Mega Kcd Aspheric

So I have been doing some experimenting with aspheric lenses and have decided to build a run at a mega Kcd thrower in the machine shop. With aspheric lenses of different diameters, a DEFT Enthusiast I own to study, and my aspheric lenses having differing focal points, the only real conclusion at the end of the day was which LED actually makes a better candidate. That's not even news, I don't believe. Heat sinking aside, of course the Cree XP-G2 92cri bins seem to be the emitter of choice for a lightsaber handheld aspheric. Whatever host it need be (2x26650/3x18650 hopefully, or minimal 2x18650 host), I will mate it to the components. I am thermally potting the driver, and turning a solid step-down-cylinder shape copper post to center the emitter die on near the focal point. I will remove the Cree LED protective lens, also.

My subject here is more shifted towards what the aspheric lenses are doing. Also, what lenses on LED lights are capable of doing for the focus of the LED projection in aspheric lights, but I use lasers here at times only as a basis for comparison in their optical properties, which are shared mostly by all lights.

In an optic system, we can compare what I see as being similar to what one might desire in an aspheric flashlight. A laser. A laser that emits light from a 0.01x0.01mm laser chip inside the diode. A single, 5mm aspheric lens is used to collimate its laser light into a focused laser beam, just like a focused die on the wall in an aspheric light is seen to be clear when in focus. The beam is small in the laser because the emitting surface is also small. Yet, the laser still has a divergence in the beam; it spreads out over distance. So, what is done is a "beam expander" is used. In this add-on beam expander, two lenses cause the exit laser beam to be much larger in diameter at the light exit, but it's divergence is then extended by the power of the lens, I.E., 10x for a 10x tighter beam. In other words, the light density is less when up-close while using the expander lenses, but the light density eventually becomes greater than the original light at a far distance and the "dot" can be seen on objects at a further distance. It's like shining a laser right through the back of a spotting scope!

The principle is the same with an LED, is it not? Say the light is focused, expanded/corrected, then refocused again to make its projection reach a farther distance? The difference between the obvious two is that the laser emitter is only a slice of the scale of the emitting surface of an XP-G2.

So, with all that said, is there a clear-cut reason this same type of beam expansion is not seen on the current mega-throwing flashlights like the DEFT-enthusiast or even DEFT-X? Why not take the output of an aspheric light, expand it to 3x, and obtain a light with 3x less beam divergence? Or, why use only 1 lens, in other words? I felt maybe, just maybe, there is some unforseen way using the right set of optics, that could make Kcd go through the roof like the laser princible. Apparently, I am not understanding the full physics, maybe the lenses' positions would need to create a light the size of a telescope...

More ideas here for a super thrower.

When is big too big, for an aspheric lens? Can it ever be too big? Say instead of a 35mm aspheric and an XP-G2, one used an XML2 and a 75mm aspheric. I would expect the total output area to be large from the XML2 of course, but would the light be any less dense over its covered area than the G2? Even similar? Of course, the LEDs being driven at their higher ends of amperage capability.

This has been done a couple times more recently. 'mash.m' has a sales thread for a light employing a lens combination similar to this. I too work with lasers in microscopy, and I use similar tricks to change the diameter of the beam at the objective.

For sure, it can be done. You should work on getting a working prototype. I too have stumbled down this road. It does work, but it also comes down to single lens pairs (if you order cheap from surplus lab type stuff) or bulk orders (expensive to do yourself, and you had better know you got the lens combination you want if you are ordering 50).

I have a few smaller aspherics using condenser lenses from old teaching scopes. At least one of them is quite impressive...but I cannot make another because I wouldn't know how to get another lens.

I do think to optically achieve more than what a Wavien reflector collar can is probably doable. The question is whether you can get those components together for a reasonable sum.

obi

I have a feeling I can get lenses. When I start to understand the entire concept, I will hunt down, through USA firearm replacement optics, HID replacement optics, through China, or whoever I must, to end up with the parts. Even custom coating custom ground glass is not out of the question. I guess you could say I start digging deep when I need a specific part. This light, this is the main light for me. This is it. I'm building the hardest throwing light I can acceptably carry (on a car seat, JK!). Also, there are no constraints I have placed on the light source such as, "It must use an XP-G2 because that's what the other guy did who made it work". In my experimenting, I picked up some XPG2s to see what the fuss was about. Also have XMLs, and an SST-90 I wouldn't mind driving hard in this. If I can use the right optics to achieve a small projection from a large die area, well, we all know Luminus has some huge lumen numbers. I chose an XP-G2, because I was under the impression this may require too much optical scale with a large-die LED. Based on some really awkward multiple lens testing I have done, I have found I can achieve smaller projections with certain lens combinations vs 1 single aspheric lens at infinite focal point. Some results showing room for improvement over single lens. What I really want to focus down, "expanded" so to speak, is a Luminus CBT-140. 50W+ LED with all that from 14mm^2(!). Sure it's a big LED but this is why I am here talking about lenses. I have a 75mm aspheric I'm getting ready to test soon with an adapter tube over the SST-90 to see what kind of throw it's doing with various lenses. The trick here is just getting a big LED to throw like it was a small one. So easy to say, eh?

I realize A/R coating is extremely important here with LED light passing through a lot of lens material with this many surfaces of glass. Yes, there is more loss of light of course. I however feel this can just be made up for in another way (cooling/more power). I indeed will not use the lens in actual prototype form until I have the right nm A/R coated glass that is worth the time spent putting into the prototype.

Swapping out a small diameter aspheric for a larger diameter one with longer fl is a form of expanding the beam. No need to collimate, then expand, and then recollimate. you could "precollimate" as a way to use one lens up close to help collect light that would otherwise be too wide to hit the larger, main collimating aspheric.

mash.m just recommended a lens simulating freeware I'd like to try, but might as well pass on the tip. It's called winlens.

I did a prototype of that concept here:







I merely shined the "Not-that-hard-throwing" ZL SC600 through an ordinary loupe - with the focal distance of the loupe optics adjusted until the LED projection was pretty clear (You can clearly see the LED wires etc).

The two lens loupe is doing what you are looking for...essentially the backwards telescope shot.


Obviously, I was only doing a proof of concept, and the ring around the LED is reflected spill from the metal of the loupe lens frames, etc...which I'm sure could be cleaned up with a cleaner optical system (Cleaner than a ZL SC600 and a 2 lens jewelers loupe laying on a ping pong table, etc...)





To really get things pumping out the cd...my next experiment would be to see how MANY LED I could get to hit a reflector and first lens, that sends the more tightly focused light to the aspheric, so that the aspheric lens "sees" a teeny point of focused light as the "source".

With the source minimized, the off center output is minimized, which should maximize the throw.

This, to me, sounds like what you are wanting to try....except I think getting a LOT of LED input would maximize the starting intensity...and require that these emissions be combined into a point source.

My idea was to focus all the light onto the effective surface to be my "point source" the way a kid would focus all the scattered sunlight onto a hapless ant.

The "ant" is then the point source that the aspheric sees and projects a picture of out into the darkness.



Try that.

Haha...

The loupe on the table! My final experiment this morning before I came to post. I tried that, what you did, with a 3 lens loupe a little bit ago just to rule out that it doesn't help "my situation". More lenses = a bigger and bigger projection at distance as the focal power increases, and the focal point becomes closer, diverging faster.

I've been playing with lenses and LEDs for the past 3 days after work just like you are: using my hands, a reflector-light or just emitter on heatsink, lenses, and of course a wall (by the way that's your orange peel reflector image on the wall, not the lenses doing something strange. ). It struck me clearly how to do it when I started mixing lenses to see an adverse effect with multiple aspherics to rule some thoughts out and off the table. I now know what will work. Two lenses, in compliance exactly with what bshanahan14rulz stated.

Owning a DEFT enthusiast, I have a baseline "megathrower" to compare to that obviously someone had some idea what they were doing when building it. Of course, like stated earlier, its throw comes from a dense LED source that is very small (XR-C), allowing an aspheric that is probably ~25mm diameter to be used, and the scale of the LED to the lens ratio works out for an efficient hand-held. But the XR-C we know puts out very low lumens, so my thoughts came down to lumen density at the die, which is just taking lumens over die area to get lumens per mm^2. So if we assume that the XR-C was highly driven to say, 150 lumens, which I don't even believe it makes (rated at like 80 lumens IIRC), 150/0.7= 214 lumens per mm^2. Here is where the XP-G2 seems to make sense, say it's heatsinked well, driven to a whopping 600 lumens, it's a 1mm^2 die so there it would be 600/1mm^2. Throw does get massively more dense with a G2. An SST-90 driven to 2500 lumens / 9mm^2 nets still higher than XR-C, 277lm/1mm^2. That tells me right there, the SST-90 has potential. It could potentially throw out about 25% more light per area, yet it's netting almost 20X more total lumens due to die size. Meaning that with a large enough lens, the Kcd could go right through the roof when focused.

After knowing that, the experimenting began with the SST-90 and other lights/lenses to get a feel for how much can be accomplished.

I took a headlight apart from DX that just came in the mail. Just an XP-E Q5 emitter with roughly a ~20mm aspheric plastic (zoomable) lens on a headlight strap. The light was amazing to me because, while not near the DEFT in output, it was still lighting up water towers pretty darn well when focused, just with its plastic lens. I sort of used that as an acceptable baseline to increase from in my experimenting, because this is somewhat new to me, I am more used to lasers which basically emit a point light source, which actually is it's focused "image" of the emitter--although it cannot be made out as anything more than a point up close, due to the light density and small size of a laser chip.

I read some physics websites about lenses, to reaffirm what is happening through concave and convex lenses so I can understand the light ray behavior, and how they invert and focus to redraw an image on a new plane of distance etc. Then I did some visual experimenting using the bright room light overhead as the 'point' source to see what a focal point is in the inverse effect, then studied what the LED under an aspheric looks like in perspective to the distant user looking into the lens from the (+) front side to the (-) focal side--where the LED exists. To put it the way I read it: A lens simply put, has the inverse effect of its shape as if it were a mirror. The focal point light is a (+) side image after refracting through a lens, or beyond the lens, while a mirror of course is the (-) side, since the light comes back to the source, being reflected of course. That put it into great perspective for me. I could then visualize what should be needed to defeat the goofy-to-me "wavien-ring" or whatever, spherical mirror. I mean, I get their philosophy in principle of what they are trying to do. The wavien ring, if understood correctly, merely attempts to do two things, (-)focus, and aperture effect. Problem, make it work right in a lens array, and--is it really worth the troubles in the end of setting it up? I find lenses to be the proper way of getting the LED's light out of the front, but all of this assumption on the wavien ring is only in theory. I don't have one to see what it does in practice. Nor a blue-print, no drawing, do not know how large one should be, etc, etc.

The idea of gathering all lost side-spill light, and trying to aim side-spill light back onto the LED die is not really amounting to much in theory, to me. Why the debunk theory, though? Because I did one small experiment. I took an XP-G at ~200 lumens, focused it's image onto the SST-90 die with a lens straight-on, and all it achieved was a bright yellow die chip you could easily look at from a foot away without burning your retinas. Just having the LED on and emitting 50 lumens per se, that's hard to look at head-on without squinting. This was not. Putting the extra light merely aiming back on the die, the increase? It's like nothing. It's also yellow light being reflected, as the main emitter doesn't stimulate that way anyways to produce light particles. In this experiment, one is basically just aiming light back at a "screen" which light particles do reflect from, but that "screen", being the yellow emitter chip, it's not a mirror it's a yellow chip! An LED is stimulated by an internal quantum mechanical reaction of the electrons exciting into photons in the "Q" or "active region" between the P and N junction layers. So it's not exactly at it's surface where you see it appear at. What I'm attempting to conclude here is that shining a chips surface doesn't stimulate more photons to emit, it's merely a yellow backdrop for light already emitted, that can be projected by the lens. If the spill light does not re-focus straight back onto the emitter chip, you're going to see some ugly refractions of reflected light hitting around it and the bond wires for example. If it can even be seen that is after all is said and done.

The LED needs a quality low focal-power lens to reduce it's initial light emission angle we know that much. Stuck at this lens choice somewhat. I plan to order lenses that are quality and coated for AR. The vehicle projector lens is more for testing you could say as its front surface is sort of like micro-orange peel when viewed through the lens and seeing the inner surface the light hits on the way out, it looks cheap for glass.

Concave lens would widen the beam. So it must be a mild convex needed very near the LED, right? Then the convex would need the appropriate focal point. Shining an aspheric into another aspheric can be done, but I believe you want to hit the exit lens with as much surface light as you can without getting too close to the edges where it defracts badly and remains trapped in the lens (so some aperture type is needed). Then the lenses and LED would all need to align to retain image quality, I.E., focal lengths would correlate in some way.

I own a shop with lathe so PVC spacers out of the tube stock work well and are quick to make once I find the correct host body. I feel it's probably going to end up looking like a movie projector. A box for a large heatsink and batteries + fan for cooling, and a large head+bezel from a host light placed on the front, threaded or glued into a cutout hole on the project body. Anyways, to get this lengthy post over, here is the test bed I started on the bridgeport for this light testing. It is currently just used as an aluminum heatsink for the different (16mm and 20mm) MCBs to swap out on quickly while testing. I am making a telescoping tube by tomorrow out of phenolic thin-wall tube sizes that fit together snug. I will probably just glue the telescoping tube during testing to play with lenses and focal points while driving the LED from my DC lab supply until I find the lens pair match that basically sets the build into motion.

Here are the album photos I uploaded on Imgur.

75mm Aspheric Project:













BTW the heatsink block in the photos under the emitter is 10x40mm 7075 aluminum bar stock. Approx 120mm length cut. I just drilled and tapped it on the knee mill (26mm CTC distance between screw holes, used 10-24 machine screws). The SST-90 heats that whole block quick at just under a measly 5A. I was thinking. I have a nice small copper waterblock with 3/8" nipples and 4 corner holes for securing. Mount it to the rear surface behind the SST-90? Hmm??? ...Also have a small waterpump for a PC, and I have two ≤60W TEC coolers that have strong cold sides. I have to add a lot of cooling for this to run for very long. I may try the TECs or I may just use a copper active fan cooler.

Well, the movie/slide projector concept is what the aspherics are doing alright...projecting an image.

The infinite focus part is where the optics differ I think...as I think the projectors I've seen at least seem to have a maximum range.

I know you said you are using what you have in terms of available aspheric lenses, and the ones I have are the ones with concentric circle texture, otherwise I'd send 'em to ya. One of the more highly regarded lenses is the ZKW lens, named for the manufacturer of the headlamp that this lens was originally found in. It's 3" diameter, but from the pictures I've seen, it's got a longer FL. But it's most sought-after quality is because it is water clear. There are replica lenses available too, aftermarket lenses modeled after the original projector's lens, but made with a clear, rather than textured surface.

Idea behind wavien is that there is only so close that you can put a lens on an LED. You WILL have light lost. What harm can it do to recycle the light? Your experiment with XP-G and SST shows this.

edit: Oh, and thanks for sharing your project with us! Let us know if you go with the Wavien or instead find a suitable precollimator. I think I remember reading somewhere here on CPF that a long FL plano-convex or meniscus has been used with decent results, but I can't find where...

Here's what's happening. I can further magnify the light using two lenses, but even though light is staying within the lens' aperture (L1 PCV and L2 ASPHERIC), I appear to be loosing light. I placed a medium wavien ring over the emitter and shined a backdrop. Took photos of it striking only the 75mm lens. Holy freaking hotspot! Then I placed a 31mm OD plano-concave (-33mm FL) lens right against the wavien ring front for minimal aperture size light entering this lens (PCV plano facing ASPHR plano). Now, the Galilean formulas state there will be a deduction in FL using a (-) lens before a (+) lens, so I had assumed the lenses would get close together. I think that because the L1 lens is not at 33mm from the LED (as it's focal length is), but rather about 12-13mm, the overall FL of the lens system grows as this difference, to achieve full focus. The total length between LED die and BFL of the L2 (aspheric) gets longer, to state it plainly. Seeing about 60-62mm of BFL instead of 52mm, overall. But maybe this has to do with the thickness of the concave lens medium being there? It's edges are probably 12mm thick. Center is maybe ~5mm thick at minimal.

Now here is the good part. The light leaving the ring, through the concave lens is expanding and hitting the 75mm lens with about 65mm of light diameter when the system comes into focus. This is good, because aspherics do no like light near their edges. So all of the light appears to be gathered, and entering the final lens instead of being larger than the aspheric. A further magnification also occurs, the hotspot at 2 meters gets about a 30% reduction in diameter. I'm thinking, this is very good, as I can make a 4mm^2 chip act like its a 3mm^2 chip in output diameter/diversion, roughly. The chip is also in full focus when this occurs, so the lenses do work together.

Here's the bad news which makes me frown. First of all, I tested the light at different levels through 1 single lens (+ wav ring). Just using the 75mm aspheric as a baseline for Kcd. Although the light reduces in diameter at 2 meters, the hotspot is NOT increasing in brightness by any certain factor. In fact, it seems to just have a smaller image of the same overall Kcd. This is where the whole idea breaks down. I don't get it. I almost wonder if the angle of the concave is causing light to get trapped in this lens, because the len's edges appear to become rather bright (which, some is expected to be trapped anyways). So I can't figure it out. It's indeed focusing to a smaller hotspot, but the hotspot does not become brighter, and I am seeing the whole LED image, not an aperture of the image. I even used vapor to see the light expanding between the lenses to see if something was missed, and it is falling fully into the aspheric lens. ??? Scratches head.

Anyways, I know that I can reduce the hotspot size and increase magification, but I need to figure out where the light is being lost (obviously appearing to be in L1). It could have to do with a larger light incident angle at the aspheric-plano side, too. So to further test this, I am ordering an array of larger, coated, grade-1 negative lenses. This time they will all be larger than 35mm at minimum, and I'm trying FLs longer than -33mm. I.E., -43mm, -48mm, etc. This would mean the expansion of light from L1 will be less, magnification will be less (hotspot will still be smaller than using no L1), but if I can get the hotspot to strengthen as it becomes smaller, I am on the right track I believe. I am going to try double-concave lenses also, with plano-concaves of same -FL in comparison, to see what this effect has on the image (always being in front of the recycling ring, of course).

Something has to be going on that's physically stopping the light, if it's projection is fully striking the asphere. It could very well be due to an uncoated -F1 lens. Or, too small an aperture of the negative lens (even though the collar has only a 12mm exit aperture, this can't be ruled out just yet without trying larger diameter lenses).

I now have the light head, 2, in hand, for a 78-80mm objective asphere.

BTW; taken from an internet published source, these are the Galilean beam expansion formulas (As you will see below, CVX + CVX lens system is a "Keplerian" formula [longer focal length, more room needed, less optimal for LEDs is my assumption, but a higher image quality with less aberration is known to be the outcome] --So I won't rule it out either as an option for optical arrangement, since I have the focal room needed in the custom large-head design.):

Galilean Divergence Formula; demonstrates a value of image size that is equal to the ratio of the focal lengths in the form: |-f1|/f2.

Plane 1 is: "negative F" ([-F absolute value] /F2)


I.E. Source divergence[(X/1)]*Focal.Ratio.Absolute = Final Divergence
Factor.

The expansion ratio:


y3/y1 = θ2f2/θ2|−f1| = f2/| −f1|


or the ratio of the focal lengths of the lenses. The expanded beam diameter


2y3 = 2θ2f2 = 2y1f2/|−f1|


The divergence angle of the resulting expanded beam


θ3 = y2/f2 = θ1|−f1|/f2


is reduced from the original divergence by a factor that is equal to the
ratio of the focal lengths |-f1|/f2. So, to expand a laser beam by a factor
of five we would select two lenses whose focal lengths differ by a factor
of five, and the divergence angle of the expanded beam would be 1/5th the
original divergence angle.


As an example, consider a Newport R-31005 HeNe Laser with beam diameter
0.63 mm and a divergence of 1.3 mrad. Note that these are beam diameter and
full divergence, so in the notation of our figure, y1 = 0.315 mm and θ1 =
0.65 mrad. To expand this beam ten times while reducing the divergence by a
factor of ten, we could select a plano-concave lens KPC043 with f1 = -25 mm
and a plano-convex lens KPX109 with f2 = 250 mm. Since real lenses differ
in some degree from thin lenses, the spacing between the pair of lenses is
actually the sum of the back focal lengths BFL1 + BFL2 = -26.64 mm + 247.61
mm = 220.97 mm. The expanded beam diameter


2y3 = 2y1f2/|-f1| = 2(0.315 mm)(250 mm)/|-25 mm|= 6.3 mm.


The divergence angle


θ3 = θ1|-f1|/f2 = (0.65 mrad)|-25 mm|/250 mm = 0.065 mrad.


"For minimal aberrations, it is best to use a plano-concave lens for the
negative lens and a plano-convex lens for the positive lens with the plano
surfaces facing each other. To further reduce aberrations, only the central
portion of the lens should be illuminated, so choosing oversized lenses is
often a good idea. This style of beam expander is called Galilean. Two
positive lenses can also be used in a Keplerian beam expander design, but
this configuration is longer than the Galilean design."

---------------------------------------------

As you can see, |-F1|/F2 is only valid for beam size reduction/expansion when |-F1| is a lower value than F2. If |-F1| becomes larger than F2, the effect, from what I can gather, would obviously break down, and the image would be widened, not narrowed.

One important other thing I will mention in my findings, a negative lens, although having a focal point value, has no "real" focal point on the LED-side. As you move the LED further from it, the beam magnifies more and more, but final image focus is not lost. It is a value effective on another lens, not a source of light behind it. LED placement behind it has no effect on focus at the final lens. So, distance between both lenses is the crucial factor for focus in the system, not distance between LED and -F1. This can be seen with your own eyes, as the die looks smaller when you are in front of the negative lens, looking inward at the LED. The more negative it is, the closer to 1:1 the LED looks when the LED is considered near the negative lens, and the less negative it is, the smaller the LED looks.

There was another thread that recently discussed why its impossible to focus the image of the LED onto a smaller simulated point source to increase the apparent brightness.

Intuitively, it does seem that this is possible, albeit, it is apparently not.

A laser can do it. It's called a beam expander. I shined a laser through a spotting scope, beam grew 20x larger, dot was visible on objects 4 or 5 times further away to the naked eye, limited by the naked eye. A laser is a much smaller emitter source though, and mostly coherent light. I think it in fact could be done, but the scale might be the same, for instance. Say an LED has an emitter 100x the area of the laser, so would be the lenses to replicate the effect. Now, if I am able to make a hotspot smaller, but it does not grow brighter, I am in fact expanding the beam. This means that the divergence IS cut down. So while light may be lost, that light will stay together at a smaller angle. Meaning at some given distance, the less efficient lens/LED setup will actually become brighter than an aspheric alone. The cost is paid for at near distances, but made up for at far distances. The question is, how far? 1000 meters? The hotspot becomes less useful because it is so small, all it can illuminate may be a 6' area at 500 meters. What I'm doing is not just making the hotspot smaller with the same divergence angle, as is the effect of say an XP-G2 vs an XM-L2 through the same single aspheric lens or reflector, what I'm doing is physically reducing the beam's output angle. An angle reduction is on much different ground than a hotspot size reduction, which keeps the same emitting angle. I think the truth is it's highly inefficient to beam expand an LED due to the optical systems size in scale to the die size (vs I.E. a laser), but not necessarily useless/can't be done. How much, is the question I'm trying to figure out. How much loss in near apparent light, is beneficial at a far actual illumination distance, when an LED is used... Hence, I am only trying to reduce the beam angle by a small amount, so that the light loss is not immediately "bad" for the throw. When beam angle is reduced, emitter size is still important, but total lumen output seems to become more important than smaller die sizes due to the angle of beam decrease. So it's counter productive at near, more productive at far, as the theory goes. But only photos are going to prove this, and right now, the lenses are all mounted to my mill table. I need to do comparison shots from my deck, to demonstrate the true effect. It might need to get a little warmer outside before I go out on the deck and do this.

matt304 said:

I'm building the hardest throwing light I can acceptably carry (on a car seat, JK!).

Click to expand...

What is it actually that you want to do? The best portable led thrower of course but how about the other qualities? Would you be happy with a very narrow pencil beam? Should it be wider? How wide?

Also, what do you mean by "acceptably carry"? For example, would you consider Databank 70 to qualify? Deft Enthusiast produces 86kcd. If you built a databank style light and managed to get 50kcd from each led, it would add up to 3,5 Mcd. The ANSI throw would be something like 4 kilometers.

I have a deft. Most of the light doesn't aperture to the lens correctly. Still a darn great little light package all stuffed in a Uniquefire R5 body!

My design so far is actually more "compact" than you may think by initial description (I didn't want to give the impression this was trying to be a small build), but I'm going to call it large for an average "big light" (SR-90 as an example, this is bigger than). I don't think I'll ever have competition against the Databank 70 (that's very extreme, and you're definitely on-point with your suggestion if I wanted to do it that way). My build I am R&D'ing you could say, has an incredibly different design goal. "One large emitter, but very low beam angle". Hence the two-lens system experimentation. I know that an LED, an aperture recycling ring, and one quality, coated lens, is as good as efficiency gets for the most part, in a one-emitter system. I have some experience with lasers, and some experience with laser optics, but optics applied to LEDs is probably the newest thing I have taken on, that I would call a hobby of its own. I realize it is incredibly possible to improve upon various Kcd figures by using an energy-dense LED die/bin (models like XP-G2, XM-L2, XP-E2), a big quality lens, and very efficient cooling to keep those lumens going, while of course supplying enough amp-hours of battery to get some run time. I also realize that the bigger the aspheric lens and longer the focal point, will give higher Kcd, but you have to fit that in something. My task is finding a balance, where "big" is not "HUGE", and the head of the light is not 8" in diameter (HUGE), but rather about 95mm in diameter. Limiting the head diameter restraints the whole system's beam angle possibilities, because now that light design becomes focal-length limited, and also beam aperture limited, to actually end up hitting the front lens with all of the light. I think that says nothing everyone here does not already know. But beam expansion, is what I am working with, or attempting to. Taking a given size limitation in a light, and now finding out how small I can get the beam angle, before diminishing returns kills effectiveness. I have a theory that a beam can be expanded in an LED system, just as it can in a laser system, though not as much. However, this could amount to something big in the world of single-emitter lights, if a point is found where the beam angle is oddly small for the size of the emitter die, yet say it is still yielding a high total lumen output within that small beam. Yes, I may run into problems as I try to defy convention. Yes, this may end up being just a big, average-throwing-for-size light. But so far, I am seeing something occur which looks like it could be promising if "tuned". I may be traveling the same path many others have realized lead to no-where, but I at least want to see what my own tests yield. So far, I am quite happy with the initial beam angle for the die size. I just think it can be tweaked a little more to advantage.

What do I want to do? I want to find, through my own experiments, where the diminishing return occurs in a two-lens system intended to throw far. I want to find the limit of current emitters, and their ability to throw, one LED source at a time. Some behind the scenes stuff is also going on, based on a patent I dug up, linking some ideas together. The patent revolved around lasers of a certain wavelength being tested through a certain lens configuration etc, that related to LEDs in their testing. My real underlying goal here is still masked by initial fidgeting going on above. Once that is over, a new stage begins. For now, that is all I can say about the light, but so far, with the help of a friend, we achieved something a little bizarre you could say. It blew our mind for the time being. But we're not sure we can really pull it off on our own. However, as I said, testing in that stage has not yet officially begun, we just know one thing: "it" worked in our bench setup. The setup is similar to what was recommended with the Databank, you might say, but light only exits one final lens system. I'll leave the rest to speculation for now.

matt304 said:

I also realize that the bigger the aspheric lens and longer the focal point, will give higher Kcd

Click to expand...

Longer focal point? I suppose you mean shorter focal point? If you can place the led closer to the aspherical lens, more light will hit it and be directed to the beam.

Secret solutions that you can't discuss yet? :thinking: I guess we'll have to wait then. While doing that, I'll throw one more idea into discussion. Have you considered recoil setup with an aspheric? A led facing backwards, a reflector to collimate the light (the focal point between the led and reflector) and an aspheric lens to direct the light forward. I don't know how that would work but it is something that I would also consider (think in more detail at least) if I had the same goal as you.

Esko said:

Longer focal point? I suppose you mean shorter focal point? If you can place the led closer to the aspherical lens, more light will hit it and be directed to the beam.

Secret solutions that you can't discuss yet? :thinking: I guess we'll have to wait then. While doing that, I'll throw one more idea into discussion. Have you considered recoil setup with an aspheric? A led facing backwards, a reflector to collimate the light (the focal point between the led and reflector) and an aspheric lens to direct the light forward. I don't know how that would work but it is something that I would also consider (think in more detail at least) if I had the same goal as you.

Click to expand...

Actually, I did mean what I said about a "longer focal point". I'll post a drawing below explaining why this is technically correct, in the system I am using now. It's all due to using an aperture. The light output--when under an aperture--has a physically defined path that doesn't wildly stray at any distance from the LED. It simply has an angle that is constant, and not a large "hump" or "mountain", as a beam shape characteristic. Without an aperture, this is where you would try to capture the largest part of that "hump" by placing the LED closer and closer to a lens (while using a shorter and shorter focal length). If you have ever found an LED flux graph from Cree etc (showing intensity as a sectioned away side-view), you know that without an aperture, light expands at too high of an angle to place an aspheric lens far away, and still capture the central, highest flux area of LED output. However, with an aperture this is no longer the case, or that problem is highly minimized at the least. This is only efficient when the "aperture" is really a backwards spherical reflector, that recombines the spill light on the outsides of the beam to re-draw an LED image back onto the LED, while lowering kelvin temp. A normal aperture plate would not provide this brightness increase through the aperture's hole, it would merely block extra light available, resulting in your suggested scenario of a closer lens. I don't blame you for being confused about this, as these are all things I have learned as of late myself through experimenting with reflective apertures I purchased as prototypes. See video example: https://www.youtube.com/watch?v=W1nD8hvlbbg

The "secret solutions" aren't really secrets the LED industry doesn't know about, but there are a few things we have learned how to do in the shop which so far, have not been found in lights as of yet. You might think, "well then it's probably a dumb useless idea". So I'll say a little about what we have done in testing. Using a "reflective aperture", or "recycling collar", we have learned how to integrate it within a larger optical system within the light head. Because the light beam is now behind an aperture in this setup, there is more room in the optical system that would normally be critical space for light to travel, that is now dark free space. In our experiments using this setup we devised, we can control light output at the chip-level (die surface). We have successfully used a Luminus LED (for large die-area and easier testing of the principle idea), and "controlled" where the wattage was applied within the chip. So when ten total watts are being consumed by the chip, we can apply those ten watts to different areas of the chip itself. Instead of the full wattage being applied across the entire chip surface rather uniformly--like all current LEDs do, we can "emulate" the die output patterns. An example with success of this concept, was applying 5 watts near the large die's center (into as small as <1mm² central chip area), and the other 5 watts around that area across the chip, so the chip acts as if it is two, separate emitters (emulating one stacked on top of the other), yet still only using one emitter plane, meaning both levels of brightness are in-focus as the projected image. This process is extremely delicate to setup I might add. That alone may be a good reason it hasn't been used in other light designs, for it could fail under a high-G drop/impact of the light, or it could over-complicate manufacturing processes of simple flashlights.

The luminous flux created at the center of the large-die LED in our testing is incredibly boosted (its gain in that area is near linear to the watts applied to that chip area), while still utilizing the entire chip surface for light output. This creates an aspheric flashlight, that when focused, has two levels of light output at the same time across the die, just like a hotspot with spill in a reflector style light. Total light output is still confined to the focused, die projection image, as well. There is no blurring effect, etc. This could have use such as: A central throw of a chip similar to an XP-E image size, but a total projected image size of a large-die LED. We are heavily invested in further testing this method, but we have other goals first to clear out of the way.

About the recoil thrower. Yes, I considered that design type myself when churning through some ideas, though not in the way you have described it setup as. What you've said indeed sounds interesting in theory. One unfortunate downside to the recoil design is the cooling ability of the mount post. For the copper post to remain small and not become a blockage of light, it would get very hot with the kind of wattage being used in the light without some form of active cooling (like water pumped through the post such as on giant arc throwers, which is an extreme complication of portable design). An important factor to me, is cooling, because I do want a light to remain stable in its adjusted output level, instead of merely starting bright, and lasting for 10 seconds to achieve a record output. Those are neat lights when done like that for bragging rights (just like the Databank), but not necessarily useful for a search and rescue mission that takes an hour+ to complete. You probably would not want 5 total battery changes to become part of your hour long lighting task. :shakehead

It is true that a recoil thrower setup has a lower limit for maximum current. However, with heatpipe design created from multiple copper bars (2mm*10mm or something) and perhaps with a heat conducting front glass, I would not expect it to be such a big problem.

So, let's see if I got you right. You say that you created a wavien collar/optics system that concentrates the reflection to a smaller area of the led? I am sure that all of us are very much interested in seeing how you managed to do that. That kind of system could have had some patent value, too (unless, of course, it has already been described in the patents that you mentioned).

A little off subject of lens talk, but on the topic of this build coming together finally...The 12.6Vin / 10A output C/C buck-driver test-board design to fit the light, for a single (1) LED mod (NOT an SST-90 emitter--to clear any confusion from my photos), is on the way now. After a few weeks of work, I found an electronic engineer to help me out, and the board required for this project, at 10A output, wasn't "anything too special" he told me. Opposite advice as I was getting here on a 10A driver, so, we'll see. Though, as I had asked before around here a simple question: How does a manufacturer drive high amperage to 1 single LED in a common high-amp product? I.E., the ST90 light, rated 8.2A of current across the Vf, it reminds that such driver boards do appear to exist around us already in "budget" lights. Getting the 10A/30W+ to the LED is not the main problem, it's keeping driver and LED's heat separated, as there will be a lot of heat in total.

The next steps: test and document the cooling capacity of the light-mod variants with this new CC board, catch any flaws in testing, and then hopefully I can get a few boards produced that are improved more, with a better cooling interface. I will get some beam shots against other aspheric lights for initial comparison. I will try to compare it to a known benchmark around here, like the DEFT-X. This chip is likely to perform well due to larger scale of everything, but, "for how long?", is the main concern. I'm confident in the cooling mass in the light design, but I will add more copper in there if I need to cool more. Roughly 3A draw from each 32650 Feilong cell (~5500mAH, 3S1P), should give a decent run-time.

Having fun AND getting some where? Looks like it i'll be watching this thread

Here is some initial testing I did back in December of the circular emitter SBT-70 against an SST-90, with the image results. The LED performance of the SBT is very high for a "big die", that actually isn't all that big. It's as if the chip really emits its light within a 2mm diameter die area. Or, that's the effect witnessed in testing, anyways. SBT-70 is 7mm² die area, as its specs go. However, in testing, either the SST-90 is bigger than 3x3mm, or the SBT-70 is smaller than 7mm² (~3mm OD circle).

This is shining a black flat piece of cardboard, placed at 2 meters from the front lens. Keep in mind, the SBT-70 emitter can produce over 2000 lumens, even over 2500 when fully driven. Everything I'm seeing, including the recycling effect, just seems to get better in a circular-emitter system such as this. A "square" of light, expanding outward, will have higher losses at the corners of the square. That seems like wasted energy to me. It leaves lens area unused at each middle side. By keeping the emitting light within a circular die area, the proof is in the pudding here you might say. The SST-90 is not even getting close to these results, even at roughly the same power per mm². Granted, they are different generations of LEDs, the Kcd "jump" seems quite large to me, going to the SBT-70. Also, the original SBT-70's color temp dropped far more, post-recycling aperture, against the de-domed SST-90 (which was also under the same recycling aperture).

At 2 meters, the SBT-70 spot is nearly no larger than the aperture of light striking the plano-side (back) of the 75mm aspheric lens (which is about a 55mm spot size). The beam looks like a darn white laser with any air moisture, using the SBT-70. Here are the test photos of the SBT-70's performance vs the SST-90's, both in the same optical configuration:

That is great work matt304 !

I really hope that you will beat the DEFT-X !

You are using 2 different lenses in your final project ?