I was sitting here explaining how things work in the projector to my roommate, she
asked a lot of good questions, like 'why don't you make the reflector bigger?' and such.
It occurred to me that there are a lot of misconceptions out here about this part of the projector.
So I thought I would address a couple of them because I can't sleep and made the diagrams
real quick to show my roommate.

First one, the spherical bowl reflector!!!!

Attached to this post is a diagram showing an ideal point light source, a reflector, and a Fresnel lens.
There is no condenser( we will get to that later).

First off, lets get some terms clarified:
Collimating light - To have all the rays of light be parallel, or going exactly the same direction.
Collimated light - Light that has all of it's rays going in the same direction.
Focal Point - The point at which collimated light entering a lens or mirror gets concentrated to.
Center Point - The center of a circle or sphere. This is in no way a focal point, though if you put
a light in the center of a sphere, the light will be reflected back to the center.

We might throw some more in there later.
Ok, in the diagram we can see how the light is at the center of the spherical reflector and at the
focal point of the Fresnel lens.
- So all the light leaving the left side, gets reflected back to the center and out the right side.
- Since the light is at the focal point of the Fresnel lens, the light hitting the lens gets collimated.
This is why the first Fresnel is referred to as the collimator and/or collection lens.

Now at first glance, a half sphere reflect seems like its an awesome idea, but the geometry of
it puts a damper on things. If you look at the green rays of light, this is light that is not hitting
the collimator, so it isn't doing any good. The reflector extending past the congruent triangle formed
across from the Fresnel is not contributing any more light to our image.

No biggie, it might help a little.... maybe the light will bounce off the sides... at least it's going the
right direction... it can't hurt....
WRONG
The truth is, you don't want any light hitting the collimator that isn't coming from the focal point.
stray light blurs the image and defeats the purpose of the collimator.

So paint the sides of your light box black and problem solved!!!!
NOT YET
You still have all that light going back through your expensive 400W bulb. Which equals more heat.
The truth here is, I haven't run 1000 tests to compare the lifespan in both cases, but in the vast
majority of the world of materials engineering, heat=death.

So any extra reflector that you aren't utilizing is actually doing harm.

In addition, the mixing bowls that people use to acheive this effect are used with the presumption that
the small flat spot at the bottom is only a small percentage of the whole bowl. But if you look at the
diagram, it is possible that it is the only area that could be utilized. This means that there is no benifit
and only harm.

'Well, I get a bright image with the reflector!!!!'
That is probably true, I am painting a worst case scenario here, but you are probably closer to the
worst case than the best case with this setup. And you could be getting so much more light!!!

I am going to have to replace this next image with two later,
but GIMP crashed and I am getting tired and want to get these posts out

Use the Reflector as a collimator!!!!

Well, here is why this doesn't work so hot.
In the following diagram pretend that the Fresnel lens isn't there, and that it is not collimating
the light that is hitting it. In fact, ignore the light going to the right of the source for right now.

If you put the light source half way between the shell of the reflector, and the center of the
reflector, you will be at the focal point of a spherical reflector. Remember, that the focal point
is where collimated light goes to when entering a mirror or lens. Well, the reverse is also true
light radiating from the focal point will leave the lens collimated. This leads to the idea of
use the reflector to collimate the light and loose that first Fresnel.

First off, in to have a chance of working, you would need a reflector that equaled the dimensions
of your LCD, including the corners that would get cut off by using something round.

Second, most of the time, when people come up with this idea, they do what I told you to do, 'ignore
the light leaving the right side of the light source'.
To address this issue, it is suggested that a second reflector be used direct all the light to the left,
so now all the light is being used!!!! hooray.

Well that is nice and under perfect conditions is true.

But what makes the 'parabolic' collimator a bad idea is the geometry again. Look at the lines that bounce
off the reflector, they represent the relative intensity of the light at those points. As you approach the
edges of the reflector, the vertical density of rays varies as traverse the surface of the reflector, giving
you uneven lighting as a result of the 'perfect condition'.

Now this variable density is a concern with all the set up we have. But it is exaggerated in this set up
because of the wide variance in angles.

Also, 3/4 of the light gets reflected once, and the other 1/4 gets reflected twice. If these were ideal mirrors
or even optics quality, this wouldn't be a concern, but it's stainless steal polished with a drill and a sock.

On top of all this, there is the functionality of your bulb being in the way, and since all of your light is
reflected, there is, in the 'perfect' set up, a perfectly collimated shadow of your bulb and socket in your light.


Now, this image implies that you might be able to combine it with a Fresnel and get the best of both worlds!!!!!
That would be nice, but remember that focal point thing? Collimated light entering a lens will converge on the
focal point. The Fresnel will collimate the light coming off the right side of the bulb, but it will converge the light
that was already collimated by the reflector.

Now, there are some more advanced concepts involving off axis spherical collimating, but these still suffer from
the ray density variation problem.

Here is a quick post to answer:
'Why do you need a condenser lens?'
The most common answer I hear from people is that changes the focal length
collimator.
This is true, and an integral part of the purpose.

The main outcome of the condenser lens is to grab more light. If you use the
formula for calculating the focal length and more importantly the back focal length
of a pair of lenses, you can find a distance between the condenser and the collimator
that will put the condenser right up against the light source. The means that the
compound lens formed by the two separate lenses now intercepts a much wider
arc of the light.

Now combine that with an appropriately sized reflector, and you double the gain.

Notice that the diameter of the reflector isn't what is important, it's the angles of light
that it spans across. With that said, the more surface area the more area to radiate heat.
the same amount of light is hitting the 3 inch reflect, as hits the 8 inch (assuming correct
placement), but the 8 inch has 4 times the capacity to deal with the heat. This is not a
problem with 'cold' mirrors. Cold mirrors let non-visible spectrum of radiation through.
Which means the don't get nearly as hot.

Now the diagram shows how it intercepts more of the light. It is also designed to illustrate
how few rays per inch are hitting the lens as the angle to the lens become much more acute.

So the balance is between even lighting and overall brightness.

This is why longer focal lengths are used in combination with more powerful lights. I will post
in a technical drawing thread a design for a light engine that does in face utilize twice the light
coming off the bulb without the uneven lighting that comes with other attempts to do so.

Now my final note for now.

Every surface the light touches has an efficiency, so every bounce, and every refraction has
a brightness cost. And if you compound efficiency losses, they add up fast.

This is an illustration of the difference between the focus and center of
a spherical reflector.

Shown is a 100mm diameter mirror with a 55mm focus.
The focus of a spherical reflector is half way between the center and
the shell. This puts the center at twice the focus from the back of the
mirror.

When light comes from the focus, it will be reflected out as collimated
light (blue rays)

What we want the reflector for is not collimating, but for doubling the
light on the target, so we put the light at the center where the rays
bounce straight back through the center and out the other side. (red)

Now, is quite common for the people selling these reflectors to get it
wrong as well. Sometimes they measure the focus off the front edge,
sometimes they measure the center off the front edge and call it the
focus. Sometimes they just pull numbers out of dark places and put
an official seal on them. Hopefully this will help you out

Very nice write-up, withe a few minor points that I'd like to add.


While the essence of this is true, the effect on the image is incorrect.

Stray light does not degrade the image, it is just that when it originates from an incorrect origin, it cannot be directed into the projection lens, which means that it doesn't help the projection any. It becomes waste light int he front half of the projector.

We want to keep that waste light to a minimum if at all possible, since if it reflects back to the LCD, and from there to the projection lens, it will wash out the image, and reduce the available contrast.

As for the image focus, that is all a matter between the LCD and the projection lens itself. We already have a somewhat scattered light source. If we had a perfect point-source light, there'd be no need for a projection lens at all, we could simply put the lamp behing the LCD, and the shadow of the LCD would become the image. Real-world lamps however would never allow that to happen, so we use lenses.

The whole point to a focused lens is that any light that meets the following criteria: 1. It originates within the FOV of the lens at an appropriate distance from the lens plane, and 2. It strikes the lens. As long as it meets these criteria it will come to a focus at a location corresponding to the source location, to the best of the lens' ability to do so. So if we take a single pixel as an example, some of the light that leaves that pixel will hit the projection lens dead center, some will hit it to the sides, and some will hit it bottom or top. In the end, though, it's the lens' job to collect all of that light, as well as everything inbetween, and put it on the projection screen at the same place, while simultaneouly doing the same thing to the light from every other pixel on the LCD. Where the light originates from behind the LCD doesn't matter, so long as it goes to the projection lens..

A good point. You do not, however want your reflector to be too small. It is better to be somewhat too large than any too small, since the limits of the reflector are used for the parts of the LCD which already get the least light, due to inverse square law. As such, it's better to err on the side of too big than too small.

Also, in terms of heat=death... That's true to a point. Heat in excess of the intended parameters = death. Appropriate operating temperature is key. Take a car engine as an example. If the block temperature is under 90 deg C (194 deg F) then the engine does not operate well. You will get excessive wear on the bearing surfaces, and overall inefficient operation. Electronic devices have both a minimum and maximum operating temperature. While your LCD will fail if you let it get too hot, it will also fail if you make it too cold, so you might not want to have your cooling system intake sucking from a pool of liquid notrogen, hey? Aside from the likelyhood of something breaking from thermal stresses, the semiconductors themselves wouldnt work. Considering that these lamps are engineered to operate in sealed, or semi-sealed boxes with no active cooling at all, the extra heat is well withint hte design paramters of the lamp. Those, too are something that there is evidence to suggest that they don't like being too cool. (I got rid of a "green push" problem with one lamp when I removed active cooling from the equation.)


LOL! Seriously.

But of course your point remains. We can get better mirrors. My own experiments with stainless steel indicate that with a spherical reflector used in the recommended method the results are that you get 125% of the light that you would get with no reflector at all. The LL pro reflector has twice the benefit, bringing that to 150%. Now I didn't go so far as to use the drill when polishing my stainless reflector, but I did use a reasonable polish. Still, that doesn't speak for a high efficiency on the reflector.


I like this diagram, as it takes into account the main reason why a parabolic reflector isn't a good choice, but it fails in one regard, and that is that the rays coming from the reflector will be bent when they hit the collimating fresnel. You can think of the collimator fresnel as a series of circular prisms which bend light coming from the focal point to being collimated. (Actually, you can treat ANY lens this way.) The properties of those prisms don't change because the light hitting them is already collimated. The end result will be that the now collimated light from the reflector will instead be warped into converging rays which will then be further warped by the collector fresnel, and directed to places other than the projection lens, which we all agree is a bad thing... Or at least not a useful thing.

And one more little quick tip. If your reflector is spherical, then it's very easy to find the correct distance from the outside edge to the center of the lamp, even if you don't know the measurement of the sphere's radius:
Click to view attachment
Measuring the reflector this way gives you the distance from the outside edge (back) of the reflector to the center of the lamp.

I won't have an effect on the sharpness of your image, but if you take an extreme case:
a Mirror placed at convenient spot passed the focal length of the combined system. You
get light projected through the system that most likely won't be in focus, but since it's off
center, you would be able to see a bright spot somewhere near the edge of your projected
image. You shouldn't have any stray mirrors in there, and they probably won't be at the
right place anyway.

So yes, the effect of light bouncing off the inside of your box and coming through the system
is not a serious concern in the vast majority of situations.



Probably, but it's not a matter of vaporizing it instantly, it's about the overall life of the bulb. They
were designed to operate without active cooling, but they weren't intended to be shining on themselves
in an enclosed space. They won't explode, but wouldn't it be nice to get an extra 10% lifespan out of
the bulb by simply trimming down your reflector? (10% pulled out of my rear, like I said,'I haven't
comparative trials')


I like that one, but people should know that will only work with a minimum size:center ratio. If the edges
aren't coming off the flat surfaces after the point of contact, like it is in the diagram, this doesn't work.
A way you can tell is if the mirror kind of swirls around instead of sliding.

1. Right. My point being though that stray light woun't affect the sharpmess of the image. It was just a minor correction to avoid misunderstandings.

2. Well, I don't have studies on it either, but the lamps are designed to work with reflectors. I thin that a "cold miror" reflector probably will reduce arc temperatures by more than a reasonable design. I'd expect this to fall under "diminishing returns" very quickly once the reflector is in place at all.

3. This should work for any reflectors large enough for our purposes. Any reflector that covers an arc of more than 90 degrees will work for this, and I'd suggest that if we're capturing less light than that, we're better off trying to get the lamp closer to the image than working with any reflector at all. The basic 15" LCD/fresnel at 220mm approaches this factor