How much energy will we save switching to LED lighting?

Factual Questions
21

Um … I’m not entirely certain what you’re saying :confused: But I assume you’re right in that the RGB LEDs will produce superior color rendering.

Yes, my fluoros are all conventional socket types, except for the stick light over the sink (which is the landlord’s bulb ennyhoo). And most of the fluoro bulbs I’m now using are less than five years old; indeed the color fidelity is greatly improved over the ones I bought ten years ago at $15-20 a pop. :eek: Deo gratias!

HowEVer. The mizzabul things still let me pick out a (very dark) green elastic to go on my braid on a day when I’m wearing (dark) blue clothes. I really can’t tell until I get outdoors. :smack: :frowning: And I have otherwise superior color discrimination.

22

Okay, this cite (warning: PDF) depicts the spectra emitted by the different phosphor chemistries. Note the calcium halophosphate types emit primarily yellow light (superimposed on the spectrum of mercury, this basically yields white by combining blue and yellow, much like white LEDs do) whereas the newer rare earth triphosphor type give off distinct red, green, and blue wavelengths.

Yes, that sounds about right for the difference between them.

Are they the type designed to mimic the color temperature of an incandescent light? The kind that glows with kind of a pinkish yellow tint? I suspect that those are manufactured with very little of the blue phosphor, which would make it difficult to tell dark blue from dark green.

23

So are there consequences for color perception to using RGB lights instead of incandescents? Incandescents produce a (red-heavy, ugly) spectrum, while RGB would produce, I assume, three wavelengths of light and naught else.

Now, if the three colors were balanced well, the light could well end up being indistinguishable from pure white, but I can’t help wondering if having no yellow light (for instance) would have some odd, anomalous effects on our color perception.

24

It might, but I’m no color vision expert. I do know that I’ve yet to see a fluorescent lamp that produces a wholly natural-looking light. These are, functionally, little different from having separate RGB emitters, since the phosphors used in them are really a mixture of red-, blue-, and green-emitting phosphors in various proportions, depending on the intended application. Arguably, the most natural-looking artificial lighting (that is, that closest to appearing like sunlight) are the new HID lamps being used more and more in new vehicles these days. Is it a coincidence that these lamps are full-spectrum? I dunno. Any color vision experts out there?

25

Self taught expert count? :smiley:

Illumination which lacks yellow light has the effect of enhancing the contrast between reds and greens (or magentas and cyans, etc). Tri phosphor lamps are commonly used in places like Kinko’s and Wal Mart, where color is either important to the products they sell or else the goods are in brightly colored packaging. Balancing the colors effectively can in theory produce a natural looking white, at least in the daytime when our eyes function trichromatically.

However, IME, lights that do not produce a full spectrum cause some odd looking chromatic aberration effects, making them and the things they illuminate look not quite right when compared to incandescents or sunlight. So a lamp which produces a wide continuous spectrum will look most natural.

26

It seems to me that it would depend on what you’re illuminating. Suppose, for instance, that I have some hypothetical substance which reflects one narrow wavelength of light in the yellow range, but which absorbs all other wavelengths. If I were to illuminate this substance with incandescent light (including natural sunlight, since the Sun itself is incandescent), that substance would appear yellow. If, however, I illuminated that same substance with a light source consisting of monochromatic red, green, and blue mixed together (this could be separate LEDs, or phosphors in a fluorescent or blue --> white LED), then all three components of the light would be absorbed, and the substance would appear black. In fact, for any two light sources of different spectrum, I could describe a hypothetical substance which would appear different under those light sources. Admittedly, this is an extreme case, but I would not be surprised if there were real-world circumstances where a trained eye could distinguish between RGB and true white color rendering.

27

I wonder… It’s probably more complex than this (in fact I know it is), but how about if we made a light source based on balanced primaries, each of which was tuned so that the peak emission was the same as the peak sensitivity of each of the cone types in the (average) human eye - would it appear perfectly white?

(I can think of two areas where it might not; fluorescent objects might not be excited in the same way thus appearing different and objects that refract or otherwise split the spectrum (like the surface of a CD) might show banding).

28

True, and this is an extreme example of why cutting out the yellow region enhances the contrast between reds and greens. A piece of red construction paper or a mark made with red ink typically will reflect a good deal of yellow light in addition to red, and not much green. Under any RGB type lamp, the eye will be summing up lots of red + a little green instead of lots of red + some yellow + a little green. Similarly, green objects typically reflect quite a bit of yellow and cyan light, and “blue” objects quite a bit of cyan and green.

If it were tuned to the eye’s opsins, then no. Our cone opsins peak at ~555, ~530, and ~420 nm roughly, whereas invariant yellow is 574nm. (If anybody can tell me why invariant yellow is not between the red and green opsin peaks, I would be eternally grateful. :slight_smile: ) Anyway, combining the first three would result in a definite blue-green hue.

Balancing invariant yellow (574nm) and invariant green (507nm), with as close to primary red (perhaps around 670nm), and as close to primary blue (maybe around 420nm) as can be achieved within the spectrum would not only produce a light that appears white but also the color rendering would be superb. (These wavelengths are from memory but a few nm off wouldn’t be all that noticeable.)

29

But hang on a minute… when I say ‘balanced’, I don’t (necessarily)mean 'of equal brightness); what I suppose I was asking is why it wouldn’t be possible to take three primary light sources, each tuned to best stimulate one type of cone, then combine them, tweaking their intensities to make something that we could not distinguish from broad spectrum white?

If it isn’t possible, exactly what would be happening in the eye under ‘real’ white light that wouldn’t be possible with tuned primaries?

30

Or, put another way, why is invariant yellow so important if our eyes are not particularly sensitive to it?

31

According to this diagram, the cones peak at approximately 575, 530, and 420nm, with invariant yellow looking like it’s around 555nm.

32

Absolutely - the red and blue would have to be considerably brighter than the others. I didn’t mean to imply that our eyes are not particularly sensitive to yellow - quite the opposite is true. Guess I misunderstood you about the peak wavelengths. :o Short answer, yes it could be done.

Four wavelengths would be a little better than three because the CIE chromaticity curve is tongue shaped - drawing lines between four points on the spectrum will cover more area than three, and the yellow should “smooth out” the effects of chromatic aberration. If the eye’s focus is not quite perfect (which is usually the case), the red and green components focus in different areas on the retina, and this effect is very noticeable if there are no wavelengths between them.

33

I’ve seen charts before that say that, yet to me 555nm looks very green.

34

I’m no expert either. But it seems to me that HID lamps are bluish, not like sunlight (which is yellowish) at all.

Here’s one expert’s take on HID.

35

One more datapoint: Wikipedia says the cone peaks are at 564, 534, and 420nm. More importantly, it makes the point that the “red” cone isn’t really red. It’s probably better to call them L, M, and S; long, medium, and short wavelength cones.

36

Wow. I didn’t expect to set off this much discussion with a mere whine. :slight_smile: Thanks, everybody! It helps to understand why. It really does.

cityboy916, they don’t have a pinkish tint. They’re just the regular fluoro bulbs from WallyWorld. Or are you talking about a tint to the lit bulb? I don’t think I’d call it pinkish, but it’s not white, exactly. :dubious:

I should add that the dark green (a dark forest green, to be more precise) and dark blue (navy, at brightest) are also indistinguishable by inspection from my black elastics when viewed under the fluoro lights in my bathroom. However, the black ones have a difference in manufacture that lets me reliably tell them from the others.

I dunno if it’s the missing yellow that’s the problem, although it sounds at least halfway plausible. <shrug>

For those who are finding some differences in how they perceive the colors on a web page:

I should remind you that unless you have gone through a laborious matching process, you shouldn’t rely on your monitor displaying colors precisely. IIRC, people who do professional graphics work have to have very good (expensive) monitors, and do color-matching so that the colors they view and select on-screen match the colors that come off the color printers. I’ve never done it, but I read through what had to happen long ago, when I got a discontinued model of a (for the time, about 10 years ago, at work) high end HP color printer. It was intimidating, to say the least.

And then, it’s also possible for a person to have anomalous perceptions of colors. I know that there are minor differences between color as perceived by my left eye and my right. My right is better - or at least more nearly matches the consensus of what any given color is called.

I can distinguish any colors that aren’t both very dark and nearly saturated in the same light levels. That’s why I’m so frustrated with the still poor color rendering with these otherwise vastly improved newer fluoro bulbs.

37

Me too, but some of that might be because most of the other light sources you’re seeing at night are yellowish(tungsten) or orange(sodium).

38

Yeah I did mean the lit bulb, sorry. :slight_smile:

I thought maybe it was the blue that’s missing, but if the green and black are indistinguishable then maybe the light is lacking blue-green. Dunno. I’d have to see the lamp through a diffraction grating and the fabrics in sunlight through a prism to be sure.

Same here, although for me it seems to have to do with being able to distinguish yellow from white in my right fovea but not my left.