Forgive some of the obvious material, this post was designed for my web site.

I'm trying to build an LCD projector on the cheap here. That means re-using a small, inexpensive LCD panel from a broken DVD player or a PSOne monitor or such. The broken DVD player is my resource of choice, find one that won't play DVDs or has no sound or something and pay little or nothing for it. DVD players can almost always take video in from an external source and display it on the screen. This gives you the screen and the input electronics, don't ignore the expense of getting RCA video to play on that old LCD computer monitor. I've actually got a couple old laptop screens laying around and a controller circuit to run them is going to cost $80, so I stick with the old DVD players for this first attempt. Besides, it makes a nice small projector.

Next cheap product is the lamp. Couple points here, first the human eye doesn't notice the color variations of different bulbs, forget about white balanced 6000 kelvin color perfection. When you are watching something from a halogen light it looks white to the eye, just like everything in your office under those fluorescent bulbs look white. I've done a lot of looking and I've found the perfect bulb. A point light source 250 watt 120 volt halogen. The style is called JCD, exactly like the JC type but at household voltage levels. $5 at LightBulbsDirect.com
The matching base is part number L80054 for $3, not listed in the online catalog.

You get about 20 lumens per watt with a halogen. That's 5,000 lumens. You can get 60 to 100 lumens per watt of metal halide, but the bulbs are $20 - $50 each and the ballast is from $60 to $150. If I can shell out $200 on lighting, I've paid for half of an eVo already (the eVo rocks, built around a $40 bulb instead of the typical $400 replacement bulbs, the whole eVo unit sells for $400 right now at lumenlabs.com).

Another feature that makes this design a little different is the lack of a collimator lens. Don't need to collimate the light if it's already collimated by the parabolic reflector. Don't need telescope quality here, just something to give a nice even field. I intend to use flexible mirrored aluminum, either by deforming it or by taking slices out of it radially. Of course you can avoid this if you prefer to use two fresnels, but you could also avoid this by going to the store and buying a nice LCD tv, but what fun would that be? Our $5 bulb is at the focal point 1" from the back of the reflector and there is a very tiny quarter inch reflector attached directly over the front of the bulb (if the bulb isn't hot enough to melt aluminum) to reflect uncollimated light back through the focus and into the parabolic mirror. The shape of the parabolic is interesting too -take a 9" deep dish parabolic and cut off the top, bottom, and sides to make a nice rectangle the size of your screen. I'll be making these available shortly as soon as I finish the parabolic mold.


Innovation doesn't stop there. Much discussion is given to controlling the noise of the fan and to controlling the leakage of light. The baffles take care of both these tasks. The fan is completly enclosed by the projector and mounted to rubber pads. Distracting fan noises are kept to an absolute minimum. Air flows in from the right, between the LCD panel and the glass heat barrier, over the hot bulb and reflector, and out the left. Extra light and fan noise are all contained within the projector.

Well a halogen lamp definitely does not look "white" to the human eye.

I have had 3 projectors with halogen lamps and all three definitely look yellow on the screen.

I appreciate the feedback, that's certainly not what I would have expected based on how the eye perceives different lights. Thanks!

The eye does compensate for white when objects are being illuminated by the entire halogen's spectrum. The problem in an LCD projector is that it filters the light to just the red, green and blue components. If there is no blue component, or very little, to start with then white will appear very yellow. That’s not to say that halogens won’t work it’s just that higher color temp lamps are much better suited.

DJ

I'm betting you could adjust the red and blue in the LCD to compensate, the only question is how to do it, hopefully there will be some pots in the circuit board.

Most allow you to adjust through the OSD, but you will be cutting alot of red and green, even with the blue maxed, so you will lose contrast in the red and green, and also alot of overall brightness.

I like the idea though, just I personally would use a different type of bulb.

i am sorry but your whole setup won't work very well.
Safe time,money and effort and stick with the standard setup..

I think that you may have trouble with a dark spot in the center of your LCD. But try it first, if there is a dark spot, then modify your lens design. Or moving the bulb away from the LCD slightly (further into the reflector) will make the light converge slightly and make the spot go away. That might be an easier and more effective solution.

Which innovation to you object to? The bulb, the baffles, the parabolic reflector? My background is in telescope optics, I know a bit about bouncing light around.

The bulb is dimmer, but the parabolic setup I am using will utilize almost all of the light, whereas the spherical reflectors typically utilized still waste massive amounts of light. Consider a thought experiment with a point light source, a fresnel lens 220mm to the left of the point, and another another lens 220mm to the right. With a spherical reflector you manage to make use of an amount of light equivalent to what would hit both lenses and absolutely no more than that. With a deep dish parabolic reflector and a very tiny secondary reflector I manage to utilize close to 100% of the light. My only concern is that I might melt the aluminum, if I do then I'll add heat fins to the back.

It's really not any "money and effort" to build a parabolic mirror of this quality (in contrast, it's hard as heck to build one of telescope quality). What I've done is print a 9" wide parabola on paper, cut it out and traced it on wood, including horizontal overhang on the left and right of the parabola. This tool will fit exactly into a 9" dish (old dog dish) and will rotate in a circle, especially if I pin the center with a finishing nail. This will allow me to shape the plaster in the dish to a very nice parabola. This then becomes the mold for a concrete negative over which I get to beat a flexible mirror with a mallet. The whole process takes a couple of evenings. It would be faster if I could find a deep dish parabolic to start from, but such short focal lengths are unusual. The short focal is desirable for this project in order to capture the maximum amount of light. See the diagram above, it's 9" wide and has roughly 2.5" of curve and a focal point 1" from the back. Notice how little light escapes the edges and how small a secondary reflector is needed to bounce even that light back through the focus and into the parabola.

If I wanted a "standard setup" I'd take my credit card to the electronics store.

I'm also considering the heat, I think a tiny aluminum reflector attached to the bulb might be in danger of melting. They make halogen bulbs out of quartz because glass would melt.

Since I'm not constrained by a collimator focal length I could shove my tempered glass right up to the edge of the reflector and then put my LCD a few inches further away. Maybe even run two sheets of tempered glass with alternating air flow if needed. Almost no light loss there and no secondary reflector. I'm liking that idea a lot.

I remember why I needed the secondary, it's not about running light through the LCD, it's about running collimated light through the LCD. Any rays that are not collimated won't make it through the projecting lens. Taking out the secondary still has 90% of the light going through the LCD but only about 60% of it would be collimated. I'll let you know if the secondary survives direct contact or if I need to back it off a bit with spider supports (super thin supports that won't be seen in the projected image).

Sorry, what I meant was a dark spot on the projection resulting from no light passing through the LCD plane in the center. If you move the bulb slightly into the reflector, the light will bend slightly inwards, eliminating the void in the center.

Thanks! I understood no light would go straight through the secondary mirror, from there I rambled on with thoughts about removing the secondary. Basically I'll start with a parabolic spotlight and see what it looks like. If it evenly lights a surface then no worries, otherwise I've got several ideas to experiment with. I might be able to move the bulb off focus as you suggest if I can compensate with the fresnel, or I might drill a 1/4" hole in the secondary center.

Well it ain't really innovations since a lot of people already tried similar things. The bulb produces unbelievable much heat, is dim and has the wrong color temperature. And believe me, you can clearly see the difference even between 5000K and 6000K.
With the reflector you won't get even light unless you use an offaxis parabolic reflector. Many people have tried it (including me) and no one has accomplish good results.
Uneven light is an absolute killer for image quality.
There is a reason why everybody uses the conventional setup. In every slide projector, film projector etc. you find exactly the same spherical/condenser lens setup. #
In commercial projectors nowadays parabolic/elliptic reflectors only work in combination with multi-lens arrays.


I know this sounds hard but i just wanted to keep you from spending money and time on something that most likely won't work.
If you still want to try it go ahead, and post results

Good luck.
And I would like to see the step by step on the castings if you have time.
I think it would be very interesting.....

Made the tool and then made the first mold. Look how shiny smooth the clay is. Of course it will shrink as it dries so I'll need to add layers to it.

To make the tool I printed and cut out a template in heavy card stock. Then I put the tool curve up in a vice and let the concave card stock rest on top of the tool. I looked through the gaps and marked the places to sand the tool more. I then sanded off the marks and repeated the process until I was satisfied the tool was perfect.

I spent a lot of unproductive time working the clay before driving the shaft end of a nail (sans head) into the center of the tool. The pointy end of the nail fit through a hole I drilled int he center of the dish (and on into my workbench). Work proceeded quickly from there, with the tool eventually turning over the clay with near perfection.

I'm considering frosting the bulb with a lite sandblasting if my initial spotlight comes out unevenly lit. I'm a bit paranoid now about getting an evenly lit surface.

As far as white balance goes, if my digital camera can compensate for any color weirdness by adjusting RGB, I figure I can get pretty darned close by turning the RGB pots on the LCD circuit, specifically by adding a bit more blue (the opposite color of yellow, and the frequency most deficient in the halogen spectrum).

The biggest advantage to the parabolic approach is that a whole lot more of the light gets usefully collimated. The disadvantage to building the parabolics yourself is obviously one of quality. You don't want to pay for a full fledged parabolic mirror, but parabolic "reflectors" are much more reasonable. Edund Optics has a 12", 18" and 24" available for $40, $51, and $79.



Edmund Optics parabolic reflectors

I have a flexible sheet of aluminum mirror, complete with protective plastic sheet over the mirror. I've been wondering how to make this stuff hold its shape once I beat it into a parabola. I just realized that when I square it up to 7"x4" by cutting off the top, bottom and sides I will be able to mount flat sides to the resulting shape, and the flat sides can be rigid enough to hold the shape. When I make the negative mold I will make it a 7"x9" rectangle so I can form the sides when I form the parabola, guaranteeing a perfect rigid shape.

An interesting chain of ill-logic just occurred to me.

First and foremost I figured out why all previous parabolic designs failed to produce uniform lighting, which quickly led to figuring out how to fix them all. It has to do with angles, the angles from the bulb to an area near the center of the parabola are much larger than the angles from the bulb to an equivalent size area near the edges. Smaller angle of light equals fewer lumens near the edges.

The fix for this was so diabolically wrong that it just might work. Remember, with this design I am taking advantage of hugely more light from the bulb than in a spherical reflector, it turns out the gods of fate and photons want a bit of that bonus light given back. So what we do is create a variable neutral density transparency to dim the center hotspot until it matches the edges. I’ll do this by taking an underexposed BW picture of the projected white image such that the corners are nearly black and the center is slightly brighter (nowhere near white). I can adjust the contrast with post processing for the perfect image. Then take the negative of that picture and print it out the correct size in high resolution on a transparency and mount it to the tempered glass. An advantage of the photographic method is that it also accounts for any vagaries of my manufacturing process. This might be an iterative process, but I’m not imagining it will take more than a few tries to flatten out the hot spots.

Might work, but hard to manage.

Just a few things:
When you place the the bulb in the focal point you get a dark spot. To prevent the dark spot you have to place the bulb before the FP to get collimating rays. Just like you want to do it. This will lead to a hot spot. But you don't get just a hot spot. Because you don't have a point source you get various hot-spots. It will look something like a car headlight on a wall.
Because of the nature of a projector the size of every hot/dark spot get's multiplied.
Also if you place the bulb for collimating rays the rays won't be parallel (of course)and the second fresnel lens won't work probably. This may result some aberration and loosing light. Although for a small bulb like the halogen this won't be that much of a problem

I want to avoid changing the focus, so I've been toying with the idea that the secondary has a smallish hole in the center. The secondary is actually not trivial. Once I cut the parabolic primary to fit in a rectangular shape and eliminate the parts that won't project my image, I have some curved pieces missing from all four sides of the parabola. I figure the secondary will have small curved sections to match. I'm not even going to build the secondary until after initial testing, the whole thing works ok without it.

It's all about having fun experimenting with different possibilities!

Thanks for the feedback!

Use a blue tinted transparency and it will help your color temp. (of course, really hard to get right)

Actually a friend of mine has a graduated set of color discs from his film processing days that one looks through to see how various amounts of color change would affect a final product. I forget what he called these, but we could use them to find the industry standard gel that corrected halogen color the best...

If you like, here are a couple of 8 inch diameter parabolic reflectors.....

Click to view attachment

$11.86 from here....Stagelightingstore



Click to view attachment

$17.90 from here....PartsExpress



Tgreenwood

what about just gluing mylar or foil over your plaster mold and using that?

Thanks for the fantastic links! I had found some "parabolic reflectors" from Testrite that were not polished and which I wasn't sure about the "parabolic" nature of, but these look really nice.

I started to get excited about the standard 600W halogen bulb for these too until I saw the average life expectancy was a whopping 3 days. Whoo Hoo!

Don't really have to wait for the ceramic clay to dry too much before adding the concrete. I blocked off the top and bottom with heavy cardstock pinned into the clay, stretched cellophane over the cardstock and clay, and carefully scooped concrete in. I should get an interestingly shaped brick out of this soon.

is it going to be able to stand up to the heat from the lamp?

Just what every wife wants to see on her kitchen table, a parabolic brick. The brick will dry for a good long time and will then be used to shape the mirrored aluminum into parabolic form. One thing I learned today is that carefully pinned and properly reinforced heavy cardstock is no match for 10 pounds of concrete, the blocked off edges bowed outward a half inch. Fortunately barely dry concrete can be sanded (crumbled?) easily. I did not try to block off the shorter ends, instead I'll probably cut those with a hack saw in a few days. I let this set for all of 4 or 5 hours before removing it from the clay mould, used lots of paper towels to suck moisture out before opening it, it's not dry but it's not going to change shape either. I removed it because I was mainly concerned about uneven drying if I left it in place with only one surface not wrapped in cellophane. The back has some small air bubbles but is otherwise nearly as smooth as cellophane. The barely wet clay mould that held the concrete is completely untouched except for the pin holes, it held its shape perfectly, too bad I have to crack it apart and put it back into the clay bucket.

Neither the clay nor the brick will be in the projector, these are tools (stone tools?) to make the parabolic mirror. Time spent so far: 2 evenings.


(the brick isn't cracked, one of the lines the cellophane made just stands out more at this angle)

This should make a nice DIY parabolic. You might want to consider dumping the halogen for something more practical though. Even with your reflector setup, you'll find you want more light, and you'll want less heat. A 600W halogen? That's an oven, and you'll be making a monster projector to compensate for it. And for what? You want white light, and like all those who've tried it, you'll probably find cutting the LCD's RGB down to a usable form for halogen unacceptable.

Like others have mentioned, you're parabolic will have obstacles, but it looks like you're on the right track. Good luck!

Thanks! The bulb I'm looking at is still the 250W point source halogen. Not too hot, and about 5,000 lumen. I figure I'll get three times as much usable light as the "spherical-reflector-with-fresnel-collator" method. So really I'm hoping it will be equivalent to a 15,000 lumen bulb in a traditional setup. But really it's just an experiment, I'll pay for better lighting when I build my "real" projector around a 13" laptop panel once I fork out $80 for a controller that can drive it.

I've pulled the front focusing assembly and some other fun optics out of a dead infocus projector I got for free, looks nice enough. The free Viewsonic wasn't even worth ripping apart, everything was tightly integrated and the front lens only had a tiny 1/2" opening on the back side - anybody want it? I've got a monster focusing lens still in a NEC MT810 that I got for free if I could fix the vertical lines in the highlights caused by oscillation in the green channel - first it looked like a red and blue problem until I figured out LCD colors were subtractive, I made some decent progress before letting the magic smoke out, killed the entire green channel, but the bulb still has 1000 hrs left on it if I decide to hock it on eBay and I'll rip the lens and I'm sure I'll find some larger front surface mirrors than the infocus. I actually pulled the motorized focus/zoom motors off the infocus lens because I figured it was harder to build the circuit to drive by reaching up and pushing the buttons than to just reach up and turn the lens.

That and the existing LCD panels knocks a big bite out of the total cost, I figure the 13" projector will cost $60 for the ballast, $30 for the bulb, $80 for the controller, and another $40 for the fresnels. That puts me around $200 for the 13" version.

I made a projector like that for a friend awhile back and it turned out really nice. For the fresnels you could save about 30 by getting them from anchor optics if you find a projection lens to match. They sell 406mm (16") 10.75" x 10.75" fresnels for 5 each. These lenses cover a 13.3 nicely, but aren't big enough for a 14.1. They probably won't be wide enough if you have a 13" widescreen though. I used a 4.5" dia x 6.5" precondensor, a 150w HQI bulb and a 400mm doublet and it was and not noticeablely different in brightness from my 400w projector with wayyyyy better colours. Here are the links to the parts I used.


condenser
400mm doublet
fresnels

The only improvement I'd advise is finding a triplet on similar length. The corners are abit fuzzy with the setup I made. Only noticeable while trying to check the time using windows.

ps... good luck with the parabolic setup. I actually have the 18" parabolic reflector from anchor/edmonds but never got it working particularly well. Hopefully I'll revisit it soon. I tried to get it working with the 10mm arc of the 150w HQI but I think the main problem with it was its not a true parabola. I'd like to get something functional with it sooner or later. Cheers

Why not use a wood turning lathe. With a stable wood like Cherry You could make the nagative parabolic then place a piece of aluminum or stainless pinned with the otherside and then lay the metal over while it turned?

DavesPlanet

I admire your DIY ingenuity in creating parabolic molds well done !

One thing I'm not sure from looking at the parabolic shape created is if it is deep enough to capture most rays.

I guess time will tell, looking forward to the reflectors progress..please keep posting/photo's as you go !

Cheers

The shape is built directly from the graphic in the first post. The focal point is an inch off the back wall. The left and right edges wrap around another inch in front of the focus, the truncated top and bottom are probably on plane with the focus. After initial testing without a secondary I'll see if a secondary can increase the brightness. I'm thinking the secondary will be angled to punch more light to the outer edges where we think the image might need some boost in order to keep it more evenly lit.

Last night I worked wet concrete over the surface to fill in the imperfections and squared up the top and bottom.

I've been experimenting with folding stiff paper over this shape and I'm quite pleased with the results I get when I cut four lines at 45 degrees from the center not quite reaching the center. The place this straight line overlaps then forms a smooth curve that is also cut out. What I'm thinking is that soon I might be able to upload a 9x11" graphic with the curves and lines that fold into a very reasonable approximation of this parabola, especially if the top and bottom flat pieces are cut out of stiffer material. I said my background was in telescopes, well my background also includes a lot of origami. I just never thought I would use origami to fold a parabolic mirror.

watching with interest..now if I ever get the time

The lenses I have available:



Pulled the small one out of an Infocus, it had motorized focus and zoom but I pulled the motors off because the lens works very nicely by hand and I couldn't figure out why I would rather push a button than turn a ring. Motors will be used elsewhere, maybe for motorized keystone correction in the larger PJ I'll build. The infocus lens has a rear aperture of just over 40mm and I fully expect to have to use a negative lens to get the best use out of this since they are made to focus parallel light going directly through those small LCDs. Anchor Optics has some nice negative lenses for under $10. This is the lens I will use to build my small PJ around an old DVD player.

The larger lens is the real beauty, removed from a very expensive (and very dead) NEC MT810 that was donated to me (I also scavaged the two good LCD panels to get D Taylor up and running again and to help fund my own PJ project). I'll use this when I build a PJ around my 1024x768 13" panel. It has 55mm rear aperture. I also kept the trim piece from the front of the old PJ, that will really add a fantastic look. I'll leave the motors on this one and figure out how to wire them up to push buttons. Probably need a small H bridge to supply +/- 5V to these. And of course I'll probably need a negative lens for this as well.

Think I've finally found the DVD player to savage, a nice Polaroid PDV-0700 with some slight case damage and 480i resolution.

Now that the focusing lens and cube thingy are out of my NEC the green portion of the light engine projects directly onto the wall.

I have to say, confusion doesn't even begin to describe my emotions when viewing this image. Obviously only a rectangle out of the center is used but it accounts for a big chunk of this light, and there are massive aberations directly in the center of the beam. I can imagine this pattern is caused by the multi-cell diffusers in the light path, but I can hardly understand why any sane and sober human would want such an effect. Neither can I make the logical leap from this pattern running through an lcd, an RGB prism, a focusing lens, onto the screen and somehow becoming a flat, evenly lit image. I just don't get it, this has got to rank in the top 10 weirdest things I've ever seen. And you guys were worried that my parabolic might not produce an even light source???

What you are looking at is the result of a lens array or fly’s eye lens or integrator lens. They are used in pairs and contrary to what you are seeing on your wall they do in fact even the light. The objective lens resolves an image from the LCD and at the LCD the light will be even. Because you have removed the objective lens, what you are viewing is the the integrator alone.

DJ

Pairs, eh? Have to look when I strip the NEC, I kept most of the interesting optics from the Infocus and I think I did come up with two lens arrays of different sizes. I suppose the second one could have been countering the effect of the first by focusing on the LCDs. I think several people have thought of making a panel of collimated light that way by putting an LED behind each eye. I'm not sure that idea would work, it would take an awfully lot of LEDs to match the lumens of a small MH.

I also came up with a flat sheet of glass made out of laminated angled slices an eight of an inch thick. Everything looked completely normal when I looked through it, I couldn't figure out what that was for. I think maybe each lamination had a different type of coating on it but that might have been my imagination.

Still sniping eBay ads for a suitable DVD player in the $20 range. No joy yet.

That one is a polarization light recycler. It lets linear polarized light of one particular direction pass through. The light that doesn’t get through is rotated by dichroic layers and reflected forward All the light that has past through will line up with the polarizes of the LCDs.

Look through it at an LCD monitor, it’ll give you an idea of its operation

DJ

Edit:
If we could find a way to get these layers in thin plastic sheet form, we could make a larger version to suit our projectors.

Now that I've had a chance to experiment with the light path I see how that fly eye pattern works, each eye is just like the bigger lenses, but because I didn't focus the light in the green image above each little beam expanded into that image. Everything does focus nicely on the LCD plane.

Below are pictures of the pattern I ended up using, a Mercadian style cut (think globes) and a picture of that pattern heat pressed onto the surface of a mirrored sheet of aluminum. The horizontal lines on the pattern are the cuts, the angled lines are where the majority of the cut lines up when folded into a parabolic, the long verticals are guides. Top and bottom quarter inch are cut off in the picture because my printer won't print edge to edge. You should be able to use the image below directly, I believe its sized for exactly 11" high. I think you need a solid parabolic back support to really make this work, it's just that same curve from the first post cut concave out of some nice wood and laid behind the long axis. Thinking of using paneling glue and some small finishing nails to hold the mirror to the support. I did learn that it's easier to be super accurate when making a convex shape, like the clay forming tool, than a concave one like the back support. I think it's because of how the sanding works out, convex with a long sanding block is much easier to control.




(yes, I am in fact a short, squat alien with hammerhead features)

Turns out the aluminum is considerably thicker than paper, I ended up cutting out the shallow V shapes instead of overlapping. The mirror fits perfectly around the brick and clay, a beautiful fit. Spent quite a bit of time with the rubber mallet (don't use a hard hammer, it marred the surface) and used a sheet of paper between the brick and the mirrored surface. I would also recommend cutting the second side of each little V once the aluminum is in place and you can see just how it actually lines up with the expected line. Now I have a considerable amount of tape holding the back side of the reflector in shape, I'll have to get a spot or arc welder out to finalize this before mounting it to the back support. All in all I'm extremely pleased with how the reflector came out. Now we'll have to see how well it collimates.

Hey Dave..How did it turn out ?!

Cheers

geezz lueezz DP,...your hideous!

Whats the update on the reflector DP, how's it coming along?

just one more note on your original design...you want the fan at the lamp end sucking air out of the enclosure..the fan on the other end blowing air in will have little to no effect.
also if you are dead set on using halogens the filter sheet would be a better option than trying to compensate for the colors in the OSD. If you live up north you might actually like the heat this thing is going to produce .

Any progress? I've been following intently hoping for some leeway to be made on the eliptical front. G'Luck

One way to make a large/deep parabolic mold is through spin-casting. I attempted this for an 18" telescope mirror with mixed success:

http://math.brown.edu/~howison/rab2004/report-node1.html

Everyone is on this guy about using a Halogen light, basically i think he is doing it for the same reason I am, I want to prove to myself I can actually build one of these things before I spend a fortune, and I got 2 bases and 5 500W Halogen lights for $17.00 shipped, if the image is a little yellow of the bat I dont care, once I prove the concept and my skills to myself we can always upgrade the light later!

Just my 2 cents

-The Trusted One