LED Color Aesthetics: The Emperor’s got no clothes. Or does he?

[zygote2k]

So from what I gather from this whole thread is that the CW leds don't give off the right amount of blue wavelength and that when you supplement the CW's with RB's, there is a weird color seperation issue when it shines onto the corals.

Maybe the answer lies within the ratio of CW to RB. Maybe someone needs to build a cluster of 6 CW's with a single RB in the center. If you make the optics of the whites tighter and the RB optics wider, at the bottom of the light cone the two colors would lie perfectly on top of each other creating a whitish light with just enough blue to excite the pigments of the corals.

Maybe this approach is way too simple- can you think up some scientific reason why it won't work?

[sseaner]

Hopefully the guys with the new AI modules will post some info and pics soon. This will give us a better idea how this combo works. Do the three different colors allow independent control or are the blues lumped together?

[jusmee]

Originally Posted by zygote2k

So from what I gather from this whole thread is that the CW leds don't give off the right amount of blue wavelength and that when you supplement the CW's with RB's, there is a weird color seperation issue when it shines onto the corals.

Maybe the answer lies within the ratio of CW to RB. Maybe someone needs to build a cluster of 6 CW's with a single RB in the center. If you make the optics of the whites tighter and the RB optics wider, at the bottom of the light cone the two colors would lie perfectly on top of each other creating a whitish light with just enough blue to excite the pigments of the corals.

Maybe this approach is way too simple- can you think up some scientific reason why it won't work?

A good way to do it, is to use modules with multiple chips on them (instead of the stars with one chip each that most have been using). I've seen them with 4, 5 or 6 LEDs, and maybe more. Then mix the colours on each module.
Like these

[zygote2k]

now you know the limitations of the leds themselves. what we have to do now is focus on optic aiming and overlapping with the right color combo to ensure a nice 14k look without spotlighting or color banding. if some of the mounting surfaces were concave or convex, it may or may not help with the color mixing.

[jknecht]

I read the new tri-color pucks from AI will still be 2 channel. So, the blues will be controlled together.

[chimmike]

Originally Posted by scolley

So that I could demonstrate what I was talking about, here are some pics. All taken within the last 48 hours. All of these shots are COMPLETELY unaltered, EXCEPT for the white balance corrections my camera imposes. Visually, I'd say these shots are reasonable faithful to the colors in real life. Not perfect. But close enough for the dialogue this might stimulate.

Only because I requested Full Tank Shots (FTS's), here's a shot of my tank under the Radium and actinics, followed by the LED with XR-E CW's and RB's.

Radium and actinics



XR-E CW's and RB's




But to get a little more detail, for a closer comparison, I decided to take some close-ups of a few different subjects under four different conditions. I've grouped them together by subject matter, for ease of comparison. The four sets of lighting conditions were all created today, all within a few minutes of each other. The first condition was my 150W 20K Radium with PC actinics (photographed last in actual fact though... and some zoas started closing up from the dark period as I had been swapping lights). Second is 12 Cree XR-E's Cool Whites and 12 Royal Blues. The CW's were dimmed until the sand no longer looked purple, but the dark white tone that is actually is. The next is just the same Cool Whites, with no blue. And finally is the Royal Blues, no white.

The subjects are three clusters of zoa's, numbered 1, 2, and 3 respectively. Followed by some GSP and a Maxima. Then a blue and green acan, and finally a red and blue acan.

Hope this at least sheds some light on what I'm seeing. Enjoy.

What I'm seeing in those pictures, especially the LED, is those optics have your whites focused on specific areas. Take the optics off all together and see what they do. Remember with those optics like that, it looks like they're focusing the light and not being allowed to overlap as they would without the optics, or with the optics on but the fixture hanging higher.

One build that comes to mind is the guy who put his fixture on the ceiling and used 8* optics for his tank. There are no 'spots' of focus, the light overlaps cleanly.

Also, again, you did a 50/50 mix of whites to rb's. I've seen numerous places here to use more rbs than whites by a good bit.

Now, I'm going to do my 3rd fixture the same as my two currently: 14 rb's, 13 xp-g cool whites. Beween those 3 fixtures I'll likely make a couple strips and try a UV and green, and maybe a red or two just to see if the color poppage occurs.

[CalmSeasQuest]

Great thread - Tagging along.

I have both the AquaIllumination (DT) and Maxspect (FT) running side by side (comparison thread w/photos and PAR measurements)

While virtually all corals are exhibiting great growth, I (apparently like many) struggle to regain the colors achieved under MH. I have found it especially difficult to obtain the reds achieved under MH or T5. For me, ORA Red Planet is a great example of a tough piece to get "colored up."

Another challenge to those new to LED is the massive PAR output. Anyone trying to match LEDS to T5 or MH based on "appearance" will quickly bleach most everything in their tank (been there, done that prior to acquiring a PAR meter.)

Obviously husbandry and water quality plays a major role in obtaining the best colors, that aside, I still believe we're missing something in current LED usage.

I had an opportunity to speak with Sanjay and Clive (Evil66) at a recent MASM event, specifically focusing on the the LED color challenges. Their consensus (to which I agree) is there is something missing from the spectrum of LEDS currently in use. The "best guess" seems to be something higher in the spectrum, perhaps in the red range. Although Clive mentioned he was planning more testing with UV, neither seemed convinced UV was the answer.

I'm still a huge LED proponent, but speaking for myself, I haven't yet learned how to replicate the colors possible through other lighting methods. I'm planning on upgrading the AIs to the new tri-color pucks when the retro kits are available, but based on the lackluster performance I've experienced with the Maxspect (which contains Royal Blue and a few near UV LEDS) I'm not convinced they will be the total solution.

[jtma508]

A slightly different way to look at the LED challenge is to consider T5's as opposed to MH's. Fluorescent bulbs work by combing different phosphors to generate their various spectra. If you look at their spectral plots you quickly see what I mean. This is similar to the way some LED's work (white for example). A blue LED is treated with a 'white' phosphor to get white light. We may need to mix-in various LEDs, much like T5 phosphors, to get what we need. The challenge is finding the right mix without unnecessarily complicating the build. I mean you can only control so many driver channels.

[jnosky]

Hey guys! just read this entire thread. Been interested in converting my 125 to LED and have been reading alot of diff articles, reviews, etc.. This particular issue is one reason why i have not gone any further on my project. My next step is to buy a few different LED and experiment with different color options.

IMO There seems to be two different issues.. one appears to be "how much of what color do we need to get the color that we're seeking" Using a 6ft tank as an example, if you only put 2 r 2g 2y led's on each side, i think you would have some coverage and blending issues.

The other is "How do we then BLEND the color to get some sort of equal colorization once we figure out what color"

One idea i've been kicking around was using a MH reflector (trapezoid, possibly parabolic type) then mounting the LEDs in the top of the reflector. This should allow me to use no optics and should allow the color to blend and aim inside the reflector. Using a seperate dimmer for the color strings, Blue and RB, and white; i think that should provide all the adjustability to see what kind of color range i can achieve.

IMO Everyone's on the right track, but the last thing i want is a big /red/green or yellow spotlight in corners or any issues like that. I'm sure they can be solved with an 80 or 60 optic, but just haven't heard anyone bring this particular issue up yet.

[chimmike]

a reflector is really not necessary. A LED without optics has a 120* 'throw' if you will. I think the OP, first, should remove all his optics. I see spotlighting based on his optics.

Then, reduce the number of cw's. Need more rb's than cw's running. Then more adjustments via dimmer to reach that good 'balance' of blue/white. once that balance is reached, THEN we should delve into trying other LED's to reach the coloration desired out of our corals. Of course this will take time to get the reaction from the corals.

I'll be building a couple, or maybe 3 strips, with various color LED's, to see what works best color wise. All the colors will have their own specific dimmer so I can isolate them as needed. I think that may be the best way for me until some more research is done by those more able than me.

[jtma508]

I think what we need is to build a test unit with maybe 1 CW (or NW) and 1 RB and to scan that output with a spectroradiometer and compare the output to a T5 (or maybe) MH 'standard'. Identify missing peaks from the 'standard' and start adding additional LED's to the test unit to approximate the spectrum of the 'standard'. We could start with two channels: <500nm and >500nm. With some effort we could figure out the necessay ratios to get close. Anyone have an Apogee PS-200 laying around or the $4000 to buy one??

[Kolognekoral]

If we consider the actual spectrum being presented by LEDs in the current fashionable combination of 1:1 CW to RB, then we are simply not getting enough red light to allow a red object to appear red. Don't forget, the colour we see is the light reflected from an object. If this part of the spectrum is not in the available light, then the object cannot reflect this colour to our eyes. Thus are red corals, (Red Planet) no longer red to our eyes. LEDs are particularly wavelength specific, even with a phosphor. I suspect coming generations of 'whites' will have a better spectrum, but we shouldn't forget we are working in a truly niche area for lighting. The need for a wider spectrum may not be generally perceived for home/business lighting. It could take a while. We may not only need to play with the various other LED colours, but consider a hybrid technology including a T5 to boost the spectrum.

At the moment, I am planning on trying neutral whites instead of cool whites, as there is a bit more red in there without loosing the violet-blue. As we know, it is finally a matter of personal taste.

[Koyaanisqatsi]

It's also important to keep separated the two concepts of color and fluorescence. I keep seeing people talk about adding more royal blue when what they really need is more red (or green, or whatever). If a color is not fluorescent, adding more royal blue will not help illuminate it. As stated above, you need red light to make red show up. Royal blue and other short wave colors are going to make the fluorescent colors glow, but they will only illuminate blues and purples.

I realize that you need the blue and violet spectrum for proper growth and such. But only so much is necessary.

[zygote2k]

could'nt we use a coated glass lense to shine the CW's through to achieve the proper spectrum?

[chimmike]

Originally Posted by zygote2k

could'nt we use a coated glass lense to shine the CW's through to achieve the proper spectrum?

wouldn't that kill the PAR ?

[zangmann]

I thought the majority of the useful par came from the RB, not the whites. Isn't that why the CW isn't as important. Meaning a NW or WW can be swapped out with no ill effects regarding the par output?

[jusmee]

Originally Posted by CalmSeasQuest

Another challenge to those new to LED is the massive PAR output. Anyone trying to match LEDS to T5 or MH based on "appearance" will quickly bleach most everything in their tank (been there, done that prior to acquiring a PAR meter.)

Yes, I have noticed that most designs use a lot of LEDs, and some basic calculations I have done suggests maybe a smaller number would be better. Too much light will contribute to the washed out effect too I think. Rather than go by any watts per gallon, or watts per length, or even watts per surface area, all of which do not take into account efficiency, i.e. light output per watt, why not use the lumen output, the geometry of the viewing angle (or optic angle) and the area and depth within you tank at which you want to place corals to figure out true light levels? When I do this, I come up with less LEDs than the watts per gallon method - for example.

[jtma508]

People don't use watts per anything as far as LEDs go. It seems the three issues are PAR, spacing (to achieve penetration and light blending) and spectrum. If we can solve all three we'll get the growth and the aesthetics we're after.

[jusmee]

Originally Posted by jtma508

People don't use watts per anything as far as LEDs go. It seems the three issues are PAR, spacing (to achieve penetration and light blending) and spectrum. If we can solve all three we'll get the growth and the aesthetics we're after.

Good to hear - but I have read a lot about LEDs recently, and MANY web sites and forum threads still talk about how many watts of LEDs they have or will need. Great that this is changing. Otherwise, it is confusing for people when they start out reading about LEDs.

[jtma508]

The thing is, LEDs can be run over a range of currents so the 'watts' measure becomes irrelevant. If you were to compare, for example, a CW XP-G R5 (139lm) and a CW XR-E Q5 (107lm) both run at 700ma you'll have two LEDs at the same wattage but the former producing ~30% more light.

[Foxy Brown]

The trouble you're having with color separation is because the cool whites have almost 0 intensity at 500nm (cyan) and at 650nm (red)... you're eye cannot see colors properly. Even though it looks white, and pigment that reflects back those colors will have no light to reflect and hence will look weird.

I actually do optics for a living so I'm used to this idea. I've recently posted on a (maybe) novel color combination that should help this out.

If you go only with CREE LEDs then using warm-white + green + blue + royal blue will get you there.

See this post for actual spectra:
http://reefcentral.com/forums/showthread.php?t=1888163

If it were me I would actually combine XPE royal blues and Luxeon K2 cyan with some XPG warm whites as the Luxeon wavelength is almost perfect.

Check my other post and let me know what you think.

[jtma508]

As I've posted previously, I'm not comfortable with the Cree spectral plots for their white LEDs. To my eye those plots appear seriously smoothed. I'd like to see a real spectral scan of each white category (CW, NW, DW, WW) before attempting to decide which way to go.

See the discussion starting here: http://reefcentral.com/forums/showth...78127&page=176

And earlier in this thread. I believe if we had better data on the whites we'd only need a couple of added 'spot' LEDs beyond the 500nm region.

[Foxy Brown]

BTW zygote, no you can't use a coated glass to do that, the cyan and red wavelengths just aren't there and a filter can only remove what is there, not add what isn't. The trouble is in your eyes that don't really tell the truth... you need to look at the actual spectrum.

You either need to do it with fluorescence, like they do inside the white (and inside fluorescent lights) or you need to add more colored LEDs. It is very possible to make fluorescent glass, but its not something you can buy and would be amazingly expensive.

[Foxy Brown]

jtma, I love that write-up you did... But, being someone who does this stuff for a living, I have to say no - the CREE plots are not smoothed. I actual know what the fluorescent materials they use are and what their spectra look like... the spectra in the datasheets are real, just the sum of an RB led plus 1 or2 different fluorescent emitter, depending on which white.

I think WW, blue, royal blue, and cyan (or green) will do it... if you get the ratios right.

[jtma508]

I've been trying to approach the problem by looking at what lighting has been 'proven' for both growth and aesthetics. If you look at the spectral data for popular MH's (Sanjay's work is awesome for this) and/or T5's (although some of the spectral plots published for T5's are suspect), using the broadband whites will produce alot of spectrum that are not found in either the T5 or MH spectra. This is especially true in the G-Y and near O-R regions.