LED Color Mixing

White Nichias have a significant blue tint to them, but they are efficiant for a white LED. Color LEDs tend to be more efficiant, especially those around the lower wavelengths. So my question is this: How many people have experiance mixing Nichia whites with other colors to achive a more neutral white and increase efficiancy? This is the only page I have come across that used this technique. Does anyone else here have experiance with this? What color (green, yellow, or amber) would you recomend to best offset the tint of the Nichias? Unfortunately I don't have a bunch of colored LEDs to experiment with, so I come to CPF for advice. Let me know.
Thanks, Eli

I'm planning on doing this when I start installing lighting in my HO passenger cars. I've already tested various combinations. By varying the proportion of amber and white (and/or the current through each) I'm able to duplicate any color temp I want from 2700K up to 6500K. I'm using mostly ChiWing white (CCT ~8000K) and amber LEDs for this project. I also mix white and orange-this requires fewer non-white LEDs to lower the color temp, and it results in white light with a slight reddish/purplish tinge similar to high CRI 4100K fluorescent tubes. There doesn't seem to be too many problems with mixing the light, either. I have the LEDs beam projecting lengthwise on the ceiling (covered in white contact paper) which gives even lighting and good diffusion. I make sure I don't have just ambers or just whites on the same side of the car. For example, for 5000K I used one amber and one white on each side. I have the amber on the right side on one end and on the left side on the other. The result is pretty good-not much indication of mixing, and a lowered color temp.

I made up an Excel spreadsheet to determine the ratios once I figured out the relative luminous flux of each type of LED. By doing this I was able to see not only the CCT but also how close to the Planckian locus I was coming. In most cases it was pretty close.

[ QUOTE ]
jtr1962 said:
By varying the proportion of amber and white (and/or the current through each) I'm able to duplicate any color temp I want from 2700K up to 6500K.

[/ QUOTE ]

Excellent!

This seems similar to the idea of mixing "white" and warm fluorescent lighting.

Are there any gaps in the spectrum when using white and amber LEDs?

I've noticed that the newer florescent replacement light bulbs/coils/folded tubes seem to bias a lot toward yellow - to the point some of my fluorescent bulb replacements seem more yellow than even the regular soft white tungsten light bulbs - anyone know if there is a good reason for this?

Aside - I have a fluorescent replacement bulb that's a cool white and although it does put out a lot of light -
I don't seem to see as well when compared to the more yellow fluorescents, and/or regular tungsten light bulbs of similar output ratings.

Similarly I have also noticed I seem to see better at similar output levels with a warmer white than cooler white LED, and using dim LEDs I see better with yellow/amber, than a white (other than color rendition)
Comments, suggestions?

[ QUOTE ]
UnknownVT said:
Are there any gaps in the spectrum when using white and amber LEDs?


[/ QUOTE ]
No more so than just using whites by themselves. In fact, the yellow phosphor used in most white LEDs peaks around 550 nm. Amber LEDs peak around 590 nm, so the amber simply bumps up the portion of the spectrum in the region from around 575 to 605 nm. It doesn't create gaps, but by the same token it doesn't fill in any, either. By using red instead of yellow you can both lower the color temperature and add output in an area of the spectrum where many white LEDs are somewhat deficient. The drawback to using any color other than amber for this is that you end up noticeably far off the Planckian locus. However, if my goal is to imitate certain types of fluorescent tubes then this is actually desireable.

[ QUOTE ]

I've noticed that the newer florescent replacement light bulbs/coils/folded tubes seem to bias a lot toward yellow - to the point some of my fluorescent bulb replacements seem more yellow than even the regular soft white tungsten light bulbs - anyone know if there is a good reason for this?


[/ QUOTE ]
None other than trying to imitate the much lower color temperature chandelier bulbs rather than halogens or standard incandescents. I believe the line of thinking here is that since these are drop-in incandescent replacements, then the designers believe that the end user prefers that type of light (or else they would already be using fluorescents). I've already had many discussions here regarding color temperature and individual preferences. Suffice it to say it's more a matter of taste than anything else. I tend to diagree that the majority would prefer 2700K once shown a range of choices. Some research tends to suggest the majority would be happiest right around 3500K to 5000K, depending upon lighting levels (higher lighting levels seem more natural at higher color temperatures and vice versa). My own preference is right around 5000K regardless of lighting levels, but I can readily adapt to anything from 3500K to 6500K. Above or below that I experience headaches and see poorly.

[ QUOTE ]

Aside - I have a fluorescent replacement bulb that's a cool white and although it does put out a lot of light -
I don't seem to see as well when compared to the more yellow fluorescents, and/or regular tungsten light bulbs of similar output ratings.

Similarly I have also noticed I seem to see better at similar output levels with a warmer white than cooler white LED, and using dim LEDs I see better with yellow/amber, than a white (other than color rendition)
Comments, suggestions?

[/ QUOTE ]
As for the first part, you need to compare tubes with similar color rendering before coming to a conclusion. CFLs usually tend to have a CRI of around 82. Cheaper cool whites around 62. If a tube has a better CRI you'll see better under it regardless of color temp. Most studies suggest the opposite of your experiences-things tend to be crisper under higher color temperatures. However, these studies don't take into account that many people's brains have in effect learned to see better under incandescent. This could be why warm white fluorescents and amber LEDs work better for you. I really wish I had a better answer for you than that. I've compared my modified bike light (20 white LEDs) to the stock halogen of about the same output. It easily appears about twice as bright, and colors seem more natural (closer to sunlight) to boot. This is in spite of the poorer CRI of white LEDs. And I just love my 5000K T-8 tubes with a CRI of 91. These provide great seeing, and come quite close to sunlight.

[ QUOTE ]
jtr1962 said:you need to compare tubes with similar color rendering before coming to a conclusion. CFLs usually tend to have a CRI of around 82. Cheaper cool whites around 62. If a tube has a better CRI you'll see better under it regardless of color temp.

[/ QUOTE ]

Thank you for that explanation - that makes sense.

The cooler fluorescent tube is a cheapo - so most probably has poor color rendering. That may well explain why I may not see as well despite the fact it seems "bright".

[ QUOTE ]
jtr1962 said:Most studies suggest the opposite of your experiences-things tend to be crisper under higher color temperatures.

[/ QUOTE ]

Yes, for me too - I do see better by normal daylight than any artificial light - perhaps it's the color rendering that also makes the difference rather than the temperature alone.


[ QUOTE ]
jtr1962 said:However, these studies don't take into account that many people's brains have in effect learned to see better under incandescent. This could be why warm white fluorescents and amber LEDs work better for you.

[/ QUOTE ]

This is something I have pondered for some time - I have read elsewhere that the human race has been conditioned to yellower colored light - as flames/fires have been the means of lighting at night, and daylight tends toward warm for most of the day - so our eyes have adapted to this bias in the spectrum.

I also realize that the brain compensates a lot - where scenes under regular tungsten lighting may seem "natural" -
but anyone who has seen a photo taken with daylight (white) balance - will immediately see how yellow the scene actually is.


[ QUOTE ]
jtr1962 said:I've compared my modified bike light (20 white LEDs) to the stock halogen of about the same output. It easily appears about twice as bright, and colors seem more natural (closer to sunlight) to boot. This is in spite of the poorer CRI of white LEDs. And I just love my 5000K T-8 tubes with a CRI of 91. These provide great seeing, and come quite close to sunlight.

[/ QUOTE ]

Sunlight seems to be my "ideal" too.

Perhaps like color temperature - color rendering (CRI) may not tell the whole story -

perhaps (I am only speculating) the gaps in the spectrum with white LEDs are not detrimental to your seeing, and the wavelengths that are emphasized may enhance your seeing?

Perhaps it is my conditioning, but for example -

I can quite definitely see that the newer HID headlights on cars are noticably brighter and throw better -
yet I don't think I see that much better with HID - than say good high intensity halogen lights - despite the fact the halogen do not appear as bright. I may have previously put this down to just color temperature - but perhaps it is combination of other factors too.

Thanks for being so helpful

UnknownVT said:
[ QUOTE ]
I can quite definitely see that the newer HID headlights on cars are noticably brighter and throw better -
yet I don't think I see that much better with HID - than say good high intensity halogen lights - despite the fact the halogen do not appear as bright. I may have previously put this down to just color temperature - but perhaps it is combination of other factors too.

[/ QUOTE ]

I was wondering about that myself, as my new car has HID's and I'd swear they are much, much brighter than standard incandescent lowbeams. My suspicion is that headlight standards for automobiles are only specified by wattage rather than lumens or lux, and as a result the more efficient HID's are much brighter.

Anyone have experiance mixing white and red LEDs? How is the color quality?
Thanks, Eli

There are the RGB phopshor UV LEDs made by Toyoda Gosei that have a higher CRI.

http://www.microlamps-tg.com/html/Prod_white_compare.htm

Look at the one on the bottom.

I've done a number of RGB solutions, and the colors look alot more vivid than normal. I believe this has to do with the color saturation and the response of the eye.

If you look at http://www.lumileds.com/pdfs/techpaperspres/SID-BA.PDF on page 6, you'll notice and RGB solution can cover a greater area than even a CCFL solution.

If it's of any help please take a look at this thread -
it's only an ad-hoc/informal experiment - but might be worth looking at:

LED Colors and Vision (pics)

Some while ago we discussed splitting a reflector into several parts, one for each lightsource, and one fellow split a mag ref. in two, removed the two parts from each other by a couple of millimeters and inserted two leds facing away from each other (sideways into the reflector).

Would that be a viable way to create a RGB white, or need the sources be closer maybe? (Weird fringes?)

Are the current high wattage colour leds proper for such a thing? What are the desired characteristics of the colour output from the leds? A narrow spectrum, of a certain frequency?

I'm not sure a shiny reflector is the best thing because you want diffuse refection rather than specular reflection to help mix the light better. Other than that the principles are sound, and resemble what I'm doing (having LEDs of different color facing each other to get lower color temperature white).

Besides mixing LEDs to lower the color temperature, I also tried to make a synthethic white by using the same principle. I used ocean blue (slightly bluer than cyan) and amber LEDs. The result did look "white" to my eyes, but it was a very strange white in that some colors (like green) really didn't show up well at all. As for the best LEDs to use for RGB white, the wider the spectrum the better the color rendering will be. I don't think the exact center frequencies of the RGB sources are particularly important except that they somewhat influence the maximum possible CRI that you can get. I've read that with three colors each having a 30 nm bandwidth you can get a CRI of around 80. With five appropriately chosen colors you can get up to 98 but it gets more complex to determine the ratios when you have more colors.

I want a light with a very tight, very hot green hotspot (a la Inova X1 only brighter) and a moderate amount of white spill light surrounding it. Is there anyway to do that?

Thanks.

That should be no problem-just put a few 12,000 or so mcd greens in the center (more if you want an even brighter hotspot). Surrounding them with as many whites as you need. If the whites interfere with the green hot spot bend them to point "outwards" from the center part where the greens are. That should give you the effect you want. In this case you don't want to employ the mixing methods I used, but rather rely on the LEDs' inherent directionality. Using both whites and greens with a narrower (15° or 20°) viewing angle will work the best here.

jtr1962:

That's a good idea thanks. Now, I'm wondering if it's possible to make an LED like those old circular fluorescent lights people used to have in their kitchen ceiling fixtures. /ubbthreads/images/graemlins/thinking.gif

At a LumiLEDs session, about three years ago, they were working on white from four colors, adding in a cyan to RGB, as their current solution is lacking for the reproduction of cyans.