Should reflectors be parabolic?

I’ve been thinking about Waiven collars recently and came to conclusion that they probably shouldn’t be spherical. I say probably, because the topic is actually quite complex and I haven’t studied it deeply enough.
The same logic applies to regular reflector, which I think interests more people.
Maybe what I write here is well known in some circles, but the parabolic size gets repeated so often that I believe it’s not common knowledge here.

I’ll make a simplified model. Too simplified to be good, but good for a start. Then I’ll refine it to the point when it’s still much too simple, but better.
The basic reason why reflectors shouldn’t be spherical is that LED is not a point source of light.
Let’s draw it as a 1-dimensional source first.
Below you can see a domeless 1d emitter as well as a pair of light rails, before and after reflection.

The angle after reflection is equal to the angle before reflection.
To minimize beam angle they should both be at the same angle from pointing forward.
What is the optimal solution? It is simply to divide the angle in half and adjust the reflector angle so that the beam from this angle is directed straight forward.
What does a parabola do? If divides the emitter in half and directs the beam from its middle forward.
Which is different, the optimal shape should be steeper than a parabola.

Now, that’s optimal if you want to minimize beam width. But usually the point is to maximize throw. That is - the amount of light pointed forward. In the drawing above, the closer end is at a better angle than the other. By which I mean - there’s more light coming this way. So to maximize the amount of light forward one should tilt the beam a bit….by making the reflector steeper.

What happens when we move to 2d? And then add a magnifier on top of it (dome)?
I haven’t fully wrapped my head around the first question and haven’t touched the other. I don’t see anything that would shift the optimal shape to become parabolic.

Also, it shows that the actual optimal reflector shape depends on:

  • LED die size
  • emission profile

Therefore it should be tuned if not to the target emitter to at least to some hypothetical average one.

Anyone seeing a hole in the logic above?

What you’re saying is all correct. To design the best possible thrower, you need to account for the individual emission pattern of the LED. This isn’t widely discussed, because it’s not relevant to everyday lights. Most beams diverge on purpose, so it doesn’t make a difference.

Let me add something to think for.

You aim at redesigning the reflector shape, something I too endorse. Now that we're there, however, how about doing it around the convergence profile of a nice convex lens? I'm not speaking of another zoomie flashlight, of course, but let me explain: imagine the beam out of a fully focused zooming flashlight. In this case, with a convex lens at the right focal lenght from the emitter we attain a tight and spill free beam, with the drawback of it collecting the intense but relatively narrow topmost emitter lightcone. What if we were to design a surrounding reflector with the convergent effect of the lens in mind? Such reflector would not need to be as steep as the typical ones because of the lens' own convergence.

I think of this as a more optimized and less space consuming way of attaining a tight, even beam light distribution, spill free thrower with maximum out the front throughput.

Just my ¢2 on this given my limited knowledge.

Cheers

A light source cannot be concentrated into a point smaller than that from which the light is emitted.
A spherical reflector over an LED projects a perfect image of the LED on top of itself.
It is not possible to have this light concentrated any more than it already is.
There is also no light being wasted, as you can see the reflection is the exact same size as the LED.


.
.
.
Your theory about the angle from the center of the LED hitting a parabola is correct, however since the LED is not a point source of light it also means that the incidence angle is far greater than the difference between splitting the angle perfectly in half and the way a parabola reflects from the center of the LED.

As LEDs get smaller and parabolas get bigger, the angle you drew in your image becomes fractions of degrees and there is practically 0 difference between having the center of the LED at the focal point or having the angle split perfectly in half. Basically, what I’m saying is it doesn’t matter.

For example, a parabola with ~10mm focal length and a 1mm LED, the angle at 22.5 degrees emission for one half of the total incidence angle is 0.3443 degrees, and the other half is 0.3440.
For angles closer to 0 degrees (directly to the side of the LED, this difference becomes smaller.
For angles closer to 60 degrees (towards the far edge of the reflector) the distance between the LED and reflector surface gets longer, also reducing the incidence angle.

What makes more of a difference is how large the parabola is compared to the LED.
If you have an XHP70 and a tiny 1” reflector then sure, if you put the LED at the focal point you probably will not have the tightest beam possible.

For real throwers that use small LEDs and large parabolas, the proportion of the angles is basically 1:1.
Unless you have some professional grade cold mirror parabola with a surface accuracy to a fraction of a light wave, you will never be able to measure a difference, let alone SEE the difference.

I easily believe it doesn’t matter for lights like BLF GT. But these are not everything. There are pocket rocket throwers too. I wonder if they use parabolic reflectors or tuned ones? The answer is “tuned”, right?
With collars it’s the same - a large collar coupled with Blackie may be a great choice for breaking records, but that’s not all there is.

They are all parabolas.
Whether the LED is exactly at the focal point, who knows.
Either way the amount of spread from the LED is hundreds of times larger than a little offset of the half angle of light, so it makes no visual difference.

Also, the quality of the reflector is not anywhere near accurate enough for a fraction of an angle to make a difference.
I have never seen a small handlheld flashlight use an electroformed reflector, or even less a cold mirror.

I thought up a ‘wavian dome’ a while back:

Led Lenser TIRs are basically this but instead of a dome (which has ~50% efficiency) there is an aspheric lens (closer to 95%)

Puts more light OTF. But because the lens is much smaller than the reflector, it can’t collimate the light as well. This leads to a 2-zone beam, where the reflected part is either narrower than the projected one or artificially widened for the sake of uniformity.
Am I right?
And will it out-throw a Wavien dome?

I guess you’re right. Still, there’s tiny untapped room for improvement. :slight_smile:

Most TIR optics have a lens in the centre.
The ‘problem’ is usually that the lens can not make a beam as tight as the reflector can, due to focal distances (from LED to optic)
(I see Argo said that too)

This is solved in this particular 20mm TIR though:

Makes for a nice little pocket thrower.
(I have some since recently, not yet tested)

No, they would perform about the same, when using a lens you get more front area, while using the dome you get higher led intensity.
Throw is a linear function of both area and intensity, so it would be almost identical, maybe slightly better for the lens since it has fewer losses.
But yes the lens TIR system has a larger angle of divergence, however not much more.
If you care about tight beam you should be using a real collar+lens, where the lens is much farther away from the LED than it is in a reflector (see jeromel’s image)

Now that’s a nice match for my S2+!
I’d love to see some beamshots.
I bought 32 lenses for this one flashlight already….I guess another 10 will come….

Wel, i just did a quick test, and it works, but it does project a square LED die, but the beam from the TIR reflector section is nicely superimposed on it.
I tried it on a dedomed SST40 by the way.
This combination will go in a Solarstorm SC01 with a Nanjg 105 with 8x 7135 on WindyFire high drain 14500
The SC03 has a large enough bezel opening to (almost) cover nothing of the front of this TIR.
An S2+ has a smaller opening, i think 17 or 18 mm.

It can be widened somewhat.
Anyway, measurements will tell how good or bad it is.

Just found a very similar lens from LedLink…looks like one is a clone of the other.
LL01CR-ADIxxL06
http://www.ekoled.com.br/files/Lentes.pdf