Light lost in aspheric lens flashlight

Hello everyone:
I have a UF-1504 host on the way from mtnelectronics, so I’ve been thinking about how to optimize the flashlight. My goal with this flashlight is of course to have a lot of throw (cd), but I’m concerned about the low light extraction efficiency due to the light not collected by the lens when it’s in focused “spot mode”.

So I am planning on using a small collector lens very close to the LED to collect a good portion of the light that would just be lost. My understanding is that while this can significantly increase the light collection efficiency, it will not result in significant changes to the beam cd. In ray optics language, this is because when the collector lens is placed very close to the LED (within the focal length), this produces a virtual image that is larger than the actual LED. The large lens then sees this virtual image and focuses it. The result is the beam with the collecting lens in place is larger, but close to the same cd. This works for me because it makes the beam wider and more usable. There is some talk about collector (pre collimator) lens in this thread: Uniquefire UF-1405 - A worthy zoomy?

The reason I made this thread was to more closely analyze the light collection efficiency. Aspheric lenses used in flashlights usually have a focal length approximately equal to the lens diameter. This means the cone of light collected from the LED has a 26.6 degree half angle. Now how can we estimate what portion of the total light from the LED lies in this cone? We can do this using the “typical spatial distribution” plot in the XP-L data sheet. This shows the luminous intensity (cd) as a function of angle. It is greatest at 0 degrees (right above the LED), and goes to zero at 90 degrees (down to the side, parallel with the LED die surface). From this plot, one can see that there is a lot of light at half angles greater than 26.6 degrees, so the lens is losing a lot of light, but this plot doesn’t tell the whole story. To get the total light as a function of angle, one must integrate the light cone (integrate around the axis perpendicular to the LED surface). Doing this effectively weights the higher angles more because there is more light in the cone at higher angles. This adds a factor sin(theta), and this factor multiplied times the spatial distribution plot gives the total light as a function of angle.

If we approximate the luminous intensity in the spatial distribution plot as linearly decreasing with angle, the graph in this link:

shows the total light as a function of half angle (in radians, so 1.57rad is 90 degrees). The area under the curve is the total light in different sized cones. The cone from 0 degrees to 26.6 degrees (0.464rad) has roughly only 25% of the total light from the LED. So a typical aspheric lens setup loses 75% of the light. Now if we use a collector lens close to the LED that collects a much larger cone, say with half angle 75 degrees (1.31rad), 94% of the light is collected.

Conversely, one can calculate the light collected by a typical reflector, which might collect all the light with half angle greater than 30 degrees (0.52rad). The light collected by this reflector would be 72, so 72 of the light would be directed to the hot spot, while the rest would come out as spill.

I just realized that the extra sin(theta) factor should be included which is why I made this thread. I think the analysis is correct. What do you guys think? Like I said, there is some talk of people successfully using collection lenses in the thread I linked above, but I could not find anyone estimating how much more light was collected.
I’m planning on doing some measurements before and after I modify my UF 1504 with a collector lens, so we’ll see how much more light it collects.

Thanks for the nice analysis :slight_smile:

What is not taken in account yet, is light loss from reflections. To get close to the same cd as without it, the pre-collimator lens should have a good anti-reflection coating.

Looking forward to your measurements!

(I did some measurements on the light loss in a zoomie a while ago, link)

Thanks for pointing out your measurements; I had come across them before but forgot about them. I think this really emphasizes the need for pre collimator lenses in these flashlights.
I have a small lens with AR coating in mind. Probably the 9mm diameter lens with 9mm FL from here:
http://www.anchoroptics.com/catalog/product.cfm?id=368
Not super cheap at $19, but I think these lenses are quality.

Go to a flee-market and buy an old super8-camera, video camera or other obsolete optical stuff for a dollar. If you open up the multi-lens objectives it is full of high quality coated lenses, there’s good chance that there’s one with the right diameter and focal length.

interestingg!
subbed

The analysis seems correct, except that I would expect the intensity to be closer to the cosine of the angle than linear.
One can increase the spot size with a shorter focal length lens. Up to a point that is shorter than single element lenses usually get, there is not much loss in throw, because you are still collecting light from angles where the LED puts a lot of light. Shorter than that usually requires some sort of Fresnel lens. Using two lenses in series will have a similar effect to using a stronger lens, so that should work, at least to some extent, if you do it right.
First order optics is simple enough to do by hand or you can write a simple program.
If you program and and know some mathematics, it isn’t too bad to write a ray tracing program to get a handle on how much light will get through. Doctor Jones has done that for reflector lights. Or you can just try it various ways.
This is a good all refracting flat Fresnel lens http://www.dhgate.com/product/flat-fresnel-lens-for-stage-light-with-size/153508269.html?utm_source=dyn&utm_medium=GMC&utm_campaign=victor_jiake&utm_term=153508269&f=bm. It has size 130 mm. and focus 50 mm. It is so strong that the light at the edges looks distinctly yellow.
POP lites, LED Lensers and Coasts use two zone Fresnel lenses.

You are right that the intensity vs angle is close to cos(theta), but wolfram alpha would not calculate the answer in that case because it took too much computation time, so I went with the quick approximate answer.

That is a strong Fresnel lens, maybe the strongest lens in general that I have seen for sale, but it would still only collect about 67% of the light from the LED, though it would make a nice light weight spotlight. I hope that a well designed two lens system can get higher light collection efficiency. I have been doing some calculations and it seems the pre collimator lens should be as strong as possible for this purpose. There is a 20mm diameter, 11mm FL lens at edmund optics.

By the way, check out this two lens thrower I put together a while back. It worked out pretty well, but I didn’t really take any measurements regarding the light extraction efficiency.

That was 2 years before I joined CPF and later BLF :slight_smile:

That was a really cool light, optically well thought out. I wonder what the throw number was (is?).

Thanks. I still have it but it is disassembled. It never got past prototype state, so it was fragile and awkward to use. As it was in the thread I would guess it had 300-350kcd. It could be much higher now I think, if I installed a brighter, dedomed LED. One of my favorite things about flashlights is they can be so portable, so this UF 1504 that I’m building, to me, is sort of a new and improved version of that light I built 6 years ago.

I thought it was interesting how Ervin used a small aspheric situated at its focal point within a large reflector to both capture and focus virtually all emitted light. It seemed pretty effective.

That is a neat idea and impressive execution.

The inefficiency of software never ceases to amaze me. That should take something like a microsecond per angle on a laptop.

I like this type of threads,

Did you seen this thread: Dereelight DBS with de-domed XP-G2 and pre-collimator lens | Candle Power Flashlight Forum

That certainly looks like a good result. It seems like he is getting good light extraction efficiency from his ceiling bounce test comparison to the 860 lumen light.

There is no direct before/after comparison for adding the pre collimator lens, but we can tell something from the beam shots. I assume the XP-G was dedomed. The XP-G with pre collimator has about the same area in the beamshot as the dedomed XM-L which has about double the die area as the XP-G. So the light collection efficiency approximately doubled by adding the precollimator lens.

I can’t help plugging my own fresnel lens contraption here that I made last year O:-) 37.5cm fresnel lens spotlight build

Vinz is really advanced modder. I wonder what happened with him? He was member here on BLF?

That was 3 years ago. Pre collimator lenses(really working, confirmed and tested by various people) and tint changing of de domed emitters are his biggest inventions… Everyone talked about it at that time and just few people managed to crack right pre coolimator lenses and to change tint with “diamond treating” as he said :).

We can tell a lot from that picture… It is 100% bigger die projection without lux drop. That is huge upgrade to led because dedomed XP-G2 grows to XML dedomed size(xml grows even bigger :slight_smile: )

I tried one pre collimator lenses and I got MTG2 size projection of XP-G2 but about 50% lux drop with poor flood mode. So in aspheric flashlight if you achieve bigger die projection with pre collimator lenses you will have extreme hotspot size with poor flood mode.

I think a lot can be achieved with right lens choice and placement which is not easy at all. It needs to be placed around 0.7mm above led as Vinz said.

So due to my lacks of knowledge in such deep details, I’ve nothing to add to the technical debate about pre collimator lenses,
but since a couple of times ago Vinz had pimped up some of my Dereelight pills for the Night Masters and X-Searcher,
I can show you a practicle kind of example.

!!
!!

The effect of the pre collimater:

!!


!!
Allthoug the X-Searcher and the UF. 1504 have nearly the same headsize (maybe the X-Searcher is a little bit smaller), I think it’s good to see that the spot of the X-Searcher is quite significant larger.

As I remember it well, Vinz said that by using such a pre lens you might have a very little loss of lux (around ca. 5 %),
but this price is worth to pay for, concerning to the amount of lumens you are gaining, which in relation should make the pre collimator
even more effective with a brighter LED.

Some Beamshots of the Night Master and the X-Searcher (and other companions) with a pre collimator XP-G2 ded. / XM-L2 ded.
you can also find here - A little beamshot comparison of my decent Thrower material.

We needed such reply in this thread :slight_smile:

About this “Vinz said that by using such a pre lens you ’ve a very little loss of lux (around ca. 5 %)”

He said once that even small lux gain can be achieved sometimes. But mostly it is 0 or ± 5% which depends on how accurate you placed lenses.

Thanks for these pics, MAC. I think that is close to the before/after comparison I was looking for.

I’m aware of the principles and compromises of using pre collimator lenses. I made a larger version of the two lens system back in 2010 (CPF thread linked above).

The objective is to collect most of the light with the pre collimator lens and have the lens project the light in a smaller cone towards the main lens. The picture below is of the light that I made, with the lenses in the optimal positions, with the cone just filling the main lens.

If the pre collimator lens were closer to the LED, the light cone would become larger and it would overfill the main lens, losing some light around the edges. If it were farther away, the cone from the pre collimator to the main lens would be more narrow, but then the pre collimator lens would not collect as much light from the LED.

The pre collimator lens should have as large a diameter:focal length ratio as possible to optimize light collection (a ratio of around 2 is what I used in the above picture). If you are familiar with ray optics or play with a calculator like this one:
http://www.photonics.byu.edu/Thin_Lens_Calc.phtml
you can see that when the LED is very close to the lens the image is close to unchanged in position and magnification. When the LED is close to the focal length, but inside the focal length, the image appears to be farther away from the lens and there is significant magnification. The light cone from the pre collimator lens is very narrow in this configuration. The magnification is what causes the beam to enlarge; the image that the main lens sees is larger than the actual LED. Some position in between these two extremes will be the optimal position for the two lens setup. The reason you want as large a diameter:focal length ratio as possible is so the pre collimator lens is as close as possible to the LED when the light cone just fills the main lens.

In the picture above, the light coming from the pre collimator lens appears to be coming from an image of the LED that is slightly below the actual LED and slightly larger.

there are two ways to use two lenses…

1) you arrange them in a typical Galilean Telescope fashion.

or…

2) you are actually making a “compound lens”. the rear lens closest to the light source is typically the shorter FL. The bigger FL lens is out front. FL of the NEW resultant compound lens? is determined by FL-1 and FL-2, and the airgap between them.

some general rules…

a) two lenses of equal FL, touching (airgap = 0mm) will make a single compound lens of exactly half. Example, two 100mm FL lenses touching? make a new compound lens of 50mm FL.

b) using example a… as you introduce airgap bigger than 0mm (not touching), the FL of the compound lens gets closer to 100mm FL, but never quite reaches it.

c) as the airgap gets bigger? the rear lens gets closer to the light source

d) the SIZE of the well focused emitter? is determined solely by the FL of the resulting compound lens. Smaller overall FL, bigger die projection size. larger overall FL, smaller die projection.

e) cigarette smoke is your friend for seeing you are not wasting light from lens to lens, lol.

f) theres no reason to limit yourself to 2 lenses, grasshopper… i have used 3 and 4 lenses before, with IR LED’s… to good effect.

g) since 2 lenses actually form ONE compound lens? and another 2 lenses form a second compound lens? using those 4 lenses, is actually like using 2 lenses, for the purposes of the math governing it.

h) if the two lenses are different FL? radically different? like… 25mmFL and 600mm FL ? the resulting lens will be a lot closer to 25 than to 600… but theres equations you can work out to accurately predict it based on both FL’s, and the airgap.

everytime i brought this idea up HERE, and on my other site? everyone poo-pooh’s the idea, and makes up ideas why it doesnt work and how it cant be efficient… even though the MATH shows it works. (also? Dr Jones described it… also? a “white paper” from edmunds describes it…)

I only use it on IR emitters… i’m glad someone started doing it with visible light…