick as a sheet of paper, it does not matter because of the wide range of colors.

The photo below shows an XM-L with phosphor removed to show an area of blue light which can be focused on. Notice what happens to the LED projection when focused on the blue side, and then when focused on the red side:

I have briefly discussed some of the common shortcomings with standard lenses used in projecting systems. Getting past those hurdles can be tough, but lets say you have a fully corrective system that not only corrects for spherical aberration, but also corrects for chromatic aberration, like the lenses which I initially discussed having in possession. Those use a spherical correction element to align incoming light, a large achromatic doublet to correct chromatic aberration, and a final aspherical element to correct what little error is left at that point to achieve maximum source image collimation. Even though the lenses themselves can provide a highly concentrated image with very good contrast ratio, they still suffer the shortcomings of all lenses. A lens always has an acceptance angle (NA) of maximum light intake to achieve collimation. Simply put, this is where the benefit to intensity hits a brick wall, as it falls back to the source producing the intensity, the LED. The lens system can become more efficient, in that it can gather more light with a higher NA value, output a larger beam with more total lumens, but the intensity/throw gain itself stops growing at this point. But there is still unused light within the system not being used!

An LED die surface should not be like a mirror. Anyone who understands collimation, would know that if the LED surface was like a mirror, the light would simply strike and leave un-collimated. A flat mirror is nothing different in essence from a clear window. The only difference is the direction of light comes from the front of the window instead of behind it. Spherical mirrors however are a much different story, as they indeed have a very precise focal point. That is exactly why it is so critical to align the spherical mirror’s return image right back onto the LED phosphor. A mirror does not suffer from chromatic aberration like a lens does.

Someone was also claiming, as I recall here recently, how spherical mirrors are so bad and just retro cruddy stuff. Luckily I get to teach why that is untrue. :wink:

The setup used in a reflective aperture is by all means, not an infinite-finite conjugate system. Rays are not arriving parallel from an infinite source, and attempting to meet a finite point anywhere. The system in finite-finite, and it’s well known that spherical mirrors can do this type of focusing incredibly well. Just how well is what I will provide data demonstrating in the ray traces which follow. Watch for how closely the rays align on a return pass near the source plane they are dispersed from initially. If you do not see rays bouncing off the mirror next to originals, it is because they are traveling the same path as they came from, stacked.

First I will demonstrate what a spherical mirror does with light rays coming from a centered point source. I have light beam count set to 18, dispersion beam count set to 3 for the colors blue, green, and red. (However, you will not see dispersion beams in the mirror ray traces, because again mirrors do not produce chromatic aberration. :wink: )
Mirror radius = 20mm, source distance from mirror surface = 20mm:

The next image is where it gets interesting. This is what occurs with a source light offset of 1mm under the spherical mirror. This would depict the scenario of using an XM-L2 or XP-L die, where light originates from one side and lands 2mm away on the other side:

Here is a zoom-in on the previous image, so you can see where the light rays exist where they would meet the LED surface, they are just barely above focal point at the die with 1mm offset, but so close that the image is still sharp. Each full block when zoomed now = 1mm. The 4 visible blocks would represent the entire area of an XM-L2 die. Based on that certainty, notice for yourself the amount of light which goes beyond the die compared to the size of the die itself (black arrows represent termination/reflection points):

If the above looked like something tragic happening at maybe 0.1mm of excess die coverage being lost, look what happens when the mirror is simply lowered 0.025mm. The furthest points from center come back into nearly complete focus, and since the other points between them are under this distance, it becomes quite possible to almost completely saturate the die with its own image. Again a full block = 1mm:

A parabolic mirror does not do the same thing, that’s right! A parabolic mirror does not invert the return image. The offset point under the mirror which it came from, is the place it also attempts to come back to. Unfortunately all I have learned from parabolic mirrors is how bad they actually distort the return image. It’s skewed, like someone stretched the corners out, and definitely not as efficient as a spherical mirror in a finite-finite focus system made to stack a uniform image. Until dies are shaped like concave domes, I cannot see how they could ever achieve what’s being covered near perfectly by spherical mirrors making a return image.

Light does not have to land on a mirrored die surface, when the die surface of an LED produces more green and yellow light just by conversion of 450nm light, not to mention growing brighter by it’s own excess light that isn’t converted, still glowing just be saturation of the colors it makes. If it were mirrored the light would bounce and bounce and bounce, yet it would not leave in collimation. The image projections from my lights clearly show there isn’t a scatted cluster of light around the die, the only thing lit is the die. The area around the die is completely dark, and I find that to mean the light is landing where it needs to be, or the red MCPCB would be incredibly lit up from hundreds of excess lumens landing on it. I have done that a few times, where focus is not spot on, projecting a backwards “NOCTIGON” out of the light. I usually fix that in a hurry though. :smiley:

Luminarium made a completely relevant contribution well beyond any elaborate theory. How many people are making well over 1 million cd in lights using lenses between 65-75mm without using a reflective aperture? Please do share when someone does this with “good” optics alone. I’d be glad to buy that set! :student: