Yeah, it’s a nice idea, and it has been discusses several pages back.
We would need a ‘blue’ (often purple) laser
I don’t think it will have better efficiency than an LED though.
And frankly i don’t see any serious cooling issues when you put some flat heat pipes across the reflector with the LED soldered onto it.
It would be an XHP35 or an XP-L2 that i want to use, so maximum 20 Watts.
It would also be a lot simpler, which is a plus in my book.
…and a lot cheaper…
I imagine the phosphor would have a mirror behind it so that the light is all reflected towards the reflector.
But shouldn’t the phosphor be cooled too?
I think so, otherwise it will be burnt by the intense hot spot created by the laser. Anything opaque will absorbs most visible photon energy. But not as severe as if you place the LED in the recoil - this is my guessing so don’t take it too seriously. I often do some laser cutting jobs in my friend’s workshop, and cutting a clear PMMA required as much laser power as to cut thicker multiplex wood plate.
As for the heatpipe layout, in my opinion, since heatpipe is orientation sensitive, it’s better to use full bridge on top the parabolic mirror. This way you can get average heat shedding capabilities from each heatpipe bars. A cross mark design with smaller heatpipe is what I’m thinking.
- Clemence
Yeah, a bridge of heatpipes across the reflector is the eventual plan.
Perhaps even a cross or “Mercedes star”.
I’m not sure if the flat ones are also orientation sensitive though, i doubt it actually.
Any heatpipes with wick lining will be less sensitive to upside down orientations (hence they designed) but still perform somewhat better in normal orientation. The general rule is heat down, cool up. But the wicking technology should already improved by now.
I remember those older laptop manuals stated not to use the laptop other than in horizontal position, perhaps because of this heatpipe issue.
Ah… I see. you are keeping the F ratio constant. I always compare lenses of the same diameter with each other (precisely because the diameter determines the throw and size of the flashlight).
No, as far as I know, that has never been shown in practise. It also doesn not correspond to what the physics/optic experts (Ra, WalterK, sma, Dr. Jones etc.) say. Only get_lit writes some weird stuff which they, as far as I know, have not agreed to.
Some posts from Ra (professional optics guy): click, click (there used to be pics there, which showed the different lenses and the lux meter display), click
A detailed (German) thread from sma: click
A detailed from Dr-Jones: click
Depending on the exact diameters and using the same LED at the same current in both cases: yes. This is only the case though, when the diameters are rather similar since the Collar only doubles the luminance and the area of a circle increases with the square of its radius.
The assumption is not false. The calculation seems simple, but it makes sense when you think about it and we/I have tried it many times. It always seemingly works, when you have correct starting parameters and the optic/reflector is completely lit.
Imagine you are standing where the hotspot is (far enough away so that the beam is fully formed). You are looking straight at the flashlight. What do you see?
You see (with a reflector) the LED in the middle and the reflector, a big yellow circle with a hole in the middle around the LED.
Now consider this: when you have multiple identical flashlights and you hold them parallel so that the beams merge in the distance, you can multiply the candela value by the number of lights. It just adds up.
Now back to you standing there, looking at the light. The reflector is basically multiplying the LED. It’s basically increasing it’s area. So in effect it is multiplying the luminance of the LED, which is Candela per unit of area, usually mm^2. So when you multiply it with the apparent area of the reflecting part of the reflector as seen from your position (you are seeing what a lux meter would “see”), you are left with Candela.
This also shows why depth (focal length) doesn’t really matter for throw. Only the area of the optic (which is determined by its diameter) makes a real difference (and of course the luminance).
Here is a post from sma in the German forum where he thoroughly studied the Wavien collars (which is really great since no one else seems to have publicly done something similar). He went to a lot of effort to get a perfect focus for his measurements. Further down he calculated how much of an LEDs lumens they collect. He says it’s 75% (simply based on the angle). From this I deduced that the collar, after “amplifying” the remaining 25% by 2x (maximum: 2,2x) emits 50% of the original light of the LED.
BTW: he he even compared all three collar sizes to find the differences. As expected the large one is the best (he guesses that it has a bigger advantage when using larger emitters).
Blue Lasers are more efficient than blue (and white) LEDs at very high current densities. LEDs suffer from Droop, which is basically an “abnormal” reduction of efficiency at high current densities. Laser diodes don’t suffer from droop. They get linearly brighter with rising current.
The separation of laser and phosphor also means that the phosphor is cooler (because it is not heated by underlying blue LED) and that is can be cooled seperately (by putting it into glass substrate for example).
BTW: I find it great that you guys are planning to use heatpipes. They are cheap and very effective. In my eyes there are not enough lights using them.
I think a clever design would put the batteries in the middle of a light and the heatsink at the back connected with heatpipes. This way it would be easy to balance a light and one would be holding the light where the batteries are, which is usually not as as hot as the heatsink.
Correct and that was the idea. Hotter phosphors also reduce lumen output
All pictures belong to Lumiphous
This truly is an awesome topic!
But I am a bit confused, the idea is to have a recoil thrower made right?
If you make the calculation for what size and shape your reflector must be and design a head to house it all and the heatpipe cooling (I suggest 1 heatpipe bend in such a way the led is in the middle and the ends are giving off their heat to the head) a light can be designed and a foundation for bringing it to life is laid.
Is that translatable into estimates?
Like say a XP-L die excited by a 15 Watt blue LED chip or a 15 Watt blue laser?
At what (estimated) point is the laser more efficient?
How can one put a XP-L size die in a glass substrate?
In this case the glass is also the lens of the light.
It should also have a mirror / reflector behind it, unless you want spill and less beam (which is not the idea behind an über thrower).
I think it’s more practical to have the heat pipes (when we use a white LED in front of the parabolic mirror) go from the centre to the outer perimeter.
Since this will be quite large, i assume it will be a good heat sink and radiator.
With a laser you can shine it from behind the reflector, through the centre (only a tiny hole) on to the phosphor.
Hmmm… So you’re saying: Back to the drawing board!
And come up with a decent design.
The thing is, we can’t even try the concept because we can’t find a suitable reflector….
A prototype should be made to test the concept first.
…on the other hand, i still think it’s obvious it will do better than an aspheric or a regular reflector light, as long as the reflector is a decent parabola…
Regarding the validity of I=LA: this relation between intensity and area seems to work for most situations. However, for lenses with very low F-number there is reason to think it might not hold. Consider the following thought experiment which seems to contradict I=LA. Perhaps it will motivate you to take a closer look at Enderman’s thread linked above, or maybe you can give us some insight and figure out the contradiction.
Assuming I=LA is correct: while keeping the lens diameter constant, as the lens f-number decreases more light from the LED is collected and the beam size increases while the intensity stays constant. But on the basis of lumen conservation the relation seems to break down as we go to low f-numbers. For example, going from 0.5 to 0.25 f-number (focal length decreases by factor of 2) causes the beam size to double in lateral size and quadruple in area. In order for the intensity to stay the same the lumens collected would have to increase by a factor of 4, but this won’t happen. The lumens collected would only increase by approximately a factor of 1.6 (assuming Lambertian LED die surface).
Enderman’s analysis in the above linked thread looks at how these quantities (collected light and beam size) change with changing lens f-number.
Reading some of the links you posted, I see Ra tested some 22mm lenses ranging from 1.6 to 8 f-number and measured the same intensity for each lens. This certainly supports I=LA, but it isn’t until the f-number goes below 1 that Enderman’s analysis predicts significant decreases in intensity. So, hopefully you now see how there is some doubt about I=LA holding in all situations. Maybe there is some property of very low f-number lenses we aren’t taking into account that would resolve the contradiction.
Sorry but I have to disagree. It is very easy to imagine the extreme case where you have a lens like .1mm away form the LED, the divergence of the beam is so great that it almost makes a full 180 degrees, a huge spot, and the extra few lumens that it collects will not increase the intensity that much. Basically, the area of the spot increases exponentially, while the amount of lumens collected increases very slowly (as the lens approaches 0mm distance)
I think the reason Ra’s tests show no significant difference is because they were done with f numbers that were like 1.5+, and as you can see in my calculations there is very little difference at that point, the big drop happens at low f-numbers like 0.5 and below.
This makes sense, thanks.
I read that post, looks like the max is 2.32 to me 
not 2.2
This is great, I wasn’t able to find it on my own. Thanks!
And yes according to my calculations the 75% lumens collected at 30 degrees half angle is correct, you can see that in my spreadsheet 
Most “cheap” lasers like the ones you buy on aliexpress or ebay use 12v 3A to produce about 7W of laser power, which is ~20% efficiency.
And by cheap I mean the $250-$500 black laser module ones.
Obviously professional lasers might be more efficient.
.
.
14W of blue laser ~= 4k lumens, so about 55 lumens/W assuming 12v 3A 7W lasers are used.
.
.
The blue laser would cost about $400 for a good 7W one (the highest you can get for continuous running) and about $250-$500 for a set of single crystal phosphors.
Some advertise 10W or 15W but in the specs it says that that is only pulse power, the average power is 7W because you can’t run it at 15W continuously without burning out.
.
.
Single crystal phosphors do not have any reduced output up to 350C, so that’s why they are used with lasers.
@Enderman: Obviously you know laser more than me. That was just flashing thoughts. I never really use laser other than my green industrial pointer. Bought at ~USD60 in 2007 and I lost it in the site.
About the phosphors reduced output: My simple test proved the other way around (blue led based). Could be the binder though, not the phosphor itself
I’m talking about single crystal phosphors, not the stuff that is used on LEDs.
If you want to use a high watage laser to light up phosphor then you definitely want it to be a single crystal because it has much better thermal properties and output.
Some university research lab tested 14W and there was no damage done to the crystal, and no reduced output.
There are two places I know of that you can obtain the phosphors from:
http://www.crytur.cz/products/cryphosphor-single-crystal-phosphor/
The prices range from $250 to $500 depending on the size of the phosphor.
Ok, maybe there is a problem with “my” calculation with F<0.5, but nobody uses those types of lenses in a flashlight. At least not high quality ones from an optics manufacturer. They usually don’t offer those, maybe for this reason. Edmunds has F0.67 iirc which is already quite low.
Yeah, so basically as long as it’s at least around F1 then you can assume it’s pretty much constant intensity.
Outside my league… 