C
curiousCrandall
Guest
Subject: long-throw LED flashlight design
This is an open letter to Peter Gransee, Paul Kim, and all of the terrific experimenters and innovators out there who are making the world of hand-held illumination such a strange, exciting, and fast-moving one these days. I want to propose and advocate a general strategy for improving the quality of the beams coming from the new-generation LED lights, like Arc's pathbreaking LS, and SF's KL4.
My proposal comes in 3 parts: 1), why incandescent flashlights are better at providing a hot-spot of light, which in turn gives the light a longer 'throw'; 2) why that design won't work for LED's; 3) what will work instead.
1) How come incandescents have such a beautiful hot-spot of light? How come even a cheapy $20. Auto-Mart spotlight can throw a beam of light farther than a $150. LED miracle? Simple: incandescents use parabolic reflectors with the filament at the focus of the parabola.
The magic is definitely not in the filament. An incandescent filament is basically a point-source of light, radiating 360 degrees in a sphere. The filament in the bare GE 60-watt bulb hanging from a work light provides no hot spot at all, and the filament in my spot-light is no different from it as far as that goes.
All of the trick comes from the reflector. A parabolic surface (paraboloid, if you prefer) has the cool property of being able to collect light from a point-source located at one of its focuses, and direct all of it outward in a bundle of parallel rays. If you want the most extreme, unmodified spot, that's the design you use; prior to the invention of lasers (which produce parallel bundles of light in a whole different way), that was the best way to get a tight, intense beam of light from one place to a far distant place. Of course, for flashlight purposes, we usually want some combination of hot-spot and corona, an intense focus dead ahead, with a decent amount of ambient light for our peripheral vision. And designers get this combination by stippling the parabolic reflector, and by shifting the filament a little shy of the focal point (like your old "spot-to-flood" aluminum behemoth where the whole bulb assembly pistoned back and forth). Still, what gives the basic character to incandescent flashlights--and what still gives a dime-store AA incandescent the long-reach advantage over an L4--is the fact that the filament is roughly at the focus of a reflector that is roughly parabolic.
2) So why don't we just put a Luxeon LED at the focus of a parabolic mirror and get all the benefits outlined above? Because of one minor problem, and one major problem. Minor first. An LED circuit like the Luxeon 5-watt is not a point-source. It is mounted on a chip, and only radiates in the hemisphere above the chip. So if we mounted the chip facing outwards at the focus of a paraboloid reflector, very little of the light would even interact with the reflector on its way out the door. It would just stream out the aperture of the paraboloid in the relatively diffuse way we're all familiar with. (Indeed, I think that is roughly what is happening with the KL4 head, assuming its reflector is roughly parabolic; I doubt if the reflector is doing that much work at all, and it certainly is not producing a hot-spot. And moving the chip forward, from the vertex of the parabola to the focus, would only make the problem worse). Still, that's only a minor problem: to circumvent it, we could simply mount the chip facing **backwards**, i.e. into the reflector, and then the reflector could capture it and focus it into a tight beam. With a little careful focusing and stippling, we could even remove the shadow of the chip from the center of the beam (you may have noticed that a lot of incandescent lamps have a frosted or even opaque apex to them, but this does not leave a hole in the beam if you make allowances for it).
But that leaves the major problem unsolved. As Peter Gransee has told us repeatedly, these new high-powered LEDs produce considerable heat, and are extremely sensitive to heat build-up. They need careful heat-sinking to keep them running. And while it might be possible to mount an LED at the focus of the parabola without the bulk of the chip obscuring the beam, we would certainly not be able to mount the chip **and** a simple and efficient heat-sink at the focus. Here's an area where incandescent filaments do have an advantage; they don't need a heavy heat-sink, so they can stand up into the focus on their own spindly leads without a lot of bulky undercarriage. Once we have provided a 5-watt LED with proper heat-sinking (and wiring), the package is too bulky to put at the focus of a parabolic reflector--at least one you could contemplate putting in your pocket.
3) What can we do instead? My suggestion is that we borrow a trick from telescope design. Keep the LED at the vertex of the parabola, as it is in the KL4 head, and keep it facing forward. Now, mount a small hemispherical mirror at the focus of the parabolic reflector, facing backwards towards the LED. (This is roughly like a Schmidt-Cassegraine telescope design). The small hemispherical secondary reflector collects light from the LED, but it makes the primary, parabolic reflector think that it has a tiny point-source of light located at its focus. This puts us in the running with the basic incandescent design; it allows the parabolic reflector to create a hot-spot of light, and then the normal modifications and stippling can both provide for an adequate corona, and provide for 'erasing' the shadow of the secondary mirror. The LED can remain where it is, in the body of the flash-light, with simple wiring and adequate provisions for heat-sinking.
The secondary mirror could be mounted in the parabolic primary with a set of spider-legs. Or, it could mount on a post projecting backwards from the front lens, like a mushroom on a stalk. Depending on the proportions and focal lengths, there might not be a stalk at all; for a turbo-head version with a 2.5 inch lens, it might simply be a matter of gluing a dime-sized mirror directly to the back of the lens.
Well, that's the proposal, in outline. Practically all engineering is in the details, and I haven't given you any. I don't have the time or resources to try this idea out and see how quickly it hits a dead end. But not all engineering is in the details; there is also having a vision to work towards and a leading idea to follow. I think this basic idea is sound--and I have some confidence in that claim, because it is not my idea to begin with. It's just a standard telescope design run in reverse, with the LED located at the eye-piece. Maybe the idea can be scaled down and put in your pocket, maybe it can't; I'm hoping that one of you people with a good machine-shop and an R&D budget--or just time on your hands--can give it a try and tell us all how it works out. (And I wouldn't say 'no' to a beta-version….)
In closing, let me take this opportunity to thank all of you out there for your willingness to share ideas, take risks, spend money, and sometimes lose money. Although I've never met him, I especially want to thank Peter Gransee, who is not only a fine engineer and entrepreneur, but also a really admirable communicator and educator. No one can read his posts without learning things and wanting to teach others. And no one can watch how he runs his company without feeling a renewed faith in basic principles of fairness, decency, boldness, innovation, and profit motive that are sometimes summed up under the heading of 'the American way'. That's another source of light in these dark times. Best wishes, curiousCrandall
This is an open letter to Peter Gransee, Paul Kim, and all of the terrific experimenters and innovators out there who are making the world of hand-held illumination such a strange, exciting, and fast-moving one these days. I want to propose and advocate a general strategy for improving the quality of the beams coming from the new-generation LED lights, like Arc's pathbreaking LS, and SF's KL4.
My proposal comes in 3 parts: 1), why incandescent flashlights are better at providing a hot-spot of light, which in turn gives the light a longer 'throw'; 2) why that design won't work for LED's; 3) what will work instead.
1) How come incandescents have such a beautiful hot-spot of light? How come even a cheapy $20. Auto-Mart spotlight can throw a beam of light farther than a $150. LED miracle? Simple: incandescents use parabolic reflectors with the filament at the focus of the parabola.
The magic is definitely not in the filament. An incandescent filament is basically a point-source of light, radiating 360 degrees in a sphere. The filament in the bare GE 60-watt bulb hanging from a work light provides no hot spot at all, and the filament in my spot-light is no different from it as far as that goes.
All of the trick comes from the reflector. A parabolic surface (paraboloid, if you prefer) has the cool property of being able to collect light from a point-source located at one of its focuses, and direct all of it outward in a bundle of parallel rays. If you want the most extreme, unmodified spot, that's the design you use; prior to the invention of lasers (which produce parallel bundles of light in a whole different way), that was the best way to get a tight, intense beam of light from one place to a far distant place. Of course, for flashlight purposes, we usually want some combination of hot-spot and corona, an intense focus dead ahead, with a decent amount of ambient light for our peripheral vision. And designers get this combination by stippling the parabolic reflector, and by shifting the filament a little shy of the focal point (like your old "spot-to-flood" aluminum behemoth where the whole bulb assembly pistoned back and forth). Still, what gives the basic character to incandescent flashlights--and what still gives a dime-store AA incandescent the long-reach advantage over an L4--is the fact that the filament is roughly at the focus of a reflector that is roughly parabolic.
2) So why don't we just put a Luxeon LED at the focus of a parabolic mirror and get all the benefits outlined above? Because of one minor problem, and one major problem. Minor first. An LED circuit like the Luxeon 5-watt is not a point-source. It is mounted on a chip, and only radiates in the hemisphere above the chip. So if we mounted the chip facing outwards at the focus of a paraboloid reflector, very little of the light would even interact with the reflector on its way out the door. It would just stream out the aperture of the paraboloid in the relatively diffuse way we're all familiar with. (Indeed, I think that is roughly what is happening with the KL4 head, assuming its reflector is roughly parabolic; I doubt if the reflector is doing that much work at all, and it certainly is not producing a hot-spot. And moving the chip forward, from the vertex of the parabola to the focus, would only make the problem worse). Still, that's only a minor problem: to circumvent it, we could simply mount the chip facing **backwards**, i.e. into the reflector, and then the reflector could capture it and focus it into a tight beam. With a little careful focusing and stippling, we could even remove the shadow of the chip from the center of the beam (you may have noticed that a lot of incandescent lamps have a frosted or even opaque apex to them, but this does not leave a hole in the beam if you make allowances for it).
But that leaves the major problem unsolved. As Peter Gransee has told us repeatedly, these new high-powered LEDs produce considerable heat, and are extremely sensitive to heat build-up. They need careful heat-sinking to keep them running. And while it might be possible to mount an LED at the focus of the parabola without the bulk of the chip obscuring the beam, we would certainly not be able to mount the chip **and** a simple and efficient heat-sink at the focus. Here's an area where incandescent filaments do have an advantage; they don't need a heavy heat-sink, so they can stand up into the focus on their own spindly leads without a lot of bulky undercarriage. Once we have provided a 5-watt LED with proper heat-sinking (and wiring), the package is too bulky to put at the focus of a parabolic reflector--at least one you could contemplate putting in your pocket.
3) What can we do instead? My suggestion is that we borrow a trick from telescope design. Keep the LED at the vertex of the parabola, as it is in the KL4 head, and keep it facing forward. Now, mount a small hemispherical mirror at the focus of the parabolic reflector, facing backwards towards the LED. (This is roughly like a Schmidt-Cassegraine telescope design). The small hemispherical secondary reflector collects light from the LED, but it makes the primary, parabolic reflector think that it has a tiny point-source of light located at its focus. This puts us in the running with the basic incandescent design; it allows the parabolic reflector to create a hot-spot of light, and then the normal modifications and stippling can both provide for an adequate corona, and provide for 'erasing' the shadow of the secondary mirror. The LED can remain where it is, in the body of the flash-light, with simple wiring and adequate provisions for heat-sinking.
The secondary mirror could be mounted in the parabolic primary with a set of spider-legs. Or, it could mount on a post projecting backwards from the front lens, like a mushroom on a stalk. Depending on the proportions and focal lengths, there might not be a stalk at all; for a turbo-head version with a 2.5 inch lens, it might simply be a matter of gluing a dime-sized mirror directly to the back of the lens.
Well, that's the proposal, in outline. Practically all engineering is in the details, and I haven't given you any. I don't have the time or resources to try this idea out and see how quickly it hits a dead end. But not all engineering is in the details; there is also having a vision to work towards and a leading idea to follow. I think this basic idea is sound--and I have some confidence in that claim, because it is not my idea to begin with. It's just a standard telescope design run in reverse, with the LED located at the eye-piece. Maybe the idea can be scaled down and put in your pocket, maybe it can't; I'm hoping that one of you people with a good machine-shop and an R&D budget--or just time on your hands--can give it a try and tell us all how it works out. (And I wouldn't say 'no' to a beta-version….)
In closing, let me take this opportunity to thank all of you out there for your willingness to share ideas, take risks, spend money, and sometimes lose money. Although I've never met him, I especially want to thank Peter Gransee, who is not only a fine engineer and entrepreneur, but also a really admirable communicator and educator. No one can read his posts without learning things and wanting to teach others. And no one can watch how he runs his company without feeling a renewed faith in basic principles of fairness, decency, boldness, innovation, and profit motive that are sometimes summed up under the heading of 'the American way'. That's another source of light in these dark times. Best wishes, curiousCrandall
