AVS Forum banner

Status
Not open for further replies.
1 - 20 of 46 Posts

·
Registered
Joined
·
578 Posts
Discussion Starter · #1 ·
Just read an article in the new Wired mag and they said LED's average 40% increase in output each year! While getting 20% cheaper.


They said a white LED produces 130 lumens. A square grid of 9LEDs = 1100 lumens. This should be plenty small enough to be 'bulb sized' and wouldn't it be better than a bulb in that it'd be square -evenly hitting a square micro-display (DLP, LCoS or LCD)?


Seems like the current price might be something like $1,500 for a grid like this, but compared to $300-$400 bulbs that need lots of power, cooling fans, pour out lots of heat (VERY bad here in AZ!), can pop at any time, and only last ~2000hrs max.... A grid that lasts 10-30 years would be WELL worth the cost IMO.


I'd NEVER shut the screen off! heh


So what's the flaw or hold up now? I can't see one.


Isn't 1100 lumens enough for at least an 'early prototype' type of projector that gets this bulb replacement issue off our backs?


If not what's a good enough minimum bulb output need to be -if you think LCD can't hit it yet??


Am I screwing up 'bulb output' w/ a 'projector's rated output'?


If so, maybe someone could tell me something like... "If a PJ itself was rated @ ~1000 lumens the actual bulb output would be (or basically be) 'X' lumens"/


If LEDs aren't really for prime time it sure looks like they will be soon at the pace they're going!
 

·
Registered
Joined
·
1,265 Posts
I think that projector lumens are measured off the screen, not off the output of the bulb. A white LED can produce alot of light but you would need a far cry more then 10 LEDs to reach a screen 15ft away. An array of LEDs to make an image on a screen 15ft away would probably take up an area about 20ft X20ft. And contain thousands and thousands of LEDs. This is just a guess. This is not technically feasable and would get quite expensive in the long run. There is no free lunch. ;)


Take Care,

Marcus
 

·
Registered
Joined
·
534 Posts
Let's assume that we could get a relatively small grid of LED's that were bright enough. One interesting thing about LED's is that they can strobe incredibly fast. Most vision systems work this way.


I don't know if there's anything that can be derived from this, but it was an interesting notion that passed through my mind just now. Maybe it's just too late to be thinking of stuff like this.


Steve
 

·
Registered
Joined
·
6,092 Posts
Light sources for projectors are typically measured by a something called Luminous Efficiency. It is simply Lumens/Watt.


Typically a UHP or NSH bulb has an efficiency of about 60 or so Lumens per Watt. Do the math and a 250 Watt NSH bulb for instance pumps out 15,000 Lumens.


It would take a bunch o' LED's to create that amount of power and a lot of work to focus their light into the tiny aperture used in a DLP and LCD projectors.


Maybe in a couple of years, but not now.


-Mr. Wigggles


Ps. 60 Lumens/Watt is higher than any LED currently out.
 

·
Banned
Joined
·
9,999 Posts
Imagine a cubic shell in which the interior surface was composed of high-output LED's and the area of that interior surface not covered with an LED is covered with a highly-reflective reflecting surface (using maybe some sort of deposition process). One of the six-sides would have what could be a Fresnel lens to focus the five-sides worth of LED light output ...


Ok, not lets take the number of sides to the extreme where the efficiency of interior surface area is maximized between the function reflective surface and LED surface. I wonder ...
 

·
Registered
Joined
·
9,769 Posts
Actually this probably has some years to go for FP...


But, for RPTV this is very doable. All they have to do is stop making the reflective devices so small now:cool:
 

·
Registered
Joined
·
634 Posts
Mr. Wigggles is right about the 15K lumens, but it is misleading. About half the light from a UHP bulb can't be collected to begin with. Of course, you couldn't collect all the LED light either, but you could do a lot better than 50%. Also, because of the colorimetry of UHP bulbs, you throw away another 30% or so. I don't know what the comparable number would be for LEDs. The rest of the optical path chews up another 50% or so (except DLPs, which are worse because they're color sequential). In the end, you'd need at least 2-3x the initial lumens azryan started with.
 

·
Banned
Joined
·
9,999 Posts
Quote:
Originally posted by steve5097
Mr. Wigggles is right about the 15K lumens, but it is misleading. About half the light from a UHP bulb can't be collected to begin with.
Why? Can you please explain? Is it that the wavelengths reside in areas not very usable? If this is the case then simply pass them through a medium and change their wavelength--make those material science folks work for once!
Quote:
Of course, you couldn't collect all the LED light either, but you could do a lot better than 50%. Also, because of the colorimetry of UHP bulbs, you throw away another 30% or so. I don't know what the comparable number would be for LEDs. The rest of the optical path chews up another 50% or so (except DLPs, which are worse because they're color sequential). In the end, you'd need at least 2-3x the initial lumens azryan started with.
 

·
Registered
Joined
·
578 Posts
Discussion Starter · #9 ·
Steve you said -

"Mr. Wigggles is right about the 15K lumens, but it is misleading.-"


Seems like you're saying because the LEDs are very directional (among other factors) that you'd lose much less light therefor need much less light output than a conventional bulb?


"-In the end, you'd need at least 2-3x the initial lumens azryan started with."


Ok... Well I started by saying 9 LEDs equaling ~1100 lumens. So seems like you're guessing that you'd need ~2,200-3,300 lumens need to hit the microdisplay?


And (like Tryg mentioned) however close/far LED is from FP app, it should be much closer for RP apps. I know this isn't the forum section for this, but since the light has to travel such a short distance wouldn't LEDs already be close enough for RP?


WanMan described a cube of LEDs... maybe that would be prob. since the LEDs are so directional? But maybe a curved grid of LEDS and a focusing lens to concentrate the light 100% onto the microdisplay. 16 LED (4x4 curved grid) maybe?

~1950 lumens according to the recent Wired article.


If Steve is guessing this 'might' be fairly close to a minumum for FP, then if should be good enough for RP right?
 

·
Registered
Joined
·
150 Posts
They are replacing traffic light lamps here in 'Vegas with LED arrays. Very bright but "pixelated" compared to the bulb/glass lens. Less power consumption and will last for a zillion hours.
 

·
Registered
Joined
·
634 Posts
WanMan,

Let me try this for an explanation:

Take a flashlight and shine a beam on the ceiling. The bright spot is the "collected" light. Now, from about 45 degrees, look at the flashlight bulb. All the light you can see from any position like that is "wasted".

UHP bulbs have excess green and blue in relation to red. If you have any material science buddies that can figure out how to shift just the right amount of light, they'll have earned their keep. :)
 

·
Registered
Joined
·
111 Posts
A really efficient single-panel DLP projector is lucky to push 15%

of the visible light off the bulb to the screen.


Every lens, mirror and filter in the optical path losses lumens. And

with a RGB colorwheel, buy definition you're throwing away ~2/3-rds

of all available light (e.g when the red segment is up, you're filtering

out all other wavelengths, like green and blue). SCR is a partial

solution to the colorwheel problem.


As for LEDs, consider this. Mercury vapor lamps can produce a 10,000

lumen fireball that is 1-2mm in size, and generate that amount of light

with about the same efficiency as LEDs. Problem is, LEDs are far far

away from generating 10,000 lumens in a fireball of only 1-2mm.


--xb31
 

·
Registered
Joined
·
320 Posts
What if you weld each LED to fiber optic and than bunch them to one tight bundle and shine on DLP or LCD, fiber will take all the light from LED with out waste. And you can use big array of LED's and focus them this way to small area. Or use tri color LED’s and sequence them instead of color wheel.

Artur.
 

·
Registered
Joined
·
6,092 Posts
Quote:
Originally posted by steve5097
Mr. Wigggles is right about the 15K lumens, but it is misleading. About half the light from a UHP bulb can't be collected to begin with. Of course, you couldn't collect all the LED light either, but you could do a lot better than 50%. Also, because of the colorimetry of UHP bulbs, you throw away another 30% or so. I don't know what the comparable number would be for LEDs. The rest of the optical path chews up another 50% or so (except DLPs, which are worse because they're color sequential). In the end, you'd need at least 2-3x the initial lumens azryan started with.
Steve,


About a half of the lumens of a standard projector's bulb are fucsed through a pin hole aperature less than 5mm square.


This simply isn't possible with LED's without some seriously elaborate optics. And what's the benefit? the use of a source that is less efficient than arc lamps to begin with? It doesn't make sense to me.


The only benefit I see with LED's is that the response time of the LED's is fast enough that the light source for a one-chip DLP could be a bank of RGB LED's with each color displaying only 1/3rd of the time, but at a very high intensity due to the reduced duty cycle. This would eliminate the need for a color wheel in a DLP projector and would increase the Luminous Efficiancy of the LED's by a factor of three versus the use of white LED's. The problem is it still isn't enough output compared to the Arc lamps. A well designed LED driven one-chip projector might make 100 Lumens (and that is a ROUGH but optimstic guess) and most people would feel that isn't enough.


Sorry to burst anyone's bubble, but LED's aren't going to be used for quite a while longer. And I'm sure once they get bright enough, you'll see them in use in the 1-chip DLPs.


-Mr. Wigggles
 

·
Registered
Joined
·
6,092 Posts
Quote:
Originally posted by kevin gilmore
The white led's are extremely defficient in the red.


This is the most efficient of the white led's, the nichia...


If you put enough of these together to get the light output you

need, you will still have roughly the same amount of heat generated

as a 200 watt UHP bulb...
Read my above remark about the use of R,G, and B LED's in 1 chip projectors. You would get a 3X increase in light efficiency versus the use of white LED's and a color wheel.


-Mr. Wigggles
 

·
Registered
Joined
·
273 Posts
Still, it's even with a 100 lumens of output, it's surprising that we haven't heard reports of a prototype of this kind...


TI! are you listening! Show us that the sales guys aren't running your company and demo something new and interesting!


-Dave
 

·
Registered
Joined
·
918 Posts
OK WanMan (with another try at the links):


Light bulbs at 15-20 lm/W, fluorescents at 60 lm/W and they hope to do 300 lm/W (eventually) with PRS-Leds.


New PRS LCD info:
PRS Leds


Another interesting one:
InGaN Nanowire LED
 

·
Registered
Joined
·
6,092 Posts
Mark,


Thanks for the link.


here is the important quote:


"Tungsten filament and halogen incandescent light sources

have a luminous efficiency of 15-25 lm/W. Fluorescent tubes and

the newer compact fluorescent light bulbs have a luminous performance

of about 60 lm/W. In comparison, the calculated theoretical

efficiency of the dichromatic PRS-LED is around

330 lm/W, which is much higher than that of conventional light

sources. Even if only 50 % of the PRS-LED’s efficiency, i.e.

about 160 lm/W, can be demonstrated within the next 5 years,

this LED would significantly outperform conventional light

sources. At the present time, the PRS-LED has a luminous performance

of less than 10 lm/W."


They are at 10 and they think 160+ is around the corner? We shall see.


-Mr. Wigggles
 

·
Registered
Joined
·
918 Posts
The PRS-LED paper was from June of 2000 so I'm hoping there has been progress since then. The Nanowire paper is newer and describes some of the problems with PRS-LEDs and how Nanowires may help out...



The following isn't about LEDs but it is interesting they rate the various lights for lm/W, CRI and lifetime.

Worth checking out European Greenlight Programme which gives some typical lamp info.


However, the sulphur microwave is suppose to be at 100 lm/W with a long lifetime. (Hopefully wouldn't take out an 802.11b network). They also give 100 lm/W for a T8 triphosphor with electronic ballist. So the number in my previous two links don't match these.
 
1 - 20 of 46 Posts
Status
Not open for further replies.
Top