Well, I am admitting defeat (for now) on using CFL's as a light source for my current box design. I have jumped on the bandwagon for the 150w CDM ceramic bulb for the time being, but I think there is a lot of good info in these posts for those that wish to try CFL's. Somewhere about page 4 I am starting my progress reports on the 150w ceramic. So peruse the threads and enjoy the journey!

After soaking in as much information as I can from this site, I have finally broken ground with my own projector! I have decided to go the less travelled route and I am attempting to use CFL's (compact fluorescent lights, to help the search engine) as a light source.

The current parts list is as follows:
- (6) GE Daylight 26 watt (don't know where I got 23 watt from...) CFL's
- (6) Daylight 30 watt CFL's
- (1) Magnavox 15MF605T, dismembered with AG layer intact (possibly being removed in the near future)
- (1) Beseler Opaque Projector lens, 18" EFL
- (1) 550mm FL Fresnel lens
- (1) DIY Front Side Mirror
- (1) Emergency Fire blanket (used as reflective material to line the light box)
- (1) Mirrored reflector box using actual mirrors
- (1) Plywood box as a "test" enclosure

I am currently in the testing phase (when are we not?), so I will post pictures as they become available. As of right now, I can get an "ok" projection, its a bit dim, but it could be watchable if necessary. When contrast and/or brightness is increased via the LCD, the picture becomes washed out, but it is more visible. Let me get some pictures up to show what I am working with...

This is a picture of the wall I am projecting onto. This sure as hell doesn't help the brightness issue. I will have to construct a screen soon.
(you can see a white piece of paper on the left, the brown wall I am currently having to project onto, and the white painted doorway trim on the right)

Good luck!

Here is the test box in all its ugliness. Starting from the top; the mirror with adjustment turnbuckle, fresnel, LCD (you can see the business cards I used to help reinforce the circuit board on the LCD), and the light box assembly. A more detailed pic of the light box to follow.

And here is a better look at what makes this projector a challenge. I have (6) bulbs, 23 watts each (100 watt equivalent each, allegedly), oriented perpendicular to the LCD (tried having them point at the LCD, but as others have experienced, not very good coverage). The board just under the bulbs is lined with the emergency fire blanket material, as are the two angled panels to the left and right in the picture. Speaking of which, those two panels MUST be angled just right to reflect light onto the LCD, otherwise the edges are much darker than the center. The bulbs are about 8-9 inches away from the LCD in this picture. Ultimately, I will move the bulbs another inch or so closer to the LCD, but I am limited by the angle and length needed by those side panels.

I am curious ... why the choice of the fire blanket material? Seems like for those planar surfaces, a true mirror would work great. Unless you are leveraging the dispersion of the reflection to even the light out?

I am very curious to see what your arc image at the triplet looks like holding a piece of paper just inside the triplet while taking the pic.

I have the blanket material already, I wasn't planning on using it exactly as pictured though. If it is beneficial, I will bend a piece of sheet metal to better focus the light on the LCD, which is what I originally had planned for the reflective blanket.

I too am curious as to what that light image looks like. It might be part of the reason why my projected image is so dim. That and I have a 90" diagonal screen on a brown wall.

daveoxide,you gotta be kidding me.6- 23 watt-(100 watt equivalent) CFL's? If you can pull this off with anywhere near acceptable results,your green as they come.I don't know why you want to try.CFL's diffuse light.6 or a dozen even with a huge custom made reflector will give only fair results at best.But I applaud you for the effort.

6 CFL's work great on my projector.

TGreenwood did some testing with CFL's as well, and had decent results ... though with a small panel and a large CRT Lens.

It's all about getting collimated light through the last fresnel and out the triplet, so if you do it with a very efficient specialized rig or generate a crapload of light and throwaway 90% of it ... same result.

Good luck to you!

Placing the bulbs about 5" away was as close as I could get without any hot spotting. Mirror also worked very well.

Wow,I'll have to check your plog out.I never would've imagined this would work.

lonewolfmichigan<<inserts foot in mouth .6 CFL's!Very impressive taco stand!I really should'nt comment on topics I know very little about.Lesson learned.
Hope it works out good for ya daveoxide.

Taco's is only 7" though.
I think more CFL's will be needed for a 15" LCD.

I really need to get my butt moving on mine.

I agree with nitrogen. Or check out the bulbs that Arcanus99 is using.

Well, I added two more bulbs. I now have a total of (8) bulbs, 26 watts each (not 23, brain fart), 1600 lumens each. That brings the grand total to 208 watts producing 12,800 lumens. Once I can borrow a decent camera, I will post pics of the resultant projection. If I do say so myself, it looks pretty good , I am quite happy with it.

I did have to add a cooling fan afterall. After running it for about an hour the screen started to get darker. Since I added the fan (a 12v fan running at only 6 volts to keep it quiet), everything has been fine and dandy.

I added a picture of the (8) bulbs burning bright. Man I really need a better camera...

I was able to take a screen shot with enough light in the room to compensate for my camera being lame. The projected image is about 96" diagonal, projected onto a Home Depot pull down blackout shade, which will make a good screen if those darn horizontal lines relax and even out. Hopefully with time it will relax and I will get a nice screen for the $35 I spent.

Here is the projected screen shot at about noon, with the sun trying to come in through those two windows on either side of the screen (and through a door you can't see behind me). What you see is pretty much exactly how it looks in real life.

I have the home depot screen you are talking about and mine have not flatten out and its been a year I dont think they will. Glad to see the cfl is working for ya

Crap, really? What thickness did you get? They had 6 mil, 8 mil, and 12 mil. I had to get the 8 mil because they were out of the 12 mil in white. I wonder if I should exchange it for the 12 mil because the 8 mil cups pretty bad on the edges, in addition to the horizontal lines from it being rolled up for who knows how long.

Thanks for the heads up.

Nicely done!

As for the screen, have you considered ironing it? Maybe you could put a sheet below and above it, while you apply a low heat iron?

I was actually thinking of hanging it out in the sun with some weights hanging from it to help relax the creases. Ironing could work as well. I wonder how the vinyl would like the heat from the iron?

What if you can insert the dismantled white sheet like thing from the LCD's BACKLIGHT ASSEMBLY?

Well a layman's logic says that if 7+7w CCFLs can illuminate 15" LCD then to illuminate 120' the ideal lighting should be 14x8=112w .What do you think?

These days OSRAM is selling square shaped CFLs of upto 150w with exactly the same lumen efficiency that is 80lumen/watt.What i think is that what counts is the lumens per square inch at the LCD.You can also experiment with and without the front collecting fresnel as i guess it doesnot work without the collimating one effectively.

Good point.
The square 64W CFL flood lights i've been playing with just seem to throw that much more light than the 64w curly bulbs I have.

I just thought it's because the curly bulbs were designed to throw light in all directions & my flood lights were made to throw light in one direction.

Well, I ended up on returning the 8 mil shade and picked up the 12 mil shade. Much better!

The new screen has horizontal crease lines, but they are minor, and they aren't noticable during a projection unlike the thinner shade. Also, the sides of the shade stay straighter (more straight?), with very little cupping. I think if I added some weight to the bottom of the shade the cupping would disappear completely.

I'm much happier with this one.

What about just some good ol blackout cloth for your screen? I have been using it and its pretty good. You could even get fancy and mix some of that maximud on it for greater results

Cheers

See the basic funda is that we want uniform illumination at the LCD panel.The same is achieved inside the LCD MONITOR by the backlight.What if we take the LCD BACKLIGHT apart and make use of the white plastic screen and place much powerful CFLs in place of the existing CCFLs?
We want a point light source as that is the requirement for proper efficient working of the FRESNEL only.Foe no other reason we need a fresnel.So leave frenel and hence the point light source WOW.These days OSRAM is selling as efficient CFLs as our acclaimed METAL HALIDES.That is 80 lumen/watt.

I succesfully removed the AG layer from my Magnavox LCD TV!

I tried the Mother's wheel polish method but it didn't seem to do much to the antiglare layer on this panel. So I figured I would bite the bullet and strip the AG layer.

And to be totally honest, I don't see too much of a positive effect after removal. My brightness/contrast *may* have been improved slightly and the colors are ever so slightly richer, but that is about it. I thought I would have noticed more of an improvement, but I guess this antiglare layer isn't that much of a hinderance compared to some.

I did notice, after removal, that the AG layer is not as opaque as some other monitors that I have stripped the AG layer off of. It is actually quite transparent (which may have to do with me polishing it before removal, but to be honest, I didn't see any change after polishing for quite some time).

At least the gut wrenching part is over. Now I am waiting on my brighter bulbs to arrive (eight 30 watt CFL's @ 2000 lumen each) and my lux meter.

So many toys, so little time!

Well, my bulbs arrived today. I still haven't picked up a new camera, but it is next on the list! So for now, you have to suffer with my 2.0 MP low light shots....

The room ambient light is brighter than indicated in the picture, due to those dang windows on either side. But the sad part is, even with the room lit as well as it is, my camera STILL can't handle the low light situations.

fantastic work! Sure this is alot of wattage, and you could get a brighter image with HID, but I see a silver lining here...

So you have 16000 lumens... but only light that passes through the lcd within ~7 degrees of the normal makes it through the triplet. even with the reflectors you have, I would be amazed if more than 10% of that 16000 lumens is collimated within 7 degrees... probably more like 2-5%... can't wait to see the lux measurements to verify (or not) this guess. Now, if you can manage to collimate a reasonable chunk, you could start reducing the wattage required and compete with HIDs. I think it can and will be done (low wattage, high lumen CFL PJ), but probably not in a reasonably sized enclosure. Can't wait to see the improved photos and the lux measurements. keep up the good work!

Has anyone tried a multiple-focus fresnel like the one here http://www.3dlens.com/multiimagelens.htm (except probably w/ less squares) and putting a reflected CFL at the focal point? Would that help at all?

Hats off to you dave.I have got to know that ONLY COLLIMATED light can be converged with the collector fresnel and that is a must i guess.So to collimate that i hope the easiest way would be to place a mirror in such a position that the CFLs reflect light onto the LCD through that mirror.

That is basically what I am shooting for right now. My current reflectors reside at a +/-70º angle to reflect the light into (onto, through) the LCD. It isn't ideal (as seen by the first drawing), but it works enough for now. The second drawing shows what I would like to try next. Basically each bulb is back on its end, only each has its own reflector to direct the light into the LCD. I am curious to experiment with this design more. I will wait until my lux meter arrives before I change anything so that I can get real data to compare.

I've been playing with ideas like that for a while, with different setups & bulbs, but it always comes down to a massive enclosure to collimate any reasonable percentage of light... anyway, here is an interesting option for light:

http://www.eiko-ltd.com/Products.aspx?CatI...ndex=DT55/65/RS

pushing 90 lumens/watt with 6500K color temp... set these on end like your last proposal and you'll have a massive amount of light (along with a massive enclosure)... a bit pricey though at around $65 for a twin ballast and two bulbs... 9600 initial lmens per pair at 110w though

Wow, that is a very efficient bulb @ 90 lumens/watt! My original bulbs were 61.5 lumens/watt, and the newer ones are 66.7 lumens/watt. Also, the 6500K is too blue of a light, the 5000K bulbs look much more natural (with the old bulbs, my blues were too blue, if you can believe that). Also the bulbs I am playing with are only $3.50 per bulb, so buying eight of them isn't going to break the bank. My bulbs cost about $0.12/watt, or better yet I get 571 lumens/dollar spent. But if we can find more efficient CFL's for a reasonable price this CFL light engine stuff just might work out pretty well for future projectors.

yup... the bulbs are not that expensive per lumen >400 per $, and even better when considering 14000 hr life, but you do have to buy the ballasts first. too blue at 6500k, huh? first I've heard that anywhere... I thought the 5000k bulbs' spectral distribution looked pretty good... might have to try that as well

the last idea you have is what i've been thinking about in various configurations, trading off between % of collimation and enclosure size. when you use the reflectors in a divergent manner like that, you get more collimated light by reflection of otherwise unusable light that is bent twice the angle of divergence of the reflector toward the lcd normal... problem is, the closer you get to the required incidence, the more distance the light needs to travel before reaching the other divergent reflector... for any real fluorescent source, one of the dimensions of the bulb inevitably leads to an enclosure in the 10-12 ft range (a bit big for most applications ... i'm currently trying to figure out what losses are associated with folding non-fully-collimated light (some light turns 'back down the tunnel' and makes another go after losing ~5% at every bounce). fun stuff... can't wait to get my controller and panel to do some real tinkering instead of just pondering.

what gets me is that there is a perception that the point light source is a highly efficient light engine... the fact of the matter is that even with a 100 lumen per watt source, if you only get 10% to the screen at the proper angle, you're at 10 lumen/watt effective... and if the spectrum is wrong, you adjust the colors from there and lose more. Get 10% of the light collimated and I think CFLs will compete well with other options... the spectrum is just right for the brewer filters, so less light will be lost to adjusting the colors. who knows, could even be more efficient some day. But, without at least an honest attempt to collimate the light, 14^2/360^2 (approx collimated light from the bulb) is a VERY small percentage

Have you tried setting your camera on a table and using the timer function (if it has one?)... looks like your biggest problem is motion blur, not resolution (2MP is certainly enough for this application... 2 page spreads in Sports Illustrated were done with 3MP back in the day). you WILL have significant noise, but at least straight edges will not be blurred (e.g. window casing should not be a blurred line). I think your camera is using a long shutter to gather more light... too difficult to hand hold. worth a try anyway.

Just tried the timer in a dark room, no go. It won't capture the projected image. But I do agree that I can't hold it steady enough for those long (that is relative) shutter openings.

I also looked up the specs on my camera... (don't laugh too much) the ISO range is 100-200! Yeah, I think that explains why it won't take pictures in low light situations.

Edit:
I just remembered that I have another camera, its a video camera, but I have no way of getting the video to the computer because the camera is a HI8, and not digitally friendly. I have a USB adapter coming (for a different project) that will allow me to connect the video camera to my computer, but until then... And the crappy part is, the video camera works great! It can capture the projection beautifully, but alas, no way to digitize it right now.

very odd... i was sure if you just tried that last shot again with the timer you'd have luck... the exposure is right on that shot, but with too much motion blur. were the conditions the same? what model is your camera? wish I were a bit closer to napa, I'd drive over and help you out with my camera

The conditions were not the same. Here are two pics with the closet light on and the closet door open (the room is darker than the first shot that was taken with light coming in through the windows)...

You can really see the noise in this picture. The camera is really straining to take the picture. And the picture is not nearly as bright as what is seen in real life (the upper left corner is clearly visible in reality, but not in the photo).

much better... that is more what I expected... you will never get rid of the noise, but I'm used to seeing noisy photos... motion blur bugs me though (pet peeve I guess). Things are looking pretty good there... can;t wait to see measurements and more results of yours.

ROFL!! I took some lux measurements... make sure you are sitting down otherwise you will fall down from laughing so hard...

Lux Measurements 27-Mar-08
3.3 5.0 3.2
4.0 6.5 3.8
3.3 4.6 3.2

Room Lux = 1.6
Avg. Lux = 2.5
Area (m) = 2.9
Lumen = 7.25


So yeah, that really sucks! Thank goodness I have a trick up my sleeve to hopefully double the brightness over the weekend (that isn't really saying much, lol). I'll post more results when I have them.

not laughing at all... confirms what others have said, and what the challenge is. Your results are consistent with the numbers I've seen thrown around for light transmission and collimation requirements in DIY PJs:... can't remember where I saw all these numbers here, but they are good enough for a rough estimate:

biggest consideration for fluorescents: ~7 degree within field fresnel normal to make it through triplet (i.e. collimated light)
fresnel transmission: 95%
LCD transmission (depends a bit on lamp's spectral power distribution and how well it matches the lcd's color filter: 6-9% for HID bulbs, 12-13% for fluorescent (can't remember where i saw these)
triplet transmission: 75%

So, to estimate what I thought you should get, I just multiply all the transmissive factors together, and make an estimate of the amount of collimated light you're likely to get with the setup you have. First the transmission of collimated light:
LCD*fieldFresnel*triplet = 0.12*0.95*0.75=8.5%, so no matter what it looks like behind the LCD, you will not get more than 8.5% of the lumens on screen. But 8.5% of 16K is nearly 1500 lumens, that would be nice!

The problem is the amount of diffuse light we can get collimated before passing through the field fresnel.

Given that light exits a fluorescent lamp from all points and in all directions, the light exiting a bulb within a +/-7 degree cone is: (14/360)^2=0.15% (ouch). So, without reflection, you would get... 8.55%*0.15%*16000 = 2 lumens.

Neglecting reflection losses, and you should get about 4X this because you have three additional bulb images (one off each side reflector and one off the rear reflector). This is a bit optimistic in that it neglects losses due to light passing through the bulbs. Anyway with all the above factors and simplifications, you should be getting: 16000*.0855*((1+3*.95)*(14/360)^2 = 7.965 lumens. Are you sure your meter is working properly

You will get more light with your new design (more reflections that lead to a 14 degree cone of collimated light), but really need to collimate a significantly higher percentage of light to get what most would consider a usable projector. Look at it this way... if you could get 10% of that output collimated, you'd have a 140 lumen projector. getting that 10% is REALLY difficult though in a reasonably sized enclosure. Another approach would be to place the highest output, slimmest bulb you can find in a VERY long light tunnel (widening from the bulb width at the back to the screen width at the other end. I've played around alot with different bulbs and geometries with this idea in a spreadsheet, and I'm currently leaning toward a single 42W 3U 4-pin compact fluorescents in a 5 foot long divergent reflecting tunnel (similar to your multiple reflectors, just deeper and with much shallower divergence creating many more reflections and ultimately collimating a much higher percentage of light. I estimate I can get about 60% of the light collimated (nearly 2000 lumens)... if I get anywhere near that, I have a usable projector consuming less than 50W. My biggest concerns are vignetting (due to a single bulb without overlapping high/low areas from multiple bulbs) and hotspotting due to the difficulty of laying the reflective mylar cleanly over such a large area. Funds and time are in short supply right now, so I will not really get started on this for another month or two. Please try it out and prove me wrong if you feel so inclined

Wow, thanks for doing all the number crunching for me! I just returned home from a buisness trip and I have come down with a little bit of a fever. Crunching numbers is not something I wanted to do right now.

I did some testing a few days ago, one CFL bulb, a fresnel lens, and one sheet of BEF (brightness enhancement film). I simulated my current setup that I have in my projector, and it was ugly. Very little light was projected onto the wall (I didn't have my lux meter then, but I will run those same tests again to get some numbers). I then installed the BEF near the fresnel, very little change if any. Next was the BEF right at the light source, BINGO! With the BEF right at the light source the fresnel projected a very bright spot a little smaller than the bulb and reflective housing. Woohoo, collimated light!

This is my plan for my next light engine. Stay tuned!

Dave,

I have some CRT projection lens.
I'm wondering if one of those in place of the fresnel would give more light.

for the Flood light CFL i'm using I stuck the CFL in a box & put the CRT lens on the side of it & I was getting a good focused circle of light on the wall.

more info please... sounds interesting. do you have any pics? I'm having trouble understanding how you've arranged things.

Your line of thinking is similar to mine, only I'm using HID
My 14.1 WXGA+ Plog

well, basically, I took the box from the CFL floodlight I bought, cut a 3" hole in the side of it, stuck the light & fixture inside with the bulb facing the hole, inserted the RPT lens & plugged the thing in.

It looked like a flash light shining on the wall, except I could read the serial number stamped into the reflector of the fixture.

Here is what i'm using



http://alltoolssupplyinc.com/home_files/65wfloodlight.html

I have two of these.
I've allready stripped them down, mounted them & stuck a dollar store aluminum food pan behind them to reflect the light.
It was darn bright.
I was focusing the light onto my 15" LCD & I was getting a projected image.
but because I was just testing I didn't have sides around my "light box" so the whole room was lit up from spillage so I couldn't even guess how bright the image actually was.
Then the FFC's on my lcd started popping off & everything went in a box.

I now have a 7" LCD from a portable DVD player & i'm looking to use just one bulb to power it.
Though, i'm retrofitting this into an RPTV so I believe I don't need it to be that bright simply because CRT's arn't that bright.

very similar, but I wanted it to be simple to construct (i.e. no tools I don't already have!)... I did a simple test using a small mag-light bulb inside a reflective mylar cone, and it certainly collimated most of the light, but because it was a point source deep in the cone, the light was not nearly homogeneous (lots of specular reflections that would probably look weird projected through the LCD). Yours will work better with a HID because it is parabolic... getting evenness could be tricky though. I did think about using a frosted HID to solve the specular reflection issue, but figured i'd be wrestling heat issues with mylar, and if I went that route would probably end up where you are anyway (aluminum). Can't wait to see your results... as long as we are sacrificing the small form factor of the commercial PJs, we might as well get more than 11% of the light collimated to the LCD... at least that is my thinking!