4Sevens Quark RGB - A Layman's Perspective

[Luminescent]

I'm afraid that wouldn't work.

What you have here is a white emitter plus three monochromatic emitters.

So by mixing the dies, you'd just get the white spectrum (either warm or cool, depending on which one you chose) with single-wavelength emission lines superimposed on it from the red green and blue dies.

What you are after is a continuous white spectrum with a variable colour temperature. Not the same thing at all. Probably not very easy to do with LEDs either, but this certainly isn't the way to do it.

First of all, the 'white' spectrum from white LEDS are no where near a 'continuous' spectrum. There is a HUGE fairly narrow blue spike created by leak through from the blue LED die used to drive the top phosphor layer. To get something even remotely 'white' they balance the blue spike with a strongly yellow biased broad spectrum hump created by the top phosphor layer.

Second, the nominal 'single color' LEDs are not lasers, and do not produce real monochromatic light. Their spectrums are quite wide compared to a laser, and though they are not wide enough to completely fill in the gaps between the spectrum spikes, they are more than wide enough to give a three color source with a CRI that will match or exceed what you can get with a 'white' LED.

Third, because the human eye itself, as well as nearly all photographic and video systems, are based on a set of tri-stimulus filters centered roughly at Red, Blue, and Green wavelengths, you can indeed get a fairly good simulation of various 'color temperatures' by adjusting the RGB balance. Sure it's not a real black body continuous spectrum, but your eye can't see that, it can only see three sample color bands, and if you tweak those levels to match the black body spectrum equivelents for say a 3200K black body, your eye would be fooled even though the spectrum would be somewhat discontinuous.

Indeed, in practical human vision and photographic applications, the well controlled wavelengths of the individual RGB LEDs, coupled with precise settings of output levels that is possible with separate emitters, could give much more precise and consitent color rendering than the typical 'luxeon lottery' found in white LEDS.

A more practical issue arises due to the off center arrangement of the 4 dies on the chip. The result would be that the colors would register and give a nice even white combined color balance only in the center. Around the edges, the colors would not register perfectly, creating a variable level of color mixing. This would create the appearance of a circular rainbow spectrum around the periphery of the beam circle.

I think 4 Sevens really missed the boat by not at least adding a 'continuous color change' mode that gradually mixes each color with it's adjacent color (rather than just hard switching to the next color) to form the complements and near complements in what would appear to the eye to be a continuously varying color rainbow spectrum.

For example, midway between RED and GREEN, with both the RED and GREEN LED's driven equally, you would get a brilliant fully saturated YELLOW. If you vary the mix, the apperent color tint would change. Want a less saturated yellow like amber? Fine, just mix in some BLUE.

User interface would be a little tricky to figure out, but the mind boggles at the possibilities . . .

- Luminescent

 

[Moonshadow]

Luminescent;

OK - I didn't realise that I was going to have to be so pedantic. Please feel free replace the word 'monchromatic' in my previous reply with 'narrow band' emission or whatever you prefer.

The point is still valid though. You are correct that displays that emit light that enters the eye directly (for example a colour TV) can create passable representations of most colours by mixing red, green and blue.

But if you want to use light to illuminate a coloured object, then you have a different problem. What the eye sees is then the reflected light from the object - a combination of the incident spectrum and the reflectivity of the object as a function of wavelength.

Imagine an LED light that emits red and green in two narrow bands centred on 630nm and 550nm. (since you mentioned it, no they are not lasers so not perfectly monochromatic - let's allow them a gaussian profile with a FWHM of 20 nm).

The emitted light might look yellow, but what do you think is going to happen if it falls on yellow object that reflects light predominantly in a range centred on 590 nm ? The light itself may look yellow but it isn't going to render the yellow object correctly.

OK, this is a bit of an artificial example - real coloured objects are more complicated than that but the point stands - for good colour rendition, you need a continuous spectrum, not one with narrow bands of specific wavelengths.

 

[carrot]

Hi, let me just pop in here:
I spoke with David about making the RGB have continuously variable output for all three colors and he says that right now it wouldn't be possible in the same form factor. The RGB already had to be lengthened to fit the current driver and to do multiple colors simultaneously they would each need a similar driver.

Also, the RGB may have found a permanent place on my belt simply due to how useful the bright white flood is to me. The flashing modes are annoying but I have been teaching myself to simply ignore them.

 

[Lightcrazycanuck]

Interesting,very Interesting.



lovecpf

 

[Luminescent]

Moonshadow;

Your point about the the reflectance spectrum of some materials having narrow spikes that can effect color rendering is valid, but the CREE MC-E Color is an RGBW device that incorporates a white LED with a broadband phosphor in addition to it's three separate RGB emitters.

The area in between the RED and GREEN emitters is filled in by the broadband phosphor layer on the WHITE emitter.

The critical GREEN emitter has a spectrum that is about twice as wide as you indicated, and it is positioned to further extend and enhance the broadband response of the white LED's phosphor emision layer in the critical GREEN area of the spectrum.

Look at the image below, and you will note that, though the spectrum looks a little spiky, ALL THE CRITICAL WAVELENGTHS ARE COVERED. What gaps there are, mirror the gaps in the human spectral response, so any 'reflection differences' that occur in these gap areas could not be distinguished by the human eye anyway.

In other words - WHAT THE MC-E lacks in spectral response, you pretty much DON'T NEED to get good color rendering.

Adjusted down to appropriate matching levels, the RGB emitters are also very well positioned not only to extend the phosphor spectrum, but also to tweak the levels of the critical pass bands for human vision, so by setting a combination of all FOUR emitters, you can indeed achieve BOTH a very good CRI and a fully adjustable apparent color temperature.

 

[Luminescent]

Hi, let me just pop in here:
I spoke with David about making the RGB have continuously variable output for all three colors and he says that right now it wouldn't be possible in the same form factor.

Not sure why that would be . . .

Obviously the CPU has switching control over the LEDs, so it would seem like you could simply implement software PWM and have the CPU continuously switch between them at a high PWM rate (say 2-4 KHz), with the PWM dwell time at each color setting it's relative intensity.

- Luminescent

P.S.
If you use this trick, you owe me a free Quark RGB (just kidding )

 

[apontes]

Being just a white floodlight would already be enough.

Having the option of a super low red is fantastic.

Green and blue are just bonus.


(my opinion, of course, based on my particular needs/uses)

 

[Mike V]

I really don't see why this LED couldn't do colour mixing?

I've used the ARRI PAX system (which by the way is completely amazing).
With this light system you can mix any colour including creating white of any colour temperature.

The LEDs that the PAX system uses looks similar to the CREE, but I believe it uses 5 colours (not three).

It looks like this:

http://www.arri.de/uploads/pics/IMAGE_light_engine_lang_02.jpg

See Rule #3 Do not Hot Link images. Please host on an image site, Imageshack or similar and repost – Thanks Norm


.

 

[Mike V]

This isn't on a flashlight, but just as an example, here is the interface on another ARRI LED light called a Broadcaster in which you can colour mix:

http://www.arri.de/uploads/pics/product_image_header_BroadCaster_02.jpg

See Rule #3 Do not Hot Link images. Please host on an image site, Imageshack or similar and repost – Thanks Norm

 

[Moonshadow]

the CREE MC-E Color is an RGBW device that incorporates a white LED with a broadband phosphor in addition to it's three separate RGB emitters.

Luminescent;

Yes, I'm well aware of that, as you'll see if you would do me the courtesy of re-reading my original post:

What you have here is a white emitter plus three monochromatic emitters. [Feel free to replace this with narrow band emission as previously discussed]

So by mixing the dies, you'd just get the white spectrum (either warm or cool, depending on which one you chose) with single-wavelength emission lines [same comment] superimposed on it from the red green and blue dies.

But thanks for the tutorial anyway.

The graph you plotted kind of reinforces the point I was trying to make: the R, G and B LEDs contribute very little in the range centred on 590nm for example so if you had a yellow coloured object that reflected strongly over those wavelengths you could vary the intensities of the red green and blue LEDs all you like and it would make no difference to the light coming back from the object. [My previous caveats about this being a simplified example apply].

Now it would be a different matter if the output of each of the coloured LEDs closely matched the spectral sensitivity profile of the receptors in the retina, but they don't, especially not the red one.

Anyway, let's not clog up a good thread about a great light. If you're convinced it'll work then by all means go ahead and make one, but don't blame me if it doesn't work as well as you think it's going to.

 

[mwaldron]

Also, the RGB may have found a permanent place on my belt simply due to how useful the bright white flood is to me. The flashing modes are annoying but I have been teaching myself to simply ignore them.

I agree with this as well, the flood is amazing on the RGB (in color or in white) and I find it useful in a much different set of circumstances than my usual lights.

If it were my project, the next revision would hide the blinky modes (like the Mini/Preon does) and fix the bright flash when it's set to low.

I'm not so sure I care about color mixing, it might be fun but I'm actually finding this light useful and I'm pretty sure anything more would make it too hard to use.

 

[Luminescent]

Sorry duplicate post - how do I delete ???
(the "edit" button says 'edit/delete' but I don't see any 'delete' feature)

 

[Luminescent]

Luminescent;
Yes, I'm well aware of that, as you'll see if you would do me the courtesy of re-reading my original post:

But thanks for the tutorial anyway.

I did read your original post.

You took 'Mike V' to task, correcting him for suggesting that this LED might offer the possibility to tweak CRI and Color Temp.

Sorry but your 'correction' was not quite correct

The fact is that a very good CRI and a fully tuneable Color Temp CAN be achieved with the MC-E Color device by using all FOUR LED chips tweaking the levels of all FOUR LED at the same time (did you get that FOUR, not three all FOUR)

I first proposed this RGBW solution to address the issue of the narrow band color spectrums of the separate RGB LED sources way back in Aug 2007 in this post. (I also mentioned the idea of simply matching the LED spectral widths to the response of the eye, but this is difficult to do, because apperently high power LEDs hit thier optimum power efficency at narrower spectral widths.)

Good points. Natural objects do reflect very complex spectrums . . .

In a pinch you could always use a variation of the CYMK trick used in printing, by including a high color rendering index white LED along with the Red, Blue, and Green chips.

In this configuration, highly saturated colors would be rendered by the R, G, and B drivers alone, and less saturated colors would be mixed with more and more 'white' light from the white LED as the color saturation decreased.

If you look at the CREE MC-E spectrums I just posted, it's obvious that CREE device is optimized for RGBW operation, just as I proposed. The GREEN and RED LEDs included in the device extend and enhance the performance of the WHITE device's phosphor spectrum so that the combined CRI can indeed be tweaked to be better than the WHITE LED's individual performance (the GREEN and RED driver chips can be used to widen and flatten the central spectrum area). By slightly tweaking the RGB bands you can tilt the combined spectrum, and tune the equivalent color balance.

Apperently CREE thought the RGBW solution was a really fine idea, even if you don't.

Getting back to the Quark RGB -

This light is a really nice beginning, but I don't think it realizes more than a fraction of the possibilities of the MC-E Color device.

How about -

Fully tunable Color Temp in White mode

Fully tuneable Color and Saturation in Color mode


and some silly but fun features like -

Rainbow sweep strobe

Random Color strobe

- Luminescent

 

[Mike V]

Yes, exactly.

Not using the RGB alone to make the white, but using the white, plus some RGB to change the colour hopefully also improving the CRI.

 

[carrot]

What is there to be gained, from a consumer point of view, by having a variable color temperature flashlight? I am curious because I can't think of an application where I would find it useful to adjust a color temp on a light? As I was told, it would add significant complexity to the driver.

 

[Quension]

A variable color temperature could make it useful as a secondary light source in photography, when not using a studio setup. It could be matched to whatever the primary light source is.

 

[Mike V]

Also a lot of people on here are fans of "Warm white" lights.

So they could dial in the colour that they prefer.

 

[carrot]

But the Quark RGB already offers that -- it comes in two versions, Cool White and "Neutral White".

 

[StandardBattery]

Having a variable color temperature would be awesome, not to mention if CRI could be improved, or at least altered for the spectrum needed at the time.

Just as the sunlight changes temperature during the day, the flashlight could do the same because our brains are already programmed to adapt to that.

Changing color temperatures when moving from Manmade to Natural environments would be very useful. Even different natural environment, the water, geographical location, altitude... all good candidates for adjusted color temperature and/or spectrum.

Is is a requirement... no are brains adapt pretty good, could it be helpful.... absolutely.

The interface for such a light and the CPU software to make it easy would be more complex... but I like the idea. :thumbsup:

Someday...

 

[jsfotografie]

wow mike where did you find that picture?
and lucky you that you can work with arri lights did you try their Arri L7-c?

best janosch

I really don't see why this LED couldn't do colour mixing?

I've used the ARRI PAX system (which by the way is completely amazing).
With this light system you can mix any colour including creating white of any colour temperature.

The LEDs that the PAX system uses looks similar to the CREE, but I believe it uses 5 colours (not three).

It looks like this:

]http://www.arri.de/uploads/pics/IMAGE_light_engine_lang_02.jpg


See Rule #3 Do not Hot Link images. Please host on an image site, Imageshack or similar and repost – Thanks Norm

.