Thank you Argo for this post, I found some cool new designs I have never seen before!
When using a pre-collimators with a lens,
I have always wondered at what angle does the light just want to “bounce off” instead of go into the lens when the angle reaches high enough.
How much does AR coating help with this?
Is there a general guideline or table that can show the transmittance relative to the angle?
Reflections of light hitting lens at angle can be calculated from Fresnel equations
Yes, AR coating helps. But each AR coating is designed for some angle of incidence and works best there. If you read that UCL is 99% transmissive - that’s just with light hitting it straight ahead, at other angles of incidence the reflections will be stronger.
But at other angles AR coating helps too…and is actually quite critical when the angle is high.
I believe that aside from professional lenses pretty much all AR coatings are designed for light that hits them perpendicularly (which is not optimal).
Simulator - Ray Optics Simulation a free program to play with lens configurations, a unvaluable help to design a system with less invest in different samples. i am designing a searchlight with precollimator system and this program helpme a lot. The main features of optics was on my mind, but whit this program you can emulate fast the possible efects of the light without a continous ( and expensive) trial and error.
Any way i need to say later of saw several designs of precolimator proyects here, the first stage would be ever a half ball lens, not plano convex lens having a part only of the ratio curve. the better system to collimate is a complete sphere, but there exist the problem o poor light collection.
I feell a litlle half ball lens like collector and a medium size plano convex with AR coated should work like the combination.
My prototype have a medium size plano convex+ 1 fresnel of 100mm dia. and i am geting a nice 1.7 millions of candelas with only 10 watts in the new osram leds. And the spot size is very adecuate for a searchlight use. Now i am waiting the arrive of the half ball lens samples, to see if i can compact the design to have a size head more litlle. My concern is not loose more light in the edges of the primary collector, and a better beam angle to the secondary to not loose any light in the head walls.
A half ball lens should be the answer.
Interesting, thanks for posting. 
Yes, spherical lenses ("half ball") are suitable for use as pre-collimators.
I know this is a super old thread but the Nebo O2 Beam was of this design as well and is the first I ever saw which sparked my continued interest in designing a light this way, which is how I came across this thread.