series connection
Excellant point. My mistake. I suppose that seeing some LED array designs with one resistor per LED just made me overlook the obvious. There is no reason you could not drive a chain of LEDs in series with a single resistor or current regulated power supply. (Assuming you are not using a common cathode or common anode array.) So the extra power dissipated by the current regulation could be neglible.
But still, even the Lamina BL-32D0-0133 only puts out 567 lumens from a 26 Watt input. That is 21.8 lumens per Watt. Compare that to 82.5 lumens per Watt with a MH bulb. So you would have to supply about 4 times as much power to the LED array to get the same amount of light, and all that extra power would mean a lot more heat.
There are much better LEDs, in terms of efficiency, but they are not high-powered units like Luxeon or Lamina.
Excellant point. My mistake. I suppose that seeing some LED array designs with one resistor per LED just made me overlook the obvious. There is no reason you could not drive a chain of LEDs in series with a single resistor or current regulated power supply. (Assuming you are not using a common cathode or common anode array.) So the extra power dissipated by the current regulation could be neglible.
But still, even the Lamina BL-32D0-0133 only puts out 567 lumens from a 26 Watt input. That is 21.8 lumens per Watt. Compare that to 82.5 lumens per Watt with a MH bulb. So you would have to supply about 4 times as much power to the LED array to get the same amount of light, and all that extra power would mean a lot more heat.
There are much better LEDs, in terms of efficiency, but they are not high-powered units like Luxeon or Lamina.
Guys im going to butt in here, i think the majority of you are missing the eficientcy thing here and how manny lumens are needed for what.
A 250w mh bulb is more eficient then a led, correct
It will take x amount of leds to make up (btw a 250w mh bulb only yealds 20000lm) 20000lm, correct.
Do we need 20000lm, NO
Why are we using 20000lm bulbs today? simple, the eficientcy of our optical system from the very begining of our source is poor. Spherical reflectors reflect a maximum of 30% of light. A condenser will only condense 55 - 65% max of the light from our source with the spherical reflector. Thats ineficent and thats on an eficient bulb!!
What do most do? go to 400w. Now is that eficent? especially when you can get the same results with the right format bulb on the right light engine as the 400w with only 250w? To me thats nuts.
So lets take a look at somthing here. Lets say we are utilisising at best 60% of light from our source (and im telling you, you dont get this unless you condense the actual arc which in most cases none of you are doing and cant do due to the bulbs format we use). 60% of 20000lm is roughly 12000lm.
Lets look at leds.
Led's maybe less eficient as in power to heat and light ratio but they are actually more eficent in the light engine department. They have their own reflector optimised and their own condenser onboard and therfore right back to the source we get the most light from the led.
Depending on the led's beam angle, you can condense this light even further. You can condense the light into a guide via a lens built into the guides end. The guide also has a higher eficentcy as we have no or virtually no inverse law and virtually no reflection or other losses. It also guides all light through the guide in the direction where we want it to go and no light or virtually any light escapes from it, (has to be well polished, the inside becomes a mirror just like a prisim). With the led on the end of a guide thats condensed it becomes extremly blinding bright, (about 2x of the led being on its own). The more you condense the lights source the more intense the light becomes (to a given point, in my case with the mh, the arc), this happend only with low F# optics and with positive focals.
I have done this btw, and spent alot of time on the design of the guide through trial and error, this is how i know it works. I also tested this design on my system and soon saw the benifits of the light guide eficientcy. I too used to be negative with leds not thinking they would work but they purley do, there is no way you need the amount of light some of you guys think, both with leds and mh.
I have also condensed the arc of a cdm-t 150w mh and a 250wHQI and im telling you guys to work on it because you guys are way way missing the conception of things here and how to get them to work in an eficient manner. Look at the pro designs for an idea.
A 250w with the arc condensed under my testing is TOO bright, it hurts your eyes to watch. Imagine sitting 1inch away from your tv screen, thats what it does to your eyes with a 250w's arc condensed. Do it to a 400w? forget it, nothing will cool the lcd or your eyes. (FYI, all screen testing is done at 100inches and on the 7inch Lilliput).
So how much light do we realy need? With mh on an elliptical or a condensed arc system via a spherical no more then 12000lm (150w mh). With leds even less, infact way less, under half, but they arent cheap to buy. Thats the downfall and why im sticking to mh, for now.
Do leds work properly with the guides and a rear frensel? i mean come on guys you can surley answer that one ya selfs lol. Dont use a rear frensel only use the top, what goes into the top frensel as paralelle light comes back out as a cone on the topside. The light coming from the guide is paralelle, you have to get the condensing system right to get it paralelle. To answer another query, the image edges are sharp as can be and the lcd is evenly lit. Also there is not just 1 guide, its 1 guide per led.
Anyway from hands on experience thats my 2cents
Trev
A 250w mh bulb is more eficient then a led, correct
It will take x amount of leds to make up (btw a 250w mh bulb only yealds 20000lm) 20000lm, correct.
Do we need 20000lm, NO
Why are we using 20000lm bulbs today? simple, the eficientcy of our optical system from the very begining of our source is poor. Spherical reflectors reflect a maximum of 30% of light. A condenser will only condense 55 - 65% max of the light from our source with the spherical reflector. Thats ineficent and thats on an eficient bulb!!
What do most do? go to 400w. Now is that eficent? especially when you can get the same results with the right format bulb on the right light engine as the 400w with only 250w? To me thats nuts.
So lets take a look at somthing here. Lets say we are utilisising at best 60% of light from our source (and im telling you, you dont get this unless you condense the actual arc which in most cases none of you are doing and cant do due to the bulbs format we use). 60% of 20000lm is roughly 12000lm.
Lets look at leds.
Led's maybe less eficient as in power to heat and light ratio but they are actually more eficent in the light engine department. They have their own reflector optimised and their own condenser onboard and therfore right back to the source we get the most light from the led.
Depending on the led's beam angle, you can condense this light even further. You can condense the light into a guide via a lens built into the guides end. The guide also has a higher eficentcy as we have no or virtually no inverse law and virtually no reflection or other losses. It also guides all light through the guide in the direction where we want it to go and no light or virtually any light escapes from it, (has to be well polished, the inside becomes a mirror just like a prisim). With the led on the end of a guide thats condensed it becomes extremly blinding bright, (about 2x of the led being on its own). The more you condense the lights source the more intense the light becomes (to a given point, in my case with the mh, the arc), this happend only with low F# optics and with positive focals.
I have done this btw, and spent alot of time on the design of the guide through trial and error, this is how i know it works. I also tested this design on my system and soon saw the benifits of the light guide eficientcy. I too used to be negative with leds not thinking they would work but they purley do, there is no way you need the amount of light some of you guys think, both with leds and mh.
I have also condensed the arc of a cdm-t 150w mh and a 250wHQI and im telling you guys to work on it because you guys are way way missing the conception of things here and how to get them to work in an eficient manner. Look at the pro designs for an idea.
A 250w with the arc condensed under my testing is TOO bright, it hurts your eyes to watch. Imagine sitting 1inch away from your tv screen, thats what it does to your eyes with a 250w's arc condensed. Do it to a 400w? forget it, nothing will cool the lcd or your eyes. (FYI, all screen testing is done at 100inches and on the 7inch Lilliput).
So how much light do we realy need? With mh on an elliptical or a condensed arc system via a spherical no more then 12000lm (150w mh). With leds even less, infact way less, under half, but they arent cheap to buy. Thats the downfall and why im sticking to mh, for now.
Do leds work properly with the guides and a rear frensel? i mean come on guys you can surley answer that one ya selfs lol. Dont use a rear frensel only use the top, what goes into the top frensel as paralelle light comes back out as a cone on the topside. The light coming from the guide is paralelle, you have to get the condensing system right to get it paralelle. To answer another query, the image edges are sharp as can be and the lcd is evenly lit. Also there is not just 1 guide, its 1 guide per led.
Anyway from hands on experience thats my 2cents
Trev
Ace, you make some good points about efficiency. That definetely has to be considered. If a properly designed light engine can achieve 60% then a cdm-t 150w mh would be enough for a 7 inch projector. What do you think it would take in wattage for a 17 inch projector? If the light engine is 60% efficient, do you think 400w mh is too big?
Ace already uses the 150watt in his 7inch projector mikey p He claims he has it brighter than a 250watt wouldn't surprise me one bit either with all the time he puts into testing, fiddling and blowing things up lol . If that is the case a 250 watt would be plenty for a 17inch projector one would think?!?
He claims he has it brighter than a 250watt wouldn't surprise me one bit either with all the time he puts into testing, fiddling and blowing things up lol . If that is the case a 250 watt would be plenty for a 17inch projector one would think?!?
image size is what counts
You don't need a particular minimum number of lumens per square inch of LCD. What matters is the number of lumens per square inch of screen image. So a 17" LCD would work just fine with a 150 Watt MH lamp, as long as you use the right lenses and throw distance so you don't try to light the whole side of your office building with it. (And of course, you need fine attention to the reflector and condensor system to get most of that light into the projection lens.) A 100" image made with a 15" LCD will be just as bright as a 100" image made with a 17" LCD, with the same lamp.
You might want to change to a longer focal length condensor fresnel (ie. 330 mm) when you go to a 17" LCD, just so the corners are more evenly light.
There may be an upper limit on how many lumens you can put through a smaller LCD per square inch before you melt it. But you will run into that limit with very small LCDs (ie. 2.5"), not LCD monitor panels.
You don't need a particular minimum number of lumens per square inch of LCD. What matters is the number of lumens per square inch of screen image. So a 17" LCD would work just fine with a 150 Watt MH lamp, as long as you use the right lenses and throw distance so you don't try to light the whole side of your office building with it. (And of course, you need fine attention to the reflector and condensor system to get most of that light into the projection lens.) A 100" image made with a 15" LCD will be just as bright as a 100" image made with a 17" LCD, with the same lamp.
You might want to change to a longer focal length condensor fresnel (ie. 330 mm) when you go to a 17" LCD, just so the corners are more evenly light.
There may be an upper limit on how many lumens you can put through a smaller LCD per square inch before you melt it. But you will run into that limit with very small LCDs (ie. 2.5"), not LCD monitor panels.
Thoughts on LEDs
First off here's a handy link I found :
http://members.misty.com/don/led.html
Secondly, I'd like to address the heat issue about LEDs. To me the issue with heat output from the light source has always been twofold :
1) heat that is internal to the light source and has to be sunk by direct conduction/convection/whatever.
2) the heat (infrared) that is shot out of the light source (radiation) at whatever optics equipment you've got in the way.
"Traditional" light sources (i.e. things we heat up so hot that they start emitting light, etc) are going to throw off significant amounts of infrared emission by their nature. This causes problems.
The arc in Metal Halides is so hot (1000's of degrees K) that it is also kicking off UV as well (mentioned before in this thread), increasing the problems.
So to me, from a thermal point of view, the main advantage that LED's bring to the table is that they are not blackbody style emitting lots of infrared/UV/junk that has to be dealt with before it hits your optics/panels/whatever. The heatsinking problem for LEDs can be dealt with entirely by direct conduction/convection away from the LEDs themselves. In other words you probably can directly couple them to a good heat spreader/ heat pipe style heat sink attachment and be done with it.
As to whether or not LEDs are there yet, they still have a ways to go, especially in the pricing area, to make them a no brainer. Right now, the fact that a good LED can last 50,000 hours+ is attractive in that it eliminates bulb replacements, but the cost to build an LED array to output the same power as a metal halide is like buying all of your future replacement bulbs up front. You would, I think, end up with a more compact (need more room for the heat sink but dont need a fan or other optics filters, can move panel closer to the lights, etc.) and definitely quieter device that should run more or less maintenance free for its lifetime. In the spirit of DIY and if you have the cash to burn it is worth doing to see exactly what the currently achievable result is.
First off here's a handy link I found :
http://members.misty.com/don/led.html
Secondly, I'd like to address the heat issue about LEDs. To me the issue with heat output from the light source has always been twofold :
1) heat that is internal to the light source and has to be sunk by direct conduction/convection/whatever.
2) the heat (infrared) that is shot out of the light source (radiation) at whatever optics equipment you've got in the way.
"Traditional" light sources (i.e. things we heat up so hot that they start emitting light, etc) are going to throw off significant amounts of infrared emission by their nature. This causes problems.
The arc in Metal Halides is so hot (1000's of degrees K) that it is also kicking off UV as well (mentioned before in this thread), increasing the problems.
So to me, from a thermal point of view, the main advantage that LED's bring to the table is that they are not blackbody style emitting lots of infrared/UV/junk that has to be dealt with before it hits your optics/panels/whatever. The heatsinking problem for LEDs can be dealt with entirely by direct conduction/convection away from the LEDs themselves. In other words you probably can directly couple them to a good heat spreader/ heat pipe style heat sink attachment and be done with it.
As to whether or not LEDs are there yet, they still have a ways to go, especially in the pricing area, to make them a no brainer. Right now, the fact that a good LED can last 50,000 hours+ is attractive in that it eliminates bulb replacements, but the cost to build an LED array to output the same power as a metal halide is like buying all of your future replacement bulbs up front. You would, I think, end up with a more compact (need more room for the heat sink but dont need a fan or other optics filters, can move panel closer to the lights, etc.) and definitely quieter device that should run more or less maintenance free for its lifetime. In the spirit of DIY and if you have the cash to burn it is worth doing to see exactly what the currently achievable result is.
Anybody want to take a stab at the design of the light engines used in the mini-projectors seen at CES this year? I think the only way these mini-PJs are even possible is the use of DLP in lieu of LCD, but I'd still be interested to hear anyone's thoughts on the specifics of these LED based light engines.
new LED projectors
From what I read, I think the key is that they don't claim to make very large images. Since the brightness per square centimeter drops by the square of the image size, a 27" diagonal image is 13.7 times as bright as a 100" diagonal image using the same light source.
Nobody ever claimed you couldn't make an LED-powered projector. They just have a way to go to be competitive with MH lamps for a real big screen application. But that distance is mostly economic: The very most efficient LEDs only need to improve by about 40% to be better than MH lamps, in terms of lumens/Watt. Then their prices have to drop by about 100 times!
From what I read, I think the key is that they don't claim to make very large images. Since the brightness per square centimeter drops by the square of the image size, a 27" diagonal image is 13.7 times as bright as a 100" diagonal image using the same light source.
Nobody ever claimed you couldn't make an LED-powered projector. They just have a way to go to be competitive with MH lamps for a real big screen application. But that distance is mostly economic: The very most efficient LEDs only need to improve by about 40% to be better than MH lamps, in terms of lumens/Watt. Then their prices have to drop by about 100 times!
I beleive they use a technology called Grating Light Valves (GLV). More can be read here about it http://electronics.howstuffworks.com/projection-tv6.htm . Here is what it says:
"The GLV chip consists of tiny reflective ribbons mounted over a silicon chip. The ribbons are suspended over the chip with a small airgap in between. When a voltage is applied to the chip below a ribbon, the ribbon moves toward the chip by a fraction of the wavelength of the illuminating light. The deformed ribbons form a diffraction grating, and the various orders of light can be combined to form the pixel of an image. The shape of the ribbons, and therefore the image information, can be changed in as little as 20 billionths of a second.
To make a projector, the GLV pixels are arranged in a vertical line that is 1,080 pixels long. Light from three lasers, one red, one green and one blue, shines on the GLV and is rapidly scanned across the display screen at 60 frames per second to form the image.
A major advantage of GLV technology is that GLV chips can make high-resolution images at a relatively low cost. For example, because a 1,920x1,080 pixel image can be achieved by scanning a 1,080-pixel linear array, a GLV chip can be manufactured to achieve this resolution with only 1,080 pixels, instead of the 2 million needed for other technology, such as DMD. Also, because the ribbons are aligned vertically, there are no horizontal gaps in the image -- there is a very high fill space on the chip. "
The web site has some pictures to help you understand. But for this case, I believe the LED projectors showcased used LED lasers with GLV technology to produce an image. Sorry to burst your bubble, but LEDs are still "out of the picture" (no pun inteded) as far as using them for LCD projection.
"The GLV chip consists of tiny reflective ribbons mounted over a silicon chip. The ribbons are suspended over the chip with a small airgap in between. When a voltage is applied to the chip below a ribbon, the ribbon moves toward the chip by a fraction of the wavelength of the illuminating light. The deformed ribbons form a diffraction grating, and the various orders of light can be combined to form the pixel of an image. The shape of the ribbons, and therefore the image information, can be changed in as little as 20 billionths of a second.
To make a projector, the GLV pixels are arranged in a vertical line that is 1,080 pixels long. Light from three lasers, one red, one green and one blue, shines on the GLV and is rapidly scanned across the display screen at 60 frames per second to form the image.
A major advantage of GLV technology is that GLV chips can make high-resolution images at a relatively low cost. For example, because a 1,920x1,080 pixel image can be achieved by scanning a 1,080-pixel linear array, a GLV chip can be manufactured to achieve this resolution with only 1,080 pixels, instead of the 2 million needed for other technology, such as DMD. Also, because the ribbons are aligned vertically, there are no horizontal gaps in the image -- there is a very high fill space on the chip. "
The web site has some pictures to help you understand. But for this case, I believe the LED projectors showcased used LED lasers with GLV technology to produce an image. Sorry to burst your bubble, but LEDs are still "out of the picture" (no pun inteded) as far as using them for LCD projection.
After a couple minutes of thought, I have to say that LEDs are not totally out of the picture after all. If some smart person was able to actually come up with a system to aim and move a powerful LED laser across an LCD display at a very high speed (similar to the way a steam of electrons are moved inside a CRT display) using fresnel lenses to direct the beam into the LCD, we could indeed use LEDs for projection. This process would not be easy though...It would require special optics, and a rotation system of some sort. Oh well, just a thought.
LED + DLP out now
Saw a production model of a palm top projector from Mitsubishi at the TI booth at the 2005 CES. Sorry I don't remember the model number. The guy at the booth wasn't to sure about the power, but he said that the DLP chip was illuminated with a Luxon Star LED (1 watt, 3 watt, 5 watt?). Not too bright, but getting better and only needing a 12V wall wart to run off of. Almost no heat either. BenQ was also showing a preproduction model of a projector that was as small as the Mitsu and using an LED as well.
Greg
Saw a production model of a palm top projector from Mitsubishi at the TI booth at the 2005 CES. Sorry I don't remember the model number. The guy at the booth wasn't to sure about the power, but he said that the DLP chip was illuminated with a Luxon Star LED (1 watt, 3 watt, 5 watt?). Not too bright, but getting better and only needing a 12V wall wart to run off of. Almost no heat either. BenQ was also showing a preproduction model of a projector that was as small as the Mitsu and using an LED as well.
Greg
I'm more into audio, and actually never thought about building a projector, but I was curious and took a look at this forum.... anyway, you might think about buying the cheap LEDs off of Ebay (100 "13000mcd" white LEDs for under $20-30), just to even test an idea. Not sure if they'd be any good for this application, but just thought I'd mention it.
cheap white LEDs
It's been mentioned before, but I will repeat it for latecomers: Most of the cheap high-power white LEDs are really blue LEDs with a yellow phosphor. We see them as white, but they don't have the red and green bands we need to match LCD color filters.
To make any sort of a reasonable image, you need LEDs that have red, green, and blue light bands. Some of the special very high power LED modules do have enough light in each of these bands, but they have a very high cost per lumen.
It's been mentioned before, but I will repeat it for latecomers: Most of the cheap high-power white LEDs are really blue LEDs with a yellow phosphor. We see them as white, but they don't have the red and green bands we need to match LCD color filters.
To make any sort of a reasonable image, you need LEDs that have red, green, and blue light bands. Some of the special very high power LED modules do have enough light in each of these bands, but they have a very high cost per lumen.
In looking at this thread it's obvious that this could be done but would be inordinately expensive at this time. Also, although it may not be a law of physics it's a rule of life that as soon as you invest in a long term system like this something will shortly come along and blow it away at a fraction of the cost.
Here's my simple-minded idea.
Here's my simple-minded idea.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.