That would be a telescope with a light source at its focus. Probably a space telescope such as the Hubble Telescope or a military surveillance telescope would throw farther than an Earth based one. I heard once that they bounced a laser off the moon.
(edit) Just to verify:
I think the thing is that an LED is not a point source.
If it would be a point source, diameter would theoretically not matter, only the ratio between width and depth.
To get close to the effect of a point source, the emitter should be as small as possible, in relation to the starting width (at 45° angle, which gets to reflect the warm coloured ‘side light’)
In other words, get a reflector that starts pretty wide, with a large ‘back hole’ (in relation to the LED)
Also decapitating the LED will help, because the dome magnifies the chip.
Still, when the parabolic (as close to it as you can get) reflector is also deep, in relation to the width, it should increase throw, because it throws more light in one direction, a higher percentage of the light emitted by the LED than when you cut off a portion of the mouth / flare / rim of the same reflector.
But that’s duh-obvious…
How would a reflector-flashlight do when you (somehow) also put a lens in it to bundle only the spill?
(I have some ideas on how to do that, but it’s not easy…)
Anyone try that yet?
No more black hole, like a TIR.
So where are the large TIRs?
I have so many questions…
Yes, a deeper reflector does catch more light (i.e. has smaller spill and more light in the collimated beam from the reflector), but no, it doesn't increase throw, instead it produces a wider corona, because you get smaller focal lengths. (Well, actually a deeper reflector does increase the throw a little, but only due to the smaller dead hole).
Also dedoming a LED does not increase throw because the dome enlarges the LED, but because of total inner reflexion and "photon recycling".
Large TIRs: these would be heavy pieces of massive material. Also the surface precision has to be better with larger optics.
My suggestion for even more throw (with a given diameter): Take several blue laser pointers and point them to the die to increase it's luminance.
But in case anyone has forgotten the point here, the question of this thread is REFLECTOR Width vs. Depth for Throw. Aspherics need not apply. Nor should telescopes or lasers or off Earth devices powered by NASA. The question is about flashlights with reflectors. The question is what combination of width/depth gives a reflector the best throw. Period.
Michaels DeftX looks sweet, it’s an aspheric, and I’d have to see it throw like he claims to believe it, as it’s an XP-G emitter. There are rules about the transmission of light. 4 times the light to travel twice the distance. The emitter has to be able to put forth sheer power. And we have all seen the XP-G footprint fall short. I can’t say it isn’t so, but I can so I don’t believe it without seeing it.
And still, the furthest throw document here is from that TK61vn. Anybody got one doing better than 622Kcd? Anyone? Measure the reflector on your best throwing light and see if it’s close to square…1:1 width vs depth. I’d like to know, really, I would. I’ve asked rdrfronty to take a look at that TK61vn. Give me some measurements.
And for the record. The driver, heat path, a whole bucket load of variables have little to do with how the reflector works to focus light from the emitter. A very nice reflector does indeed throw even Lo mode light surprisingly far. That being said, gotta have light to throw light. Load up a paint brush, make a stroke with it. How far did it go? Get a longer narrower brush, yada yada yada…you’re gonna be running out of paint in the brush regardless. Similarly, light can only travel so far without a whole lot more light pushing it. Or so it seems to me. But then, I’ve got a pretty good track record these days at being wrong….
I would like to verify again:
…because the parbole is relatively smaller with a deep reflector of the same width, so relatively the emitter is larger, not at all like a point source which would be perfect (with the perfect reflector) but more like a “cob LED” :^\quote].(Well, actually a deeper reflector does increase the throw a little, but only due to the smaller dead hole).Also dedoming a LED does not increase throw because the dome enlarges the LED, but because of total inner reflexion and “photon recycling”
[/quote]
My point would be that a domeless emitter is more like a point source than one with the dome, because with the dome, the light comes from a zone rather than a spot, scattering the light.
I’m not sure if PMMA is so heavy for making a, say, 40 or 50 mm diameter TIR.
That’s not on my to do list yet…
But yes, I have to do more stuff in practice and then talk about it… 
In the mean time I have a ‘lengtened C8’ on its way with an extremely deep reflector (60mm in stead of 30mm) and I already regret it… :Sp 
Dale - You are the one who asked "What flashlight has the longest known throw?" ?? Okkk, I'll forgot you asked 
, but I certainly did not forget the point here...
I have a TK61, belongs to DayLighter, but I resistor mod'ed it for him and still have it while working on another light for him.. I had it all apart - it had no glue, was easy to get access to the emitter, driver, etc. I can measure the reflector if you want. Like I said previously, it looks like a HD2010 reflector on a larger scale, roughly same proportion of width to depth. Also I posted on it in this thread on the TK61: https://budgetlightforum.com/t/-/23474.
Again, primary, main reason why the TK61vn is king of "reflector" throwers is because it's reflector's diameter is the largest, plus easily resistor modded to high amps. Of course other factors of design/shape - for example, though the NiWalker had a slightly wider diameter reflector than a TN31, it could not throw as well - it did good, but not as good as a TN31 - thinking the amps was also a factor - could not get the amps up as high as you can with a TN31.
Unfortunately many of us (me too! ) don't invest in the high end name brand lights to tinker around with. Vinh can of course because he makes a living on mods and has a large customer base that will pay big $$$.
If that's true,then dedoming color LEDs wouldn't increase throw.
That is the more fundamental reason, in that it applies to optimal optics, but the visible fact that it enlarges the led is also relevant. The dome narrows the angular spread, which reduces the throw of a reflector light.
I'd like to point to the dedoming thread where that is discussed, to avoid going off-topic too much.
I bet designers are not trying to give us perfect “Paraboloid” reflectors, If they were then by definition, the patterns of light would be perfectly even across the beam (assuming no Optical aberrations or imperfect dimensions). setting aside the known devation from perfect of all the dead space behind an LED emitter, something else is going on.
I think all the “personality” we see. Is from two sources (or maybe controlled sides of the same coin. We know that as explained by wiki below that a “perfect” Paraboloid designed for visible light is incredibly precise and we know our light source is not a perfect “point source”. So I bet either or both are happening.
- The designers are trying to give us a hot spot (because if we have shown a preference for focused light in the middle).
- The the “out of focus” portions of the LED end up emphasizing uneven distributions of light aka the hot spot/corona/spill.
- Artifacts are simply because of deviations from the ideal in application.
I was noticing on one of my lights that there are four distinct areas of light. Hotspot, Corona, Inner Spill, Outer Spill.
I think what we call Corona is actually the light from directly out the front, then where does the less then perfectly focused light show up? The rest of the variations I think might be from the design intent to give us hot spots and simple material deviations from ideal for the other artifacts.
The below is from good ole Wiki:
In contrast with spherical reflectors, which suffer from a spherical aberration that becomes stronger as the ratio of the beam diameter to the focal distance becomes larger, parabolic reflectors can be made to accommodate beams of any width. However, if the incoming beam makes a non-zero angle with the axis (or if the emitting point source is not placed in the focus), parabolic reflectors suffer from an aberration called coma. This is primarily of interest in telescopes because most other applications do not require sharp resolution off the axis of the parabola.
The precision to which a parabolic dish must be made in order to focus energy well depends on the wavelength of the energy. If the dish is wrong by a quarter of a wavelength, then the reflected energy will be wrong by a half wavelength, which means that it will interfere destructively with energy that has been reflected properly from another part of the dish. To prevent this, the dish must be made correctly to within about 1⁄20 of a wavelength. The wavelength range of visible light is between about 400 and 700 nanometres (nm), so in order to focus all visible light well, a reflector must be correct to within about 20 nm. For comparison, the diameter of a human hair is usually about 50,000 nm, so the required accuracy for a reflector to focus visible light is about 2500 times less than the diameter of a hair.
Optical aberration
To make a good hot spot, a parabolic mirror is best.
The non-zero size of the LED (i.e. the out-of-focus parts) makes the size of the hot spot.
Different parts of the reflector cast images of the die of different sizes, the bigger ones causing the corona.
Fortunately, for simply collimating the light of a LED, the required reflector precision is much much lower...
I think the outer spot is sometimes caused by flattening of the reflector at the back, relative to a parabola. This may be done in order to get an outer spot, or it may be done to shorten the reflector. I think I have a light that forms an outer spot this way, and the outer spot is warmer colored than the main spot or the spill, because the light hitting the back of the reflector has a longer path through the phosphor.
Yes, 20 nm. deviations from a parabola blur the spot by an amount that is comparable to the diffraction aperture limit, which is much smaller than the image of the LED.
I heard from people I was working with on an X-ray telescope that there are other possible choices of shape that handle an extended focal plane better than a parabola does, at the expense of loss of perfect point focus. Apparently, the improvement is minimal, even for a telescope, as they did not expect to pursue that line.
No, at least none of my reflectors have that. But you are right, the parts of the reflector near the LED catch the warmer light - and since they are nearer to the LED, they also cast a bigger image of the die, and these bigger images form the corona. This happens with a prefect parabola.
I can’t remember right now which it is. It isn’t the large reflector light I now have a battery in.
That is interesting about the nearer center back part of the mirror casting a larger image of the LED. I need to think that over.
I had an illustrative idea to show. In the image below notice how deep the reflector is? It is a very, very shallow reflector. These lights throw a very tight beam a very long distance (originally built to find enemy aircraft in the sky at large distances).
So thusly a wide and shallow reflector throws the tightest beam and makes the case for the best thrower design in general. I believe the light source's location above the reflector makes no significant difference in this particular example. The light source is at the optimum focal point for a tight and long throw.
As a side note it seems that in a flashlight experiment placing the LED suspended above the parabolic reflector might be very interesting and lead to an extreme thrower if the reflector is shallow and wide.
The parabolic reflector/mirror is 60" (152.4 cm) in diameter and the searchlight has a peak output of 800,000,000 candlepower or candelas driven by a 15kW generator the light drawing 150 amps at 78 volts DC and has an effective beam visibility of 28 to 35 miles (45 to 56 km) in low humidity.
Even though the light at the edge of an LED is much smaller in area, because of the mirror’s proximity, it’s converging angles are much wider. Angle in = angle out.
I am sorry to be off topic but what if the LED emitters were made in a semi-spherical shape instead of planar?
That would be very useful indeed, and if that shape would be globular you would have a led shining in all directions which is great for household bulbs. But such a led is not there and that must have something to do with a more difficult or impossible production process, and perhaps also with the more difficult heatsinking of such a led.
So, who wants to join me in an experiment that costs roughly $33.00? 
I have found a new optic that produces a 1º beam angle with an XP-G2. Usable with XM-L2, SBT70, XHP70, whatever. All producing the absolute lowest angle of any collimation device seen to date.
Designed for throw, the XP-G2 is estimated to have a 16M hot spot at 800M. SBT70 31M at 1200M! But, this one is gonna take some creativity. It’s 157mm in diameter, 79mm depth. It has screw mount capability through it’s holder, uses a dual lens system with the large lens being of the Fresnel variety.
It’s new, and has a 20 piece MOQ. :~
I want to try this so bad! But not $660 worth of want…
Nice picture. When I was a kid, a lot of car dealers and some others had these as signs.
Since the source must be nearly isotropic, or at least a hemisphere, it seems that it would have had the same throw and wider beam with a shorter focal length mirror. So I think the very long focal length must be for practical reasons such as size and weight, cost or to be able to use a sphere as an approximation to a parabola.
Extreme thrower flashlights either are retro-reflectors, as you say, or else have lenses.