Materials :
- 4* SSC P7 C-Bin LED Emitter (Max 900lm each, Thus 3600lm)
- 4-LED Reflector for Cree Emitters
- 4* 21mm*7.3mm Acrylic Lens (F ~= 10mm)
- A concave lens (F = -75mm Dia = 50mm)

Ok here is the plan :



Explanation & Calculation

Leds working through driver. They're all fine Rays going through reflector and reaching Acrylic lens(of course its convex by the way)
Because Leds are positioned in the focal point of acrylic lens(1cm), Rays going parallel.
If this part is wrong I think there must be a way to set leds so rays will go parallel. It shouldn't be impossible
Rays reaching the concave lens. Because they're coming parallel they're refracting like they're coming from concaves focal point.

So heres the calculation : Concave lens FL is = 75 mm. Concaves diameter = 50mm Fresnels Fl = 120mm And fresnels height is 90mm

We want to know if the ray is sufficient for fresnel or reaches the fresnels height

X = 4 So that means its not reaching. x*2 = 8cm
8cm means the full height
LCD's height = 77mm so its okay.

What do you think?

3.6kLum is not so much for projector with big LCD.
These LED engines are suitible only for small industrual LCD projectors. (they do not "eat" too much light).

Any LCD is going to "eat" about the same amount of light. 90-95% of it.

If you can start with 3600 of [b]collimated[/i] light, which you put through your concave lens and re-collimate with a fresnel, then you'll be fine. If you could get 3600 lumens into the LCD then you'll have between 180 and 360 lumens on-screen, which is plenty. (I've got 250 for a 130" diagonal screen, and it's plenty.)

3600 lumens from a more traditional lamp is a problem because of inverse square law, but if you're starting with directed light, then there are fewer losses.

However... Just because you can draw a nice diagram doesn't mean that the real world will cooperate with you.

What I'd suggest instead is to eliminate the concave lens entirely. The LEDs won't produce collimated light without additional lensing anyway, so instead what you want to do is restrict the "beam angle" from the LEDs instead. The narrower you can keep that beam angle, the better, (so long as it properly covers the LCD.) Keep the LEDs as close together as possible. It's best if you can keep all of the light sources within about a 1" circle (However, if you can fit more than 4 in that circle, then I'd say go for it!)

Build the rest of the projector conventionally, and voila. I'd say that you want at least 1,500 lumens to hit the LCD screen for a projection of about 100" Remember that light which does not hit the LCD, and does not come from within that 1" circle doesn't count, since it won't be projected.

Yesterday, I found a single LED(about 0.4-0.5" diameter) which provides about 2000 lumen Max. At 3.7v 9A. (9A needs a serious cooling system,I know)
And a 45 degree collimator. And in description of the collimator it says its efficiency is up to %90.
Lo lets says its efficiency is about 80 percent.

According to my calculation , 45 degree in the focal point of fresnel is getting almost all rays that LED provides.

That means i can provide almost 1600 lumens to hit the LCD screen. I want about 60"-80" projected image.
Is that really enough? If it is then i'm going to order the led and other stuff like cooling system and drivers.

And by the way i wanted to know what is wrong with 4 LEDS in a 1.8" diameter circle. Because that's a better option for me (like economical and availability)

SupraGuy thanks for your replies by the way. You're really helping.

Okay, let's run some numbers...

A 45 degree cone of light at 220mm produces a circle of light 220mm in radius

A circle with a radius of 220mm covers an area of 1,520cm^2

A 15" LCD is approximately 30.5cm by 23cm, or 701 cm^2.

This means that a 15" LCD would capture about 46% of the light from your LED.

Your 45 degree beam lens is 90% efficient, so that means about 41.5% of the light hits that area.

Start with 2000 lumens, and assume an even distribution (Which it won't be, but we can pretend.) That means that approximately 830 lumens hits the LCD, so you will project between 40 and 80 ANSI lumens to the screen. Not too bad... Not great, but not that bad.

My LCD's size is 4.8 inch. And Fresnels focal length is 120mm

45 degree in the focal point of fresnel creates 12.7cm diameter circle light

That makes (12.7/2)^2*3.14 = 128.1 cm^2
LCD dimensions = 11cm*6.5cm = 71.5cm^2

so LCD will capture 71.5/128.1 = %55 of light

%90 efficient so = 55*90/100 = %50 of light

I'm planning to use 4 leds (750 lm each) thus its 3000 lumens

%50 of 3000 lumens is = 1500 lm

i guess that makes 75-150 ANSI Lumens and i think its good

but i want to know 4 leds in 1.8" diameter circle is OK?

Most Lcd's pass only 5%. I think only 2 or 3 people here have documented a 10% Lcd.

I know that. What i'm talking about is reflectors efficiency.

The 1.8" circle is problematic.

Basically, the size at the projection lens is the ratio between the distance from the lamp to collimator fresnel over the distance from the collector fresnel to the triplet.

Let's say that you're using a 300mm triplet, which would be about 330mm from the LCD. Split setup, so that's 320mm from the fresnel.

120mm / 320mm = 0.375. Invert for a magnification factor so it's 2.67 factor in your 1.8" initial diameter, and you'll find that your LED circle image is 4.8" at the triplet. This then becomes the minimum diameter to capture all of the light from your light source. I don't know of too many triplets that large in diameter, though they do exist.

Of course, if you're using a 200mm objective lens, then the magnification factor becomes more favorable, but I still think that 1.8" is too big to work well.

My triplets FL = 137 mm and collector fresnels FL = 160mm. And i can use a 120mm FL or 150mm FL Fresnel too.

Does it works with these? Or it's just about objective lens' diameter?

Okay, plug in those numbers to what Ive done above. The magnification fator won't be that large, but you'll still need a 3" or larger diameter triplet to make it work.

I have an another idea. As you know, my 4 leds was in a circle
Now i think its better to place leds alongside.

I calculated and it seems I can capture almost 7/10 of light with this pattern :

(there is a fresnel between lcd and light circles of course)

But i'm confused about needed triplet's diameter size. We are capturing light from where light placed and then we are making the rays parallel. Right?

Why wouldn't our triplet can't capture light when its diameter size is not enough to capture from light circle with magnification factor. Rays are going through. and then they are converging.Aren't we trying to focus the screen? Not the lights?

The problem isn't in getting the light onto the LCD, it's with collecting the light back to the triplet. That's what the fresnels are for.

The "needed diameter" of the triplet is because the fresnels will magnify the source of the light in exactly hte same way that the triplet magnifies the LCD image. The result at the front of the projector looks like this:

Click to view attachment
(Blue dots represent LEDs, the green line is LCD, cyan lines are fresnels. thin blue lines are the light cones fromt he LEDs, which are collected back to points at the front of the projector.) This is assuming that you use one fresnel in front and back of the LCD, though an unsplit configuration woudl be the same. You need to gather all 4 of those points into the triplet in order to project the image within that light. The image is a LITTLE exaggerated as well, but that's the principle.

If it were just a matter of getting light to the LCD, we could just use flourescent tubes, like the original backlights...

There's another problem there, too.

Click to view attachment

As you can see, the light isn't going to be very evenly distributed. Some areas of the screen will be illuminated by 3 LEDs, while others will only be illuminated by one.

I understand the diameter issue. Thanks. and i know about the distribution problem. I thought that if i expand the space between circles, distrubtion would be fine.

Here's another question : What if i put 200mm FL fresnel on the light side and 120mm FL fresnel to the triplet side.
Then the magnification would be inversed. Line width of leds is = 8cm. With this setup it would be 4.8 cm on the triplet side when its un-magnified right?

And if i use 4.8cm or 5cm diameter triplet it would be ok?

Best design choice for big LCD ir much more small LEDs - 5mm or 10mm in matrix and of cause if you can - in two (2!!) floors
Yeah, the working examples are found in some forums

Those 5mm or 10mm LED's efficiency is about 40lm/W. What i'm talking about here is almost 80-100lm/W
And that needs a wider diameter triplet(very wide)

Someone posted up in the Trading Post a very wide triplet (195mm FL, and 6+" diamter) at Surplusshed which would work for a large variety of lamp/fresnel combinations. If it wasn't so freakin' expensive, I'd DEFINITELY want to experiment with that.

The idea of reversing the fresnels... Well, points for thinking on your feet, but the problem is that those cons that converge back to dots start to diverge again after that point, which means that the light scatters again, which is back to the original problem.

The next idea is the "fankenfresnel" (You can use that as a search term, it'll com up with a few hits) which is a segmented fresnel to collimate two or more light sources, which can then be gathered by a single collector fresnel to the objective lens.

This is the same concept as the "flyeye" which is what the idea for LL's LED light engine will be based off of. This one is more suited to only small panels, though, since the overall size has the be dealt with.

Oh, and for hte large bank of LEDs, 40lm/watt isn't so bad if you can capture a much larger percentage of the light. Since we only get 10-12% of the light from a radiating source with a "traditional" build, it takes 400W to get a couple hundred lumens on-screen. If you can capture 70% of the light from an LED array, 100W is plenty, and will probably out-perform the 400W single lamp. This is also spread out over a larger area, which means lower temps, even for the same amount of total heat.

i hope i will test today my 10W/20W LEDs.. i have only 3 qty but it will me give approx "how stuff works" with standard triplet and 10.6" 720p screen.

So did you ever buy that 4.8" lcd? I was wondering if you could review it if you had it. Swapping it in for a psone screen in a delta II/CFL build would be pretty sweet on the cheap.

I have all the equipment except the light source

http://www.vitrolight.com/specs/LMS480KF02.pdf

This is the datasheet of the lcd. I bought it in eBay. Sellers name is vitrolight.

Yeah I saw the ebay link in your other thread. I was wondering more about the picture quality and such as experienced firsthand.

Picture quality is fine but if you mean the picture quality with the projecton ; Well, we'll see that when i get the LEDs