Does it have a moonlight mode and does it come with a pocket clip? 
Thank you for the tips djozz! I will keep that in mind.
I am going to build something like this with an CXB3590 @140 watts
though I need a collimator lens right in front of the COB to narrow the spread of it, I got a 90mm diameter and 90mm focal length lens so far but that is not enough, I will need a near (half)sphere to get a narrow enough beam of the COB for my fresnel.
I see, so you are going for maximum output in your build. I have asked myself that, more light is more useful but smaller light sources (less light) give a more intense hotspot. Iām still debating if I should rebuild the light with a Cree XHP70.2 for more light and still a pretty intense hotspot.
the collimator should compensate for the die size, it would make it look like a 3x3 mm LED from the focal point of the fresnel, which is pretty small, given lens loses and spill there should still be >60% of the LEDs light remaining.
I was also thinking about using a fresnel mirror to narrow the COBā¦
glad to see this thread revived, I missed it on the first lap. Nice build! I really like the speaker box megathrower. I have a smaller fresnel that I have been playing with and have been tempted to grab a free projection tv off craigslist or kijiji to salvage the ~60ā fresnel out of to have some fun.
You want to use the spherical lens as a pre-collimator (mounted right above the LED)?
Then itās the other way around. The apparent size of the LED as seen from the main lens is enlarged (probably doubled).
Thatās not how it works.
The apparent size of the spot will depends on the diameter of the led and itās distance from the FIRST optic.
If you have the LED 90mm away from the first lens, it will be a much higher divergence beam than if it was 1m away from the large fresnel lens (or however long that focal length is).
What the precollimator does is increase the luminous efficiency by collecting a higher percentage of total flux (lumens), while simultaneously creating a larger spot.
I may have said this before in some related thread. Light spreads outwards in a wide angle range from the led by design. I believe you should build a triple lens setup for maximum throw:
- A powerful Ć¼ber short focal distance precollimator, also with enough size to lose as little light as possible. Output: a wide straight cone of light.
- Another collimator: the concentrator, another plano-convex style of lens rotated 180Ā° with respect to the precollimator. Output: the cone of light re-concentrates to a point of light (or sort of) at focal distance. Diameter of this lens should match precollimator.
- Diverginator: the final focusing diverging lens, concave sort, which at the correct optimal distance from the concentrator would create a very tight cylinder/cone of light.
Mmmkay, now I'm gonna shut up and go see if some sort of trampy lass asks me for love services btw. 
Minor influences of lens systems ignored, throw depends on the diameter of the largest lens, in this case the fresnel. Whatever number type or position of lenses you throw at the system may affect many parameters but hardly the throw.
An LED is not an infinitely small point of light.
The closer you put an optic to it, the more divergence there will be since the LED has a finite size.
Putting stuff close to the LED is good to collect lumens, not to make highly collimated beam.
Once you have put a lens close to the LED, it doesnāt matter how large or far away the large lens is because the divergence has already been massively increased due to the close lens, and you cannot ācorrectā divergent rays of light.
djozz I am never disappointed reading about your sometimes odd, but always awesome projects. I like how you go into a decent amount of detail all the while not being so technical that a relative FL newbie canāt follow your narrative. I have spent many quality hours perusing your build threads and to a lesser degree reviews. You and a handful of other people on the blf have expanded my flashlight/batteries/electronic testing instruments knowledge base exponentially. I would love to have 20 percent of your FL knowledge. Thanks and may I say keep up the good work.
-Joey
Well, I ought to admit I am intuitively focusing the problem. However, the question here is: are you absolutely sure that wouldn't work? I believe it would.
Related: Afocal photography @ Wikipedia.
Precollimator: L3.
Concentrator: L2.
Laserator: L1 (well, I've renamed this one; of course, as I previously said a diverging lens could be used much closer to the concentrator). This one would be a small diameter lens, doesn't it?
System for the ultimate full spectrum laser Farsight Flashtorch. 
Take care.
Take a good look at your image again and imagine what it will look like when the source is NOT an infinitely small point.
Every time a ray hits a piece of glass it will diverge more and more because an LED has a finite size.
I think this is obvious enough for you to understand.
Red is an extreme scenario, green is slightly more realistic.
The convergence point before the last lens will be different depending on which point on the LED surface you trace the rays from.
This means it is impossible to focus the last lens for the LED unless the LED is an infinitely small point.
EDIT- there is a slight mistake with the long green line at the upper left, rather than bending more than the gray area, it should be bending less.
The focal point should be slightly below that black point.
That was a bit extreme, Enderman. That LED of yours is nearly as big as the precollimator, while in my mind I am imagining a precollimator of very large diameter and short focal lenght with a relatively tiny led behind.
Of course, this may still be bullsheesh but hey, maybe some workaround could be achieved. 
The closer the LED is to the precollimator, the larger the angle of divergence.
This is why you want very long focal lengths for anything dealing with collimation.
The longer the distance is, the closer the LED gets to an infinitely small point.
Osram black into that one please 
Then again not sure with that lens design, it sure is interesting.
If youāre ever close to Brussels take that thing with you and come visit me
Iām not sure if with such large lens faults a 1x1mm die is a good idea. There will be a under-limit in hotspot size because of the crappy lens, and it may not be so bad to have a somewhat large light source to begin with.
This box is a bit large to bring along but I have vague plans to make a 20mm diameter version that should do 3.6 Mcd, which will be way more portable.