A while back someone showed an alternative light engine using fresnals and mirrors in place of the reflector. Here is an idea that takes it one step further. It uses fresnals on 3 sides of the lamp to direct all of the light towards the condensor fresnal through a point source. I think that theoretically it should perform much better than a reflector. However, the primary drawbacks is that heat can melt the fresnals and we can't find fresnals with small focal lengths (approx. 5 inches or smaller).

I guesse another alternative would be to try putting the 220 fresnals closer than 220mm to the lamp. However, I don't know what the ray trace would look like, since the light would no longer be parrallel on the mirror side of the fresnals.

Oh yeah, one of the biggest advantages of this setup is that it should eliminate or severely mitigate the hotspot in the middle. Notice that light going up or down ends up nearer the the edges of the lcd. This should provide a nice even lighting with much more lumens (actually all of the lumens from the lamp) directed towards the lcd.

Hmmmm...I've been trying to do ray tracing for this design for a couple of hours with little success. I really like the concept, but it seems that the slightest imperfection will send my ray off to never-never land. Apparently, precision is paramount. However, I have never seen a more efficient design concept. All roads lead to the condenser fresnel. Bravo, phutton. Most impressive.

The concept is pure genius, but look at the path of this indiviual ray. First it moves downward and too the right. Passes through the fresnel, and hits a flat mirror. It comes straight back off the mirror the way it came, passes through the lamp arc, and goes through the top fresnel. It then reflects off three mirrors and enters the righthand fresnel, passes through the lamp arc a second time, and finally leaves the light engine. Apart from any slight offcentering of the original point of lighsource which will be magnified as the light path gets longer (Which we cannot avoid since our arcs have significant thickness) there is also a great deal of filtering.

The light in that particular spot, passes through the lamp arc twice, reflects off four mirrors, and passes through three fresnels. Ouch.

Of course light traveling in the opposite of that direction from that point will save a step and end up in the same place, so they will help to reinforce eachother. Still thats two fresnels, three mirrors, and a lamp arc to pass through.

But in my oppinion the biggest problem is this. This light engine is only efficiant on a 2d plane. It will only reinforce the rays from side to side and not vertically. (or vice versa) the only way for it to work bothways is if you were to rotate that frame about the center and draw out a circular reflector with that shape. Such that if you cut in in half, thats what you would see.

Here we are, draw this up on paint. It would have to look something like this:

You've hit the nail on the head, Mikau. This is a hard concept to translate to 3d. You need a curved fresnel completely surrounding the bulb. I still like it, though. The possibility of using all the available light is just too freakin' cool.

More like this.

Of course, now we've got to figure out how to get the bulb in there.

In my oppinion the concept in theory is pure genius, but in reality it doesn't seem possible.

You know what? I'm wrong. It does not need to be 3d like this. Because light will flat after it leaves the fresnel. Gimme a sec I'll draw a pic.

(edit) here we are. I guess it would work with flat mirrors.

Sorry for the idiocy.

Its a great design. Its basically the lighthouse fresnel. Speaking theoretically, the sides are better handled with the parabola - direct forward.

Hows about a spherical for behind, a parabola chopped in half for the sides and a fresnel for the front? You could even do a parabola slice for top and bottom. Cut your parabola into 4 pie slices for the perimiter and a regular setup front /back. You would need a very deep parabola, not the easily available kind.

Actually, the design does not have to be 3 dimensional. The actual shape of the bulb produces a 2-dimensional light output. I know someone showed the light distribution somewhere. It looked somewhat like a modified figure "8" with the lamp in the middle laying on its side. Here is a rough sketch.

As you can see, the light is emitted primarily radially. Almost no light is emitted along the axis. Also note that the axial component of any light will be bent by the fresnal into parrallel rays on the mirror side also. This will then be directed via a bunch of reflections to the lcd.

You are right, though, about the reflections being detrimental to the concept. all of the mirrors would have to be highly reflective for this to work. A simple 10% loss at each reflection would absorb 20-40% of the light, depending on where it originates.

I was seriousely considering the possibility of incorporating this design in my 2nd build. However, I cannot find fresnals with a focal length of about 4 inches at reasonable prices. If anyone knows of such then please let me know and I may be able to compare this design with a typical reflector.

Click to view attachment

Why do designs like that when the sphere reflector accomplishes what you are trying to do?

A sphere reflector only reflects the portion of light that is within the arc of the condensor fresnal. As much as 60% of the light does not even reach the condensor fresnal. In theory this concept would direct all of the light towards the fresnal. Another advantage would be that the projected lighting would be more evenly distributed, unlike the spherical reflector, which has a hot spot in the middle of the image.

However, in practice it is hard to say if this would work better or worse than a simple spherical reflector. The overall concept is much more complicated, with many more reflections to account for.

But, like removing the a/g layer, all potential breakthroughs start with an idea.

As an idea for aligning the reflector. One way of going about it would be to do the opposite of what we are currently doing. Most of us currently set up the condensor fresnal, then the lamp, and then we align the spherical reflector.

For this method it might be better to first setup the lamp and reflector arrangement. Then move the condensor fresnal back and forth until we get the maximum light through the triplet.

Absolute genius in 2D theory.

Probably impossible to position the light source correctly in 3d theory.

In practice, you're not going to be able to deal with heat, to get perfect mirror corners, to do a alot of things you'd need in order to make it practical.

It would seem that an ellipsoid would be much more efficient and only requires a single bounce on a front side mirror.

For example light leaving the bulb in the 11o'clock direction
1) passes through the top fresnel, looses 10% (90%)
2) hits the upper left mirror, looses 5% (85.5%)
3) hits the upper right mirror, looses 5% (81.2%)
4) bounces off the upper middle rear mirror, looses 5% (77.1%)
5 passes through rear fresnel looses 10% (69.4)
6) passes through bulb outer shell rear looses x%
7) passes though bulb inner shell rear looses y%
8) passes through arc looses z%
9) passes through bulb inner front shell looses y%
10) passes through bulb outer shell looses x%


I'm not sure how much light is left from the original but it can't be more than 30%. This also neglects bump mountings that remove portions of frensels or mirror and block some light paths. Ellipsoids capture a lerge pecentage of light with a single bounce per source ray.

You nailed it buddy.