I want to build a BRIGHT LED variable beam light

[tobias]

Hello,
I thank you all for hanging out in an odd world.

I wish to build a variable beam and dimmable light using LEDs as a lightsource, that has an equivalent luminance output to a focused 250w + Halogen. I am a photographer and filmmaker, and I often need a good portable source of light, and for some reason I am always attracted to the DIY approach to innovation rather than the purchase of an expensive dedolight, or equivalent.

My thoughts were moving towards the use of a large LED array, using mirrors to narrow the beam of the array, into a mirrored box (a system used photographically to uniform a lightsource), which would then output through a condensing lens. My thoughts suggest that the mirrored box should contain a short focal fresnel facing the condenser which, ideally, would be a long focal fresnel (though they are hard to find).

Do any of you have any pointers, or possible suggestions that might reduce the complicity of the design, or shall i post a drawing? I guess i am looking for a beam that can be adjusted between about 20 and 60 degrees, and can produce as much light as possible. I am not too worried about power consumption, as my other alternative would be a 24v 250w halogen, which would burn more power and have more heat to dissipate.

Many thanks,
Tobias

 

[SafetyBob]

What you are talking about is probably to tall an order for here.

What stands out for me is the ability to adjust your beam from 20 to 60 degrees. That will take a lighting design engineer and a couple of weeks to figure out. It would be very, very complex as I understand your needs.

Now, on to the variable output.....that's not that bad of a problem. The main question is how deep are your pockets? I am talking electrical here and nothing else for now. Because of the amount of light you are trying to copy with LED's, you are going to need a bunch. I see the best solution would be an Advance dimmable LED transformer running on street power (mains), if you went portable, for the amount of power you are needing, a car battery (at least one) would be needed mainly for runtime. DC power would require a dimmable driver too, which is available from the Luxdrive folks (Buck Puck), but you would need a system to move all the dimmers together which could involve a gear system or belts, but could be done and probably be usable.

There are nice optics (spot and flood) made for the high powered LEDs that would work for you, but depending on how picky your needs are you might have to have a flood set and a spot set of lights.

Some of the other smart guys here might be able to approximate how may P4 Seouls or Q5 Crees it would take to be equivalent to a 250W light.....my bet is it will be a bunch.

Wait. If you wanted to go from a 20 to 60 degree beam, you could put a mask infront of the lights and just simply block the beam for your desired pattern, but you would be wasting much of the limited light of the LED.

Hope this helps some.

Bob E.

 

[VanIsleDSM]

You'd need a minimum of 25 LEDs. More at lower current would be better. You could use some 25 degree optics, and then some diffuser material to get a wider beam when you needed it.

 

[TorchBoy]

If you assume a Cree Q5 is four times as efficient as a halogen light bulb then you'll need to have 250/4 = 63.5 W LED lighting (ignoring losses to optics) etc. If each LED is running at a nominal 3.7 W then you'll need 62.5/3.7 = 17 Cree Q5 LEDs.

Edit: Hm, Van got in while I was suffering database down time. I think his figure is probably more accurate since at high power the Q5 isn't 4x as efficient as a halogen.

 

[tobias]

Hi guys,
thanks for your input so far. Maybe I am making my life more complicated than necessary. It sounds like getting the hypothetical output i am after would need about 6A/h, compared to the 10A/h that the equivalent halogen would need. maybe more portable and cooler, but much more costly in the short term.

the kind of focus that is used in the kind of lights that i would like to emulate is really primitive, as the bulb+reflector is moved in and out from a cast glass Fresnel with a pebbled back. I am not trying to project a slide, but simply vary the throw of the light witout losing many lumens. much like how the Maglites would move the bulb in and out from the reflector to vary an imperfect beam.

Breakfast time.

t

 

[snarfer]

It isn't exactly what you're looking for, but if you haven't seen them before maybe you should check out the Nila Lights, just to see what a similar commercial solution looks like.

I am afraid I don't quite understand your scheme for collimating the light using multiple fresnels and mirrors. Can you point out a light that does that? Are you proposing to use an array of fresnels coupled to an array of LEDs?

The way Arri and Mole-Richardson type lights vary the distance between bulb/reflector assembly and fresnel is kind of different, because that is a single light source. You would need to put all the LEDs together at one point somehow. But then how would you ever cool them? You'd need a custom heatpipe, and that isn't exactly DIY kind of stuff.

Anyway those old fresnels are horribly inefficient fixtures, and way too heavy.

Here is another interesting scheme with LEDs and reflector where they used multiple LEDs to simulate a single source into a reflector. Actually that light (the BebobLux LED60) would probably be exactly what you're looking for, except you'd have to exchange lenses instead of spot/flood knob. And of course it probably costs an arm and a leg.

Well anyway, how important is the spot/flood knob versus just changing the lenses anyway?

 

[tobias]

good point, i did not explain the purpose of the mirror in my mind. I had though that the use of a diagonal mirror would have allowed me to use a larger rectangle of LEDs to light up a square: the rectangle running parallel to the output beam, and the mirror reflecting them into the barrel, essentially changing the shape of the beam.

The Nila is nice, seems like a new version of a KinoFlo. Oddly neither the BeboLux nor the Nila indicate comparable light output, either comparing to existing lights or in Lumens. Creating a flat bank of LEDs would be a feasible option, though this would not allow for any control of the light. I would be interested in the architecture of the Bebolux.

Though the Arri fresnels are not entirely efficient, they are still the industry standard for most film and television, and remain pretty much the only continuous lighting that has an array of modifyers. It was for this reason that I was thinking to build an LED equivalent, as the simple lenticular structure should not be too hard to work with, and it is easy to use. The downside is that they are heavy and very pricey. Thus, for studio use, I am using old stage CCT Minuettes with 500w halogens and a lot of heat ( and they seem to be a good equivalent to the Arri, at a lower price and no modifyers other than barndoors ).

Thanks again,
Tobias

 

[snarfer]

Actually there is a link on the Nila homepage for photometrics. They have measured footcandles at various distances and with various lenses. I think Nila is not comparable to fluorescent such as Kinoflo, because Kinoflo has large flat source, whereas Nila has many discrete sources with ability to focus in to a narrow spot, which Kinoflo cannot do. Also Kino is soft light whereas Nila is hard light.

However, regarding the Bebob I could not find any photometrics either. I am not entirely convinced that this manufacturer is not playing some games about "tungsten equivalent" and so forth. It is an extremely interesting light from a thermal engineering standpoint though, because it appears that somehow 60 watts of LED power is mounted at a single point, possibly at the end of a heatpipe.

One advantage of LEDs is that there is very little heat in the projected light. This makes it possible to use plastic fresnel. So perhaps one interesting solution for you would be to construct an array of LEDs that would be matched with an array of Fresnels. That is, you could use something like part number 570 in this catalog (pdf), which is a 6"x6" array of 16 fresnel lenses on one sheet. Then you could make a panel with an LED corresponding to the center of each of the Fresnels in the array, that is, in the case of using that particular part, an array of 16 LEDs to shine through the 16 lenses.

You could make a sort of box with mounting slots at different distances for spot/flood control. Maybe Nichia 083 1W LEDs with high CRI would be appropriate.

Of course you might run into some problems with multiple shadows, so there would be an advantage to using lots of 1 watt LEDs as opposed to fewer 3 watt LEDs.

Really you should also think about what you are trying to control with spotting and flooding. Are you trying to control the intensity of the output, which you could also control with dimming, or are you trying to control the area of the output, which you could also control by adding or removing sources...

 

[TigerhawkT3]

I agree with SafetyBob in that this light is a pretty tall order. The variable beam part makes it much more difficult, as evidenced by the few lights that manage to create an acceptable (and reasonably efficient) beam at more than one focus (or de-focused) point. If you were to drop that requirement, you could try something simple like the TVL2.

I totally hear you about expensive off-the-shelf film lights. $1200 for a little LitePanel? Not on my budget, thanks.