hi,

Just wondering if any one knows roughly how many LED's would be required in an array configuration to come close to the light put out by the Metal Halide Lamp thats available on the website? assuming say each LED has a Luminous Intensity rating of 40cd

thanks

LEDS will not work as they are not a point light source.Anyway,even if you could get enough at a high enough lumen rate,it would actually burn hotter than the MH bulbs we use now.

ya there is lots of post around here that are realy long telling about his idea i will try and find some for u

Why would they burn hotter? I was under the impression LED's run practically cold... you could leave them on for hours and still be able to touch them with out being burnt.

Short answer: lumens/watt is too low.

The LED idea came to me too. I was convinced that it could work and I even ordered some super bright white LEDs from China. Then I did more research on the forums and found that many other people have had the same bright idea . What convinced me the most that it may not work well is the amount of lumens per watt is too low. Or in other words, the "brightness" per amount of heat is too low.

This said, I would still like to see some results of someone who has tried it (if anyone). I still might experiment with the LEDs I got from my friends in Hong Kong.

Yes. One or two LEDs run quite cool. They also each dissipate only a very small amount of power.

I hvae been experimenting for other purposes with LEDs, and have found that the amount of available illumination is really quite low. Puting an array of 36 LEDs together in order to make a lamp works okay, and doesn't get hot enough to BURN, but it certainly gets very warm. In addition, and probably becuase of the heat build-up, the life of the LEDs is shortened. The array of 36 lights, I'd put at an approximate equivalent of a 10W incandescent light for illumination.

If the 5mm LED cases were filed down to (for example) a hexagonal casing, and each LED lens were re-shaped to narrow the viewing angle even more, then the LEDs were packed into as tight an array as possible, I still don't think that you could make a projector from that. It would be brighter than the regular backlight, but not as much as you might think.

Not that it would be practical or anything...what about a couple led bulbs using one of those matrix fresnals...maybe a 4x4.

There are loads of threads about this, and every one has concluded it wont work.

Firstly, its pointless because LEDs have a LOWER EFFICIENCY than MH bulbs, so they would be hotter.

Secondly, an array of LEDs is not a point source, so you would have to evenly light the whole board like a backlight and you would lose an awful lot of light.

well, really i was just wondering cause i saw this

http://www.lumileds.com/products/family.cfm?familyId=15#docs

and they said that the white LED puts out 120 lumens

the light forsale on this website said it out puts 33,000 lumens

33,000/120 = 275 LED's

if you make a square matrix it would say be 17x17 LED's

assuming the LED's are 10mm thats a 170mmx170mm

still point source prob though, just wanted to know before i wasted my time, and the forums search function would not let me search for LED cause it was only 3 letters long, so i just posted.

Lumens from LEDs is somewhat misleading, because of the nature of the lights.

You don't really require a point source light. The point source is only so that you can get a collimated light throught he LCD. You can acheive much the same effect from regularly spaced LEDs. Most LEDs have a fixed viewing angle (Which is why the lumens rating of an LED can be somewhat misleading) You can re-shape the LED to have a narrower viewing angle, and you get almost the same result as you'd get after the fresnel. I'd base a design using LED as a light on the idea of not using a collimator fresnel at all.

Give it some time on LED's. It's true you wouldn't need the collimating fresnel. And LED's make more efficient use of the light as it's already directional with it's own reflector and condensor lens so I figure you could use nearly half of those LED's. They perform quite well for backlighting though and are more efficient than cold cathode fluorescents, but have a long way to catch up to compact fluorescents. One of the biggest downfalls of led's is cost. If you want to perform a three letter search, then use the the astericks as wildcards like *led* .

I'd go with no collimator AND no Field. :D

Kirk

You'd still need the field fresnel unless the objectives field of view was large enough like using a CRT projector lens with small LCD's.

No, the field lens is used to "herd" the light into the triplet. So if you had your array set up to fire the LED's light in a convergent cone through the LCD onto the projection lens, you don't need the field lens either.

I think everyone seems to be ignoring the fact that LEDs are not as efficient as MH bulbs.

Modern LED efficency is about 20-35 lumens per watt

Metal Halides are more like 70-110 lumens per watt.

This means that unless you can double the efficiency by capturing twice as much of the light from the LEDs, which is highly unlikely, the LED based projector is going to be significantly less efficient than the Metal Halide projector, which means more cooling and a dimmer image.

I guess we have to wait until LED or OLED technology improves and the efficiency is better. I wish they would hurry up, I would love an instant-on projector with a re-strike time of 0.

Though I am not promoting LED usage...

We are capturing more of the light, therefore dramatically increasing effective efficiency.

Though the Metal Halide light is producing 2.5-3X more actual light than the LED, watt for watt, well over 75% of it isn't going where we want it to, (through the LCD) and is, as such completely wasted. (I'm sure I remember calculating the percetage of the light from the bulb that actually hits the LCD, and it was less than 20%. We can probably get better than 80% of the light from an LED array through the LCD, which more than makes up for the reduction in light output. There are, of course still more technical faults with the idea, which would need to be resolved before it could be put into practice, in addition to the expense of the LEDs themselves, not to mention the lifespan of an LED array of adequate size to produce the light required.

It's technically doable... but I don't think it's practical at this point.

Not ignoring it just letting it slide to the bottom of the stack while discussing other ideas. I know LED's will never cut it at current levels (pun?) but IF they do ever reach similar power / lumen output levels then we have an idea how to set them up.
I think an even bigger problem will be the colour spectrum. White LED's are grossly difficient in some colour bands, even though they look white they can not be used for RGB filtered LCD projection. Better would be an LED array that did not have to go through any LCD - just the raw RGB elements that are then collimated (aimed) at the triplet . This alone would make them 5 times more powerful than the exisiting MH setup. I guses if you have that many led's jsut screw them all to a wall.


Kirk

Here you go, read all about it:

LED

1120 LED's
1,400 watts? Damn....



I'll bet that's cheap.

I also wanted to comment... I thought you guys were being VERY conservative on the amount of LED's it would take to make a backlight for a 15" LCD.

My guess would have been some where around 1000 LEDs. 33*33

And that's 10mm LEDs.

What? Were you going to space them out? hehe. That would result in one spotty display I'd think.


This is a completely insane idea.

I don't see how it's conservative when it's putting out more than 33K lumens every bit of it being forward. Even if only 5% of it were to pass you'd still have at least a 1.6K lumen projector. And imagine the heat, wattage, and cost. Nope! Definitely not conservative. Keep in mind most DIY projectors optics engines are very inefficient so not as many lumens from LED's would be needed. I don't think it's insane at all to consider luxeon LED's as a light source. They're currently the most efficient and cheapest out there, but are still too expensive for my taste. If I could afford to I'd throw down the money right now on a RGB array to play with on my DLP proxima dp4200. Having a third duty cycle would prolong their life even more.

I think what Agent707 was referring to was the fact that a "point source and fresnel" setup gives you an even distribution of light, but using LEDs you need to consitantly cover the whole of the LCD with light from an array of LEDs, in a simular way to a backlight, though obviously much brighter. Getting a perfectly even covering using a small number of large LEDs could be tricky.

And as far as I know (correct me if i'm wrong), LEDs are not collimated light sources, so not all of the light is going to go through the LCD anyway.

Pretty good read here

Maybe in a few more years this will be viable.

Well, even though they "claim" this thing doesn't produce much heat, I still have a problem with "20 lumens per watt".

If that's as good as LED's get, then it's not for me.

I would much rather deal with the heat of the MH than the "cost" and power consunption of this LED array.



----------

But, I guess until someone from this forum ACTUALLY builds a PJ using LED's and/or Flourescent bulbs AND posts the lack luster results, we'll continue to beat this dead horse, even though the flies are swarming around it by the droves now.

Big five" discuss growth areas for power LEDs
Breakthrough Technology Accelerates Solid-State Lighting
Seoul Semiconductor

Thought those would be interesting to some people. We currently have the technology to make an LED powered projector more powerful and efficient than metal halide with DLP/DMD's. I don't think there's anything wrong with pondering and theorizing when it will become a reality in the near future. Yes MH is currently the best option, but it's still not perfect which is why people put alternatives in their minds. LED's are still too expensive for my taste though. Really I don't see DIY LCD projection lasting much longer. Good thing there are other DIY projects like CNC.

I was looking in to doing this some time ago. An array of superbright LEDs and possible the use of the small fresnel array sheets. I spoke to a guy at EYE lighting that was working on a lumenlab projector about it and the conclusion was that they simply lack the intensity to carry all the way to the screen. However, i would love to see someone try it. Knocking off a good 8 inches off the box and loosing the ballast weight is a good thing ;-)

Even a small array of about 10-15 of them would be a halfway decent test to see if that small area that is being projected at a halfway decent level of brightness.

10 to 15 LEDs wouldn't even be as bright as the factory back lighting that comes with LCD's.... What are you thinking?

Or do you mean 10x10 to 15x15? That could be something to play with.

Though... If I had extra money like that - to play with researching and developing an LCD projector, I would just go buy a commercial and be done with it. LOL.

----

One thing I will say about hind sight. Now that I've watched all my movies 3 and 4 times, I rarely use my PJ. Nothing like the first 3 months. Now it's more like 1 movie a week or so.

Given that, a commercial PJ would last me a long time, as I don't use my for anything else. Period. In fact, the Lumen Lab pj "Isn't good for anything else."

He said that it was for a small area test. Not the whole lcd just one small portion to see if a 10x10 15x15 ... was worth trying. If that one spot could be made to work it might be worth considering expanding to hundreds of LEDs for the whole screen.

Correct just a tiny small patch of the lcd to see if that portion would be projected. I cant remember all of my math from last time looking in to this but it would have been rather expensive but worth it in the long run to use the LEDs if they did work. If i remember correctly it would have been like 250-300 LEDS with very little spacing. I have 1 super bright LED at work. Its so bright you cannot even look at it. I could not imagine 300 of those things in an aray. Hot or not if they could work the loss of the extra length and balast would be good.

this is something that is bound to work eventually if you have the time and money to put into it I think thats great! I honestly believe that this could work as I think many others do to, I am just not smart enough to figure out how.

Hmmm...

As I mentioned before, I've played around with LED lighting. I have a bag of 1,000 "11,000 mcd" white LEDs from eBay. I have various projects in mind for using them, many in relation to my car, and some for general lighting.

For a general lighting application, I have a PCB that I designed which holds 36 LEDs in a fairly dense circular array. It operates on (nominal) 12V, and seems to generate about as much illumination as a 15W halogen. The diameter of the array is just under 3" Current draw at 12V is estimated at a bit under 400mA.

In an area 12" by 9" (the size of a 15" LCD panel) I'd expect to be able to fit about 4 per linear inch, though it could be possible to squeeze in 5. That's then 25 per square inch, best case. 108 total square inches to cover is a total of 2,700 LEDs. Assume that the 11,000 mcd rating is correct, and thus we can get 275,000 mcd per square inch, or a total light output of around 29,700,000 mcd. (This works out to a brightness of 20,694,960 mcd/m^2 -- Let's round that to about 20,700 cd/m^2 (Change units to compare to stock backlit brightness, typically 250 cd/m^2) Now, of course just because that's leaving the LEDs doesn't mean that we can have that at the screen. The light still has to pass through the LCD, and we know that has an efficiency of maybe 20% tops. That gives an "actual" brightness of about 4,140 cd/m^2 The good news is that at least this is reasonably collimated, as compared to the standard backlight. Far from perfect, but it'll do. Space the LEDs from the LCD a bit to let the light scatter and blend a bit, and if you haven't found out yet, you're going to very quickly discover that aiming 2,700 LEDs so that the beam is aimed at the triplet is no small task. (It was bad enough directing 36 LEDS!) Straight ahead would be easier, but then you need a field fresnel again. (A very small price to pay, REALLY.)

Okay. Now let's SUPERSIZE that. Coming through our 15" LCD, we get 4,140 cd/m^2 Lots of brightness (compared to the standard 250 cd/m^2) Let's round that to 4,000 to factor in losses (being very generous!) getting through the triplet. To then get back to the 250cd/m^2 (Being a reasonable brightness) we'd be able to get to a 240" screen! Or at least we could if we had zero absorbtion at the projection screen. I don't see that happening.

Assume the screen is only 10% transmission.

Straight is always easier. I would go with the fresnel because the beam from each LED will spread as it aims at the triplet. If you use the field fres then the short distance should make up for some divergent loss. 10% loss from the lens will likely make up for the larger scattering loss from the led beams.

I've said this before ....but have you actually checked the colour rendering of a white LED after it has gone thru the LCD? It seem to me it is missing the basic colour output required. If using an LCD as a filter then you NEED to have the three colours accuratly represented in the source light. Close is not good enoough.

Cheers,
Kirk

This might be the wrong thread to ressurect, but as has been noted many times, searching for LED is a pain because of the 4-letter restriction. Could we maybe have a sticky on LED stuff?

My interest in a LED array light source has increased lately - as in last night. I'm putting together a simple, compact PJ with a 4" Lilliput LCD coupled to an HD-6 CRT projection lens. I'm starting to experiment now with light sources. Last night I tried my "naive approach": quickly put together a light tunnel out of aluminum flashing and used it to couple the output from a 500W Halogen worklight onto the LCD.

I was hoping that I might get off easy, since the reflector on the worklight is so close to the size of the LCD. Of course, no bones.

It was dark, very unevenly lit, and with poor color. Definitely going to need a collumating lens back there to make this thing have even a remote chance of looking acceptable.

On a whim, though, I shined the 3 LED light from my bicycle through this setup, just holding it behind the LCD about 1 cm. I think that the areas that were illuminated by the 3 LEDs were brighter and had better color than any part of the screen when I had the halogen light in place. This, coupled with the compactness of the LED array (a big deal for this little PJ) has me thinking hard about wiring about a hundred bright white LEDs up and seeing how it looks.

I know that some people have built LED arrays with varying degrees of success... But is there any reason why I shouldn't take a shot with something like this:

Cheap eBay LEDs

They're saying 30,00-40,000 mcd per LED, with a 5mm diameter. Is that even possible? It's been a while since I've looked at this stuff. Blows away the Sunbrite LEDs I used to light the lic plate of my motorcycle a few years ago (which look like the same thing SupraGuy's referring to above).


Thanks,

I think you should be in pretty good shape, led brightness is getting up there where in the right setup(like yours) it can be used, mostly because you are using a lens that will cover your entire lcd..no need for fresnels ...... it sounds doable... no idea on its brightness since I never can figure out mcandles to lumens, but i thnk its worth a shot definitey....

Yeah, I'm going for it. For occasional use on a medium-sized screen in a darkened room, I think that this will make for a nice quiet, compact setup. I'll order the LEDs this week and get to this sometime over the next month.

Next question: where do I find pointers on building an array of LEDs?

good question... did you see tuans' ( I think) pj with LEDs? search on hi member name..also I think the same pj is over at diyaudio.com forum also....

ps: you can use google to search lumenlab too , thereby avoiding the dreaded 4 letter minimum ) like so :

LED site:lumenlab.com in the search box

Yes, Tian - saw it before, found it again today. There are no details of how to construct or wire the arrays, but I'm thinking that with 100 LEDs, some perfboard, and a lot of patience, I should be able to work something out. I'll start a PLOG when I get into building.

Syscrush - you might want to take a look here:

http://www.dansdata.com/caselight.htm

Its a good primer on LEDs and arrays.

1000leds will get you something watchable
not 100

My 100 number came from an assumption that the 11,000mcd LEDs were 10mm in diameter (the ones I've used in the past were). But now I've got a source for 35,000mcd LEDs that are 5mm in diameter, so I'll be able to pack about 300 of them behind my 4" LCD.

So, that'll be about 3x the soldering for about 9.5x the light. As I'm soldering up my 300th, 400th, 500th lead, that'll be my mantra. "9.5x the light. 9.5x the light. 9.5x the light".

I should note that this PJ is meant as an experiment and "occasional use" toy kind of thing. It will be used only in a completely darkened room, for movie watching.


Take it easy,

Good luck Syscrush. I'm interested in how this works out for you. Let us know when you start the plog.

300leds
0.1w each so 30w total
20lumens/watt
so up to 600lumens hitting the lcd
maybe 60 lumens gets through
and 30 lumens on screen could be feasable

now 30lux is nicely watchable in the dark (flat white screen), so for up to 1m² image you'd probably be ok
seems quite reasonable considering the resolution limits of most 4" lcds

a high gain screen would make a lot of sense for a led projector...

Good to know that the math works out in such a way to make it look like it's got a chance. This coincides with my LED flashlight experience, and yes, a 3-4' wide screen is the goal.

I'll get a PLOG rolling in a few weeks once my LEDs arrive.


Take it easy,

PLOG is started, but it will be a while before there's real content there.

See for yourself.