This is topic white l.e.d.s as a light source in forum Film Handlers' Forum at Film-Tech Forum ARCHIVE.


To visit this topic, use this URL:
https://ft-forum.com/ft/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=004682

---

Posted by Matt Hale (Member # 1518) on 03-19-2003, 10:07 PM:
 
Here in Vancouver, they are in the process of replacing all our crosswalk signals with new ones made out of light emitting diodes. Now the dont walk signal is made out of the conventional amber ones we all know, but the little walking guy is made up of 25 or 30 white diodes - and they're bright. I mean really bright. You can see it 3 or 4 blocks away in broad daylight.

This has got me thinking. If 30 of these things are that bright, how bright would an array of several hundred, or even thousand be? In comparison to say, a small xenon bulb? The colour temperature certainly seems about right for film projection. I imagine you would need some type of fresnel lens in front of the array to smooth out all the point sources and make the light even, but is there any reason this couldnt work?

Using diodes for projection would bring the same benefits as in other areas they are appearing:

low-voltage
vitually no heat
highly energy efficient
not explosive
100,000+ hours rated life

Am I completely off my rocker? Is anybody out there working on this kind of thing?
 

---

Posted by Manny Knowles (Member # 1171) on 03-19-2003, 10:20 PM:
 
Assuming that your assumption about color temperature is correct, I would think that the problem would be "collecting" and then focusing the light from such an array. The light from a Xenon is concentrated around a point and, as you can see, it lends itself to collection and focusing via a dish-shaped mirror.

I would imagine that the size of your proposed contraption would soon rage out of control.

Lamp replacement would also become a source of irritation. How many are we changing out at one time? All 900,000 of them, or just some? In that case, which one(s) is/are blown?
 

---

Posted by Matt Hale (Member # 1518) on 03-19-2003, 11:20 PM:
 
I had envisioned something more on the order of one or two thousand diodes, mounted into the parabolic reflector itself. Exactly how bright this would be with the current technology I dont know, but these new white ones seem a good deal brighter than the red, green & amber ones that have been around for years.

As to bulb replacement, the idea is that you dont need to very often at all. With 2000 light sources, one burnout is only going to diminish your total light ouput by 0.05%. If the light from each individual diode is sufficiently "mixed" with the light from all the others then a burnout is not going to produce a noticable dark spot in the image, just an imperceptable dimming. You could probably run it with quite a few burn-outs, until the image got too dim to be useable, then just replace the whole thing.

Besides, with a rated life of 100,000 hours, thats almost 11.5 years of continuous 24/7/365 operation.
 

---

Posted by Josh Jones (Member # 447) on 03-19-2003, 11:29 PM:
 
Several problems will arise.

1. With any electronic device(noting is 100% efficient) heat will become an issue.

2. The color temperature is wrong, at least in the signs they use around here. it is shifted too far to blue.

3. Cost. Have you seen how much those little bastards are? Close to a dollar a piece if I recall. then there is the whole ordeal of focusing that light. at this point I dont think it will be practical. But who am I to say? I have a pair of carbon arc lamps in my basement

Josh
 

---

Posted by Bill Enos (Member # 440) on 03-19-2003, 11:38 PM:
 
In a conversation with a county traffic engineer I was told that every point of light you see in these new traffic lights contains five to nine super bright l.e.d.s depending on the color and some other considerations. Power supply is another consideration as leds will draw all they can get and commit suicide if current is not limited.
 

---

Posted by Matt Hale (Member # 1518) on 03-20-2003, 07:46 PM:
 
Well I've done a little more research, and it turns out white leds come in different colour temps. As I recall, 6500k is about the temperature of tungsten light, which is too yellow. I would imagine you could produce the required colour temp by mixing some that are too yellow with some that are too blue. The exact ratio would have to be calculated. Does anybody know what the colour temperature of a xenon bulb is?

The power supply would definitely have to be custom. We're talking about a totally different circuit here, operating on a totally different theory. According to the spec sheet of the 6500k diode, it operates at 3.6v, 20ma. At that spec, 2000 leds would pull 40 amps of current, or about 144 watts of power total. What I still dont know is whether that would be bright enough. 2000 diodes would produce roughly 12,800 candellas. Does anyboady know how many candellas a 1kw xenon puts out?

Cost would be an issue, but I look at it this way:
If a 1kw bulb is available for $386.07 and runs for 2400 hours, it will cost $16,214.94 to get 100,000 hours of xenon light (not counting taxes & shipping.)
If a white led is available for $1.80 (in quantities of 500) then 2000 leds will cost $3,600.00 which leaves plenty to cover the power supply & optics before you even reach half the cost of the xenon.

Am I crazy? (dont answer that.)
 

---

Posted by John Anastasio (Member # 607) on 03-20-2003, 08:44 PM:
 
Incandescent studio light sources are balanced to 3200K, which is an ORANGE color. 6500K is the color balance of daylight at high noon. An much more intriguing idea has already been invented utilizing three cavity lasers operating at red, green and blue wavelengths. They produce enormous amounts of light with relatively low heat. The problem is still the cost, but such a device could be the heart of an extremely efficient laser projector as well.
 

---

Posted by John Pytlak (Member # 331) on 03-22-2003, 06:21 AM:
 
AFAIK, even "white" LED's do not provide a continuous spectrum of light output. Xenon and carbon arc provide light that closely approximates natural daylight, with a relatively continuous spectrum with no "gaps" that would cause poor color rendering.
 

---

Posted by John Anastasio (Member # 607) on 03-22-2003, 06:40 AM:
 
Good point, John. They didn't arrive on the scene until the invention of the blue LED, so they're probably a combination of blue and yellow (which is a combination itself of red and green LED's). They more than likely have extremely high peaks at those wavelengths. I'm going to run one through a spectrograph on Monday just out of curiosity.
 

---

Posted by Steve Kraus (Member # 476) on 03-22-2003, 10:54 AM:
 
I doubt even 500 would provide the number of lumens required but as to the color issue I think that's a red (hehehe) herring.

Color imaging, whether electronic or photochemical, is based on the fact that the color receptors in the eye (retinal "cones") work on a 3 color basis and we can fool it into seeing colors that are not there. We represent, for example, yellow or cyan as being combinations of green + red or green + blue and we present those colors to the eye and yellow or cyan are perceived by our eyes.

But yellow is NOT a combination of red and green and cyan is NOT a combination of green and blue; they are actual colors--you can describe specific wavelengths of light that are yellow or cyan.

Yet the way our vision works it is not necessary to deliver any actual yellow or cyan or whatever to the eye to have the eye think those colors are there.

So with that in mind I'm not sure that it's necessary to have a continuous spectrum in a projection lamp. I doubt very much that the light emitted by the 3 color phosphors on the CRT I am looking at offer continuous spectra when the 3 combine to form white. A continuous spectra is necessary on photography but not on display.

Without a continuous spectra one might find that, depending on the wavelengths and the density curves of the film, a difference depending on the dyes, that might not be there with a continuous spectrum, like maybe the difference from color positive to IB Technicolor could be more pronounced with this sort of discontinuous illumination than with the continuous light of an arc source but that aside as long as the source is producing the three colors and these colors are properly modulated by the dye image I would expect it to work.

BTW, is a white LED a single LED giving a range of colors or is it really just a closely spaced combination of individual LEDs?
 

---

Posted by John Anastasio (Member # 607) on 03-22-2003, 12:45 PM:
 
Thanks for clearing that up, Steve. Yes....yellow is yellow. A wavelength of around 575 nanometers to be more precise, but if you look at the spectrum of a "white" flourescent lamp, you'll find that it has two very distinct peaks at around 435nm and 550nm....blue and yellow. If you used that as a light source, you'd get some odd color reproduction on the screen. No, you don't need a continuous spectrum of white light, but you do need the right combination of red, blue and green fequencies to get accurate color with an additive system...and I'd be surprised if a so-called "white" LED had them. It's a moot point...the whole idea is ridiculous.
 

---

Posted by John Hawkinson (Member # 1135) on 03-22-2003, 03:16 PM:
 
If you're going to have hundreds or thousands of LEDs, you can certainly vary the color spectrum by having different quantities of different colors. There's no requirement that you use only "white" LEDs.

--jhawk
 

---

Posted by John Pytlak (Member # 331) on 03-22-2003, 05:28 PM:
 
A discontinous light source (e.g., fluorescent lighting) usually has poorer color rendering compared to one having a continous spectrum. An extreme example is to look at the color of automobiles under sodium vapor or mercury vapor street lights -- some colors look fairly normal, but others are very desaturated.
 

---

Posted by Gordon McLeod (Member # 33) on 03-22-2003, 11:03 PM:
 
Discharge lamps have a CRI (colour rendering index) number that is based on the fact they are not a contious spectrum compared to colour temp which degree kelvin relates to a black body
 

---

Posted by Steve Kraus (Member # 476) on 03-23-2003, 12:33 AM:
 
Well, lets face it: "Color temperature" has always been a crock anyway. A useful shorthand only when the source resembles a black body radiator. Therefore it's only meaningful for incandescent lamps which are becoming less and less significant as light sources.
 

---

Posted by Randy Stankey (Member # 64) on 03-23-2003, 08:28 AM:
 
How about tri-color LEDs?

They are coming out with all sorts of neat-o devices that have three (sets of) colored elements inside them. By varying the intensity of the three colors, any color can be made.

I can't imagine that LEDs would be good enough for projection on anything but the smallest screen... If any. But I can imagine that LEDs would be a nice light source for small viewers. (Movieola style)

Combine a small-screen viewer with the variable color LEDs and I imagine you would have a nice way for directors, editors and film technicians to judge the quality of scenes shot on film under different conditions of light. (Color temperature?)

Question: If the words "Color Temperature" only apply to light generated from certain types of devices then wouldn't the words "Color Quality" (or something to that effect) be a better term for light that is not emitted from black body ratiators? (Just thinking...)
 

---

Posted by Steve Kraus (Member # 476) on 03-23-2003, 10:42 AM:
 
Soitenly. But it's not going to be a single number when you've got some sort of strange curve.
 

---

Posted by John Pytlak (Member # 331) on 03-23-2003, 12:16 PM:
 
"Color Rendering Index" is usually used to quantify the ability of a discontinuous spectrum light source to render colors accurately.
 

---

Posted by Bobby Henderson (Member # 840) on 03-23-2003, 12:40 PM:
 
LEDs are costly and that alone would make them impractical for use as a light source where many thousands would be needed.

We deal with LEDs a lot these days in indoor and outdoor electronic signs. They've all but replaced wedge based "light bulb" style lamps and the flipper type message centers are all gone. Lately, banks and schools with time/temperature signs or message centers with flipper based discs are needing them replaced with LED units when new maintenance issues arise.

I would really love to see prices on the things take a dramatic drop. Then such items like full color outdoor video boards would lower into more practical price ranges. Right now customers must pay hundreds of thousands of dollars (or even millions of dollars) for large outdoor LED video displays.

LEDs are costly, but when that cost is amortized out over the total maintenance cost of standard lighting the LEDs often win. This is why many new traffic signals and crosswalk lights are converting over to LED. LEDs last longer, with some rated for 10 years of life. But not really all that many LEDs are used for things like an LED cluster in a traffic light.

Oh, and I do agree some of the newer LEDs are very bright.
 

---

Posted by Matt Hale (Member # 1518) on 03-23-2003, 01:02 PM:
 
Well I have done a little more research and, unless I am not understanding something, it looks like it would only take 469 leds to equal the light output of a 1kw xenon. I have lost the links to the spec sheets, but as an exmaple, the Osram XBO1000 bulb has a rated light output of 3000 cd. The leds I am looking at are rated 6.4 cd apiece. That is significantly less leds than I thought it would take.

Building the array out of seperate red green & blue diodes would definitely give you better control of the colour spectrum, and probably reduce the cost a bit. Where it would complictae matters is in ensuring that the light is sufficiently mixed before it hits the film that you do not have coloured blotches in the image. I have been trying to figure out what sort of lens(es) would be needed for this, and to convert the multiple beams from the many sources into one parallel beam, but I do not know enough about optics (like that wasn't obvious already.)
 

---

Posted by Steve Kraus (Member # 476) on 03-23-2003, 03:59 PM:
 
While LEDs have made great inroads into signals and indicators I haven't seen them being used yet for general lighting although that may very well be next. I'm just mentioning that because I think you'd need to reach that point, where they are practical to produce a mass amount of light, and then be able to go beyond that and produce it in a way that is efficiently collectable.

How boring an LED lamphouse would be!
 

---

Posted by Gordon McLeod (Member # 33) on 03-23-2003, 04:00 PM:
 
lets see that would be one awfully big "point source" so i doubt you could collect it very efficantly
 

---

Posted by Matt Hale (Member # 1518) on 03-23-2003, 04:47 PM:
 
Check out

http://pro.colorkinetics.com/showcase/

They make some very cool architectural lighting for various applications, including theatres. More general leds lights, including incandescent replacements, flourescent replacements, even streetlights, is at http://www.theledlight.com/120-VAC-LEDbulbs.html

This stuff is coming very quickly. I remember hearing that BP (or maybe it was Shell) is going to convert all their gas stations to white led light.
 

---

Posted by John Pytlak (Member # 331) on 03-23-2003, 08:12 PM:
 
Check out the new Organic Light Emitting Diode (OLED) technology from Kodak, coming soon to a digital display near you.

http://www.kodak.com/US/en/corp/display/index.jhtml

http://www.kodak.com/US/en/corp/display/kodakAndOEL.jhtml

http://www.kodak.com/US/en/corp/display/overview.jhtml
 

---

Posted by Gordon McLeod (Member # 33) on 03-23-2003, 08:18 PM:
 
John I remember reading abut some organic memory simms that kodak was developing that was faster than any current computor memory
 

---

Posted by John Pytlak (Member # 331) on 03-23-2003, 08:49 PM:
 
All thinking humans have "organic memory", don't they?

Link to Kodak R&D:

http://www.kodak.com/US/en/corp/researchDevelopment/
 

---

Posted by Steve Kraus (Member # 476) on 03-23-2003, 09:27 PM:
 
Finding a thinking human would be a problem there...
 

---

Posted by John Pytlak (Member # 331) on 03-23-2003, 09:32 PM:
 
At Los Alamos? Lots of non-military research now.
 

---

Posted by Christopher Seo (Member # 7) on 03-24-2003, 05:19 AM:
 
Steve made some good points earlier. What if the LEDs were engineered so that the R,G,B spectral emission peaks of the LEDs involved coincided with the spectral response peaks of the C,M,Y dyes in the film? Or, taken to the hypothetical extreme, what if only the peak wavelength for each dye were used as illumination, as with lasers? Would the projected image of a well-shot film look any less rich or vibrant? Granted, discrete spectra are bad for viewing real-life objects, but it would seem to be different for projection, where the "object" we are viewing is color dyes with precise spectral characteristics to which the discrete light source can be optimized. And lasers certainly make a good point source....
 

---

Posted by Steve Kraus (Member # 476) on 03-24-2003, 07:06 AM:
 
Oh, "there" referred to the concept of finding a thinking human (!) in general not anyplace specific.
 

---

Posted by John Pytlak (Member # 331) on 03-24-2003, 01:04 PM:
 
Christopher Seo wrote:

quote:

---

What if the LEDs were engineered so that the R,G,B spectral emission peaks of the LEDs involved coincided with the spectral response peaks of the C,M,Y dyes in the film?

---