Color Temp on CFLs

[Lightmeup]

Most of the cheap CFLs are usually around 2700 degrees. I've noticed that the ones with higher temps usually cost more. What is the best choice for indoor usage? Does it make a lot of difference, are the higher temps better? What do the ones called 'full-spectrum' do that makes them more costly? Isn't a bulb always set to emit one frequency only, so what does full spectrum really mean? Thanks.

[turbodog]

When we moved into our new house last year, I bought an entire house's (is that a word?) of CF bulbs.

I got them at home depot. They were 60w equivalent bulbs that actually used about 14w of power. They were "warm white" 2700k rated.

We hated them. They made everything look brown or yellow-brown.

I got online about 9 months ago and replaced them ALL with some 4200k (or was it 4100k?) CF bulbs. They are great. They are the same color as a "cool white" fluorescent 4' tubular bulb.

I bought them at http://www.lightbulbsdirect.com/

They are 4100k bulbs (I just checked). They are NOT listed on the web site. I have only had 1 go bad (out of over 40), and it was in an enclosed fixture (overheated I think).

They come 4 to a box.

Anything much over 4100k will start to have a strange blue look to it. I have some 6500k in my office and they take a little getting used to. They gave me headaches at first.

[turbodog]

Also, the cost.

I think I paid the same or LESS for these than I did at home depot.

They were worth twice what I paid. Now the colors stand out in our house. You can tell the kitchen is cheerful yellow instead of vomit-yellow.


Just call them on the phone. They made me a deal on an entire case.

[Lightmeup]

Originally Posted by turbodog

Also, the cost.

I think I paid the same or LESS for these than I did at home depot.

They were worth twice what I paid. Now the colors stand out in our house. You can tell the kitchen is cheerful yellow instead of vomit-yellow.


Just call them on the phone. They made me a deal on an entire case.

Do you remember which ones you got and what they cost? I looked at the site, their cheapest ones seemed about 5 or 6 bucks each. This kind of stuff is available on ebay for under two dollars when buying a case. Were yours anywhere near that price?
http://cgi.ebay.com/CASE-of-20-FLUOR...QQcmdZViewItem

Your site had 2700 and 5100 degree bulbs generally available, and I think 4100 was a special order. Do you think 5100 is too high?

[Wim Hertog]

The difference between 2700K, 4100K, 5000K, 6500K and other CFL's is the phosphor mixture inside the tube. Normal CFL's use a triphosphor mixture: a blue, green and red phosphor is mixed to obtain a certain color temperature. By adding more blue and less red, the color temperature raises.

Triphosphor CFL's have good color rendering properties (CRI 80-82) and good lumen maintenance.

The term "Full Spectrum" doesn't mean anything without the color temperature and CRI. Good full spectrum CFL's have a balanced 5000-6000K color temperature and a very high color rendering index (CRI >90). This is possible by adding extra (and expensive) phosphors to the triphosphor mixture. These extra phosphors fill in the gaps in the triphosphor spectrum and raise the CRI.

I'm using 5200K fluorescents and CFL's everywhere..I just can't stand the yellow light of normal 2700K CFL's. If you have a modern house (white walls etc), I would certainly go for 5000K CFL's. If you have lots of wooden furniture I would try something around 3500-4000K. Definitely avoid a fluorescent lamp or CFL under 3500K or above 6200K.

Also, try to find CFL's with a preheat or warm start function: the lifetime is increased drastically by preheating the electrodes of the CFL before ignition!

[turbodog]

Originally Posted by lightmeup

Do you remember which ones you got and what they cost? I looked at the site, their cheapest ones seemed about 5 or 6 bucks each. This kind of stuff is available on ebay for under two dollars when buying a case. Were yours anywhere near that price?
http://cgi.ebay.com/CASE-of-20-FLUOR...QQcmdZViewItem

Your site had 2700 and 5100 degree bulbs generally available, and I think 4100 was a special order. Do you think 5100 is too high?

I say again, call them. There is a lot that is in stock but is not listed on the site. The price? I think it was around 2.95 a bulb. These have been very good bulbs. They are starting right up, and don't appear to need to warm up.

[turbodog]

5000k is not too high probably, but you're gonna mismatch the color temp from your long tube fluorescents. Most 4' long tubes are cool white, and 4100k temp.

[Wim Hertog]

Originally Posted by turbodog

5000k is not too high probably, but you're gonna mismatch the color temp from your long tube fluorescents. Most 4' long tubes are cool white, and 4100k temp.

True.

Maybe you should change all the linear tubes to high CRI 5000K? (H)

[turbodog]

Most of the 4100k bulbs I have are CRI 85. That's pretty good for a non-specialty light.

[brickbat]

Don't conclude that higher CCTs are 'better' because some posters here personally like them. It's purely a subjective thing. It's not unlike asking your buds what color car is best. Some like red, some like blue, just as some like higher CCTs, some like lower.

Also, don't conclude that since noon sunlight has a high CCT, that high CCT lamps are the best for interior lighting. The illuminance levels from interior lighting are an order of magnitude or more below sunlight.

Wim makes an excellent point about the interior colors in your living space having an influence on the 'best' color temp lighting. If you have blues, greens and whites, higher CCTs will enhance them. OTOH, if you have reds, tans, and natural wood, lower CCTs look better, at least to my eye. (In my experience some types of wood pick up an unnatural greenish tone under high CCT fluorescents)

I believe the term 'full spectrum' is marketing hype. (hey, who, in their right mind would want partial spectrum lamps?) They do not, have a continuous power spectral density as their name implies. I've looked at them through a spectroscope. They have big peaks and valleys just like any modern rare-earth (a.k.a. RGB) phosphor fluorescent.

Select your light sources based on how they look to your eye, in your environment. CCT and CRI are important, but in my experience there's no substitute for evaluating the lamp in your living space. Lamps from different manufacturers with identical CCTs and CRIs frequently look different.

As you may have surmised, I’m no fan of high CCT lamps for general interior lighting, but I wouldn’t suggest that my preference for warmer color lamps is because lower CCTs are some how ‘better.’

[jtr1962]

Originally Posted by brickbat

Wim makes an excellent point about the interior colors in your living space having an influence on the 'best' color temp lighting. If you have blues, greens and whites, higher CCTs will enhance them. OTOH, if you have reds, tans, and natural wood, lower CCTs look better, at least to my eye. (In my experience some types of wood pick up an unnatural greenish tone under high CCT fluorescents)

Yes, but also note that he said to definitely avoid a lamp under 3500K (or above 6200K for that matter). The 2700K of common CFLs and incandescents is just too warm regardless of the decor. Things just look too yellow and completely unnatural. There is a limit over which your eye can autocolorbalance. For most people it's probably in the 3500K to 6500K range. Any lower and things just look too yellow. Any higher and they're too blue. The greenish tone you mentioned is just a factor of poorer CRI tubes, particularly cheap halophosphor cool and warm whites. Wood looks just fine under my GE Chroma 50s or my Paralite Maxum 5000s. Oddly enough, it even looks OK under Luxeon and Lamina LEDs, probably because most white LEDs suffer from a slight green deficiency rather than a spike as many fluorescents do.

My guess as to why 2700K CFLs are the most widely available probably has to do more with the manufacturers assuming that people's complaints about fluorescent were mostly because it didn't look like incandescent. Hence they tried to imitate incandescent rather than just finding out the real reasons a lot of people didn't like fluorescents. It turns out that these reasons have little to do with CCT and more to do with poor CRI and flicker. Modern tubes overcome the CRI problem, electronic ballasts overcome the flicker. Interesting how GE's Reveal incandescent is actually preferred by something like 90% of the people who are surveyed. Maybe the CFL makers could learn a thing or two. I'd personallly like to see 3500K offered as "warm white" rather than 2700K, and more options in stores for 4100K and 5000K. It seems the only non-2700K CFLs commonly available in stores are the bluish 6500K daylight.

A lot of color temperature preferences also has to do with what people are used to. People who are used to incandescent lighting often first complain that 4100K or 5000K lighting is too blue or too harsh. After a few weeks most like it better than their old lighting and would never consider going back. No surprise there since eons of evolution means we fell more comfortable under light similar to the CCT of the sun.

[BVH]

I use Sylvania 3500K, 13 watt CFL's and I just love the color. They flank my PC and when its dark/cloudy outside and they're on, its just like being outside in the bright sun and I get a "happy" feeling. Tried a higher temp and didn't like it. Funny, cause I love my 5000K 4' tubes in my office at work.

[Wim Hertog]

Originally Posted by brickbat

I believe the term 'full spectrum' is marketing hype. (hey, who, in their right mind would want partial spectrum lamps?) They do not, have a continuous power spectral density as their name implies. I've looked at them through a spectroscope. They have big peaks and valleys just like any modern rare-earth (a.k.a. RGB) phosphor fluorescent.

Full spectrum is a marketing hype indeed. The problem is that some manufacturers sell really good multiple phosphor, high CRI "full spectrum" fluorescents and some just rebrand a normal triphosphor tube.

There are actually 2 different phosphor mixtures used in high CRI fluorescents:

One is a mixture of three (RGB) phosphors with 2 other rare-earth phosphors to fill in certain valleys in the spectrum. This gives good efficiency and a CRI of 90-95. There are still certain phosphor and mercury spikes in the spectrum.

The other method is to use a very expensive phosphor mixture with particles in it that filter out the mercury spikes in the spectrum. This gives a low efficiency, but a CRI of 98-99.

Below is a comparison of the different phosphors and their spectra in fluorescent lamps.



-> Normal 4500K cool white tube. CRI = 62. Notice the lack of red!



-> 5000K triphosphor tube. CRI = 82. Notice the enormous spike in the green.



-> 5200K Pentaphosphor CFL. CRI = 92. Notice the smaller spikes and the emission at 400-500nm and 620-660nm.



-> 5200K philips graphica pro phosphor. CRI = 98. This is the smoothest possible fluorescent spectrum.

[brickbat]

Nice work, rounding up all those spectra for us to look at. Several observations on my part: I have read elsewhere (sci.engr.lighting?) that a pretty decent high CRI source can be made by blending halophosphates (old cool whites) with incandescent. Makes sense – and probably could be tweeked to offer a PSD curve closer to a blackbody than any other low-cost source. Also, in the last few years, my understanding of CRI has changed. I used to think that a CRI of 80 or 90 was pretty good, but in looking at the spectra, 82 CRI and 92 CRI seem to be a far cry from ideal. Even the 98 CRI lamp, only 2 points from the ideal 100, looks little like a blackbody radiator, or sunlight. It’s not hard to imagine how these sources can taint color perception –particularly with colors that have high reflectance peaks in between the sources’ peaks or where the sources’ peaks don’t match those in the viewers’ eye. The perceived color then becomes highly dependent on the ‘tails’ of the reflectance peak.

Originally Posted by jtr1962

The 2700K of common CFLs and incandescents is just too warm regardless of the decor. Things just look too yellow and completely unnatural.

I’d be a fool to argue with your personal taste in color...

Originally Posted by jtr1962

There is a limit over which your eye can autocolorbalance. For most people it's probably in the 3500K to 6500K range.

Interesting – I’d like to read more about this 3500K limit. Where can I find more details?

Originally Posted by jtr1962

Any lower and things just look too yellow. Any higher and they're too blue.

But, I see the world through my eyes, not yours.

Originally Posted by jtr1962

The greenish tone you mentioned is just a factor of poorer CRI tubes, particularly cheap halophosphor cool and warm whites.

Yep, those are pretty bad, but I note it under SPX41’s too. Maybe I need to switch to a different species of wood. I’m using pine now.

Originally Posted by jtr1962

My guess as to why 2700K CFLs are the most widely available probably has to do more with the manufacturers assuming that people's complaints about fluorescent were mostly because it didn't look like incandescent.

Also, the DOE mandated that ‘energy star’ qualified self-ballasted CFLs be 2700-3000K unless marked with specific wording, such as the term ‘daylight’ that apparently doesn’t help sell lamps...

Originally Posted by jtr1962

Modern tubes overcome the CRI problem

CRI numbers certainly look better for rare earth phosphor lamps, but the PSD is still far from ideal, IMHO.

Originally Posted by jtr1962

Interesting how GE's Reveal incandescent is actually preferred by something like 90% of the people who are surveyed.

In a survey conducted by whom? GE’s Reveal marketing team? Also interesting that GE has been making Neodymium-doped incandescent glass lamps the 1930’s. (they used to market them as ‘daylight lamps.) Took a while, but I guess they’re finally catching on...

Originally Posted by jtr1962

It seems the only non-2700K CFLs commonly available in stores are the bluish 6500K daylight.

Yeah, that’s strange.

Originally Posted by jtr1962

A lot of color temperature preferences also has to do with what people are used to. People who are used to incandescent lighting often first complain that 4100K or 5000K lighting is too blue or too harsh. After a few weeks most like it better than their old lighting and would never consider going back.

Most, maybe. But not me. (“try it, you’ll like it” This sounds a lot like what my parents used to say about skim milk, which I still find abhorrent, and note that by coincidence that it’s BLUE too – maybe I just don’t like blue!)

Originally Posted by jtr1962

No surprise there since eons of evolution means we fell more comfortable under light similar to the CCT of the sun.

The CCT of sunlight in the morning and evening dips well below 3000K, and I happen to like that light and am able to see very well in it.

[jtr1962]

Originally Posted by brickbat

I used to think that a CRI of 80 or 90 was pretty good, but in looking at the spectra, 82 CRI and 92 CRI seem to be a far cry from ideal. Even the 98 CRI lamp, only 2 points from the ideal 100, looks little like a blackbody radiator, or sunlight.

Actually the CRI 98 lamp is pretty close to sunlight outside of a few spikes. There is another measure of how close a lamp is to ideal called the FSI, or full spectral index. Basically, it measures the differences across the entire spectrum from an equal energy spectrum, squares the differences, sums them, and then takes the square root. IMO it's a far better measure than CRI because it also takes color temperature into account whereas CRI does not. An incadescent and the sun at noon both have a CRI of 100 but things hardly look the same under them. It's virtually impossible to differentiate deep blues and violets under incandescent despite the supposed 100 CRI. See this table for some typical FSI values of various light sources. The closer a source is to zero the better by the FSI measure. Daylight at 5500K has an FSI of 0.35. Incandescent has an FSI of 5.3. Surprisingly, the FSI and CRI of GE's Reveal are worse. Many of the more common types of fluorescents such as the SPX don't fare any better by this measure, and many are even worse than incandescent. The 2700K CFLs are among the worst. However, note the low FSI values for some types of full-spectrum fluorescents. GE's Chroma 50, which isn't even marketed as full-spectrum, comes in at 1.8. Some of the other T12 full-spectrum lamps do even better. Interesting how poorly the T-8 full-spectrum lamps do, coming in at 4.7 and 6.2. Also note that CRI isn't correlated at all with FSI. Both the 32W Lumiram Lumichrome 1XX and the 40W Duro-Test Daylight 65 have a CRI of 93. The former has an FSI of 4.7, the latter 1.8. Among the absolute worst sources of light period are sodium vapor streetlights, especially the low-pressure ones. Even the HPS has a FSI of 22 and a CRI of 12. Small wonder the general public can't stand them.

What can we gather here about all this? In summary, it seems most of the common light sources we use are rather poor, whether incandescent or fluorescent. Most seem to have an FSI in the 4.5 to 7 area, making them all more of less equally bad. In short, incandescents are no better or worse than many common fluorescents but since they use more energy they're better off not being used at all. They have no real advantages over most other light sources in terms of quality of light. Also, among the fluorescent sources it was interesting how the worst ones had low CCT. No surprises there. A balanced spectrum, which would have equal distribution among all visible wavelengths, has a CCT of 5457. No surpise then that light sources which deviate very far from that CCT have a poor FSI even if their spectrum is continuous, like an incandescent.

Conclusions? The best possible artificial sources for home/office use seem to be mostly T12 full-spectrum fluorescents which unfortunately have fairly low efficiency compared to common T8 although they are still 3 to 4 times better than incandescent. I would imagine that CRI 98 T-8 fluorescent Wim mentioned would also fare well on the FSI but here again it has lousy efficiency. If color rendering isn't absolutely critical then the more efficient full-spectrum T-8s are "good enough". I use some which have a CRI of 92 and an efficiency of 92 lm/W. I have no idea what the FSI is but I'd hazard a guess that it's in the 4.5 area like the 32W Lumiram Lumichrome 1XX. That's still better than incandescent, and a good 5 to 6 times more efficient.

What should we avoid? Low CCT fluorescents are very poor light sources. The 2700K CFLs should definitely be avoided. The 3500K lamps even seem marginal. If you must have low CCT then it seems that ceramic metal halide is the way to go. They have a 3100K lamp with an FSI of 5.1 and a CRI of 93. More importantly, it has an efficiency of 93 lm/W. In other words, these are very similar to incandescent in terms of rendering but much more efficient. Incandescents? Just avoid them. They really have no compelling advantages in terms of color rendition over the better discharge lamps but have horrible efficiency.

One fascinating thing here is that LED does rather well on the FSI despite having a CRI in the 70s. I've always found LED light rather pleasing and natural despite the supposed low CRI. This shows how misleading CRI is. An SP fluorescent has a similar CRI to an LED yet I find the light from it severely lacking whereas from an LED I don't. All one needs to do to solve this mystery is the look at a typical white LED spectrum. Yes, it has peaks and valleys, most notably peaks at 470 nm and 590 nm with a valley around 550 nm. It's also deficient in deep reds. Despite these drawbacks, it is a smooth spectrum with no sharp spikes like many cheap fluorescents. I imagine better phosphors could fill in the valley at 550 nm and render deep reds better. This would decrease the FSI and increase the CRI, provided the color temperature didn't shift much from 5000K or so. No matter how continuous the spectrum, good FSI values (<2) at low color temperatures are impossible to obtain. The best we can do is around 5 whereas we can get FSI values under 1 at 5000K.

It’s not hard to imagine how these sources can taint color perception –particularly with colors that have high reflectance peaks in between the sources’ peaks or where the sources’ peaks don’t match those in the viewers’ eye. The perceived color then becomes highly dependent on the ‘tails’ of the reflectance peak.

Hence the reason FSI is a better measure of light quality than CRI.

I’d be a fool to argue with your personal taste in color...

Likewise. Just don't invite me to your house. I get a terrible headache under low CCT sources.

Interesting – I’d like to read more about this 3500K limit. Where can I find more details?

Autocolorbalance is actually done by the brain, not the eye. Everyone's brain is different so some people are more biased towards high CCT sources while others are towards low CCT sources. The best I could find to give more details is an article about the Kruithof Curve. In summary, there is a range of CCTs which people find pleasing. Yes, at lower light levels people find lower CCTs pleasing. For example, at full sunlight levels the lower end of the range is 4000K, meaning that almost everybody would agree that <4000K light looks too yellow, and more would consider 5000K to 6000K to be white. However, at typical indoor light levels (let's say 400 lux) the range falls to 3000K to 6000K, with the average being maybe 4500K. In other words, most people would consider cool white (4100K) to be perfect for typical indoor lighting. Those who like somewhat bright interior lighting like me might go for 5000K. Almost nobody would like anything less than about 3500K (note that 3000K and 6000K are the extreme outliers of the range). Amazing then how despite this curve we've standardized on 2700K to 3000K for our interior lighting. Most people just find it too yellow although I'm guessing the standardization was more for practical reasons since decent higher CCT lighting didn't exist until fairly recently. Using 2700K now is probably solely through force of habit more than anything else.

But, I see the world through my eyes, not yours.

Actually, more through your brain than mine. Our eyes probably have a very similar response. It's how our brains handle it that differs.

Also, the DOE mandated that ‘energy star’ qualified self-ballasted CFLs be 2700-3000K unless marked with specific wording, such as the term ‘daylight’ that apparently doesn’t help sell lamps...

Let's hope they don't make the same mistake with LED energy star lamps. Just put the CCT temperature on the package, along with pictures of what it looks like. I think it's interesting how many LED manufacturers are standardizing on 3500K or so as "warm white". Lamina even calls their 4300K LED "warm white". I'm not saying 2700K/3000K shouldn't be made as there are a small number of people who prefer it. However, the number is probably no more than those who like 7000K. I'd rather see a range of choices, or better yet a lamp with user adjustable CCT using RGB LEDs so everyone is happy.

In a survey conducted by whom? GE’s Reveal marketing team? Also interesting that GE has been making Neodymium-doped incandescent glass lamps the 1930’s. (they used to market them as ‘daylight lamps.) Took a while, but I guess they’re finally catching on...

Yes, it was by GE's marketing team, and the results were something like 88% in favor of it. I can't find the link any more. Also interesting how the Reveal has both a lower FSI and CRI than regular incandescent yet it was still preferred. Maybe we need yet another measure but I'd say for now FSI is much better than CRI.

...maybe I just don’t like blue!

That's entirely possible. I personally like blues and greens but I also like yellows and reds as accents (hence my preference for full-spectrum lighting). White is another color I use quite a bit. Maybe that's a reason I just can't stand low CCT lighting since white never really looks like white under it.

The CCT of sunlight in the morning and evening dips well below 3000K, and I happen to like that light and am able to see very well in it.

Actually it dips to ~3000K right at sunrise/sunset, and maybe 3500 to 4000K during the first/last hours of daylight. The sun looks yellower than it really is because it's contrasted against the blue sky. As an experiment, try holding up the whitest incandescent light you have next to the sun. The whitest incandescent flashlights would be roughly 3500K. The sun will look whiter than the flashlight until probably a few minutes before sunset. At sunset it should match a halogen lamp.

I'm thinking you probably fall on an outlier as far as CCT preference. I often happen to be cycling right around sunset. I find that's when the seeing is the worst. Conicidentally, I have the same trouble under incandescent light, especially something like a dimmed chandelier. You'll never catch me in establishments with lighting like that, or at a candlelight dinner.

[NewBie]

As folks get old, their eyes actually have very different responses.

5500K is very white at the age of 18.

By the time a typical person ages to 40, the 5500K source looks like 4000K would have looked at 18.

Also, the gel in your eye yellows with age.

A smooth color spectrum response may not be the ultimate light source, and can cause alot of colors to look bland, with poor saturation. Remember, the eye has three types of receptors, with red response being determined by subtracting the green response from the "yellow" cone. The other two being blue and green cones.

Common phosphors are not very saturated, due to the spectral output not being very pure, and do a poor job of covering the entire gamut of the CIE chart.

You will find that RGB LEDs actually cover more of the CIE chart than tri-band phophors, with smooth spectrum fluorescents being amongst the worst.

Very carefully chosen tri-band phosphors are used in LCDs, some mixes are better than others. Even though the color filters used in LCDs have yet to be optimized for LED backlights, just putting a RGB LED source behind an LCD can result in considerable improvements. Here is an example:
Tri-Band Fluorescent vs. RGB LED

Something to consider....

[jtr1962]

Interesting link about the LED backlighting, NewBie. When I was playing with options for my taxi light project last year, one thing I briefly considered instead of white LEDs was RGB LEDs. I abandoned the idea as unworkable because of the difficulty obtaining a consistent white without some sort of active feedback. However, I did play with mixing RGB LEDs and manually adjusting them to get a decent white. I also compared the quality of light of this source to regular white LEDs. My first impression was "wow, RGB really makes the colors jump out". This was surprising to me, because the 3 narrowband emitters probably left very large gaps in the spectrum yet the resulting light hardly seemed deficient at all. I liked it better than regular white LEDs, and perhaps even more than my full-spectrum fluorescents.

Maybe this is telling us that we shouldn't necessarily be obsessed with obtaining a "blackbody" spectrum so as to duplicate sunlight exactly. It might be that all we need are three or four properly chosen narrowband emitters. The resultant spectrum may have humps and valleys, but it won't have the severe spikes of many fluorescents (which should definitely be avoided).

[Wim Hertog]

Originally Posted by jtr1962

standardizing on 3500K or so as "warm white". Lamina even calls their 4300K LED "warm white".

How is the color rendering of those high CRI lamina LEDs? Any beamshots?

A comparison between the normal white laminas and the 4300K part would be nice. For CRI and color temperature comparison I usually set my digicam's white balance on "sun" and take a picture of my hand under the light. Flesh tones can be very revealing.

[Wim Hertog]

Originally Posted by jtr1962

Maybe this is telling us that we shouldn't necessarily be obsessed with obtaining a "blackbody" spectrum so as to duplicate sunlight exactly. It might be that all we need are three or four properly chosen narrowband emitters. The resultant spectrum may have humps and valleys, but it won't have the severe spikes of many fluorescents (which should definitely be avoided).

I've observed this also when playing with RGB LED combinations. However, I must say that I did experiments with different phosphors for fluorescents and a fluorescent with a smooth spectrum and high CRI definitely looks better to the eye. There is quite a difference between a triphosphor fluorescent lamp of 5200K (CRI 85), a pentaphosphor (5 phosphors) fluorescent of 5200K (CRI 92) and a philips graphica pro of 5200K (CRI 98).

Maybe CRI values are not valid if we use truly monochromatic light sources such as LED's or lasers?

Here's an interesting paper on the color rendering properties of LED light sources: http://www.lrc.rpi.edu/programs/soli.../CRIForLED.pdf

[brickbat]

Wow, much good info here in the last day. Here are the ideas I’m walking away with at this point.

FSI is an interesting concept, but since it uses an equal energy/nm reference (akin to acoustic white noise), I’m not sure what practical significance to give it. The reference CCT is stated to be 5457K, and anything differing from that can not have a perfect FSI. Interesting concept. Seems like a good tool for selling 5100K true full spectrum lamps...

I am a believer in the validity of the Kruithrof curve, since it appears to have been based on data gathered from real observers on what they considered ‘pleasing.’ And to me, that is what is generally most important for lighting in my home.

So, thanks for digging up the Kruithrof curves. I come to a different conclusion than you do, JTR, however. For example, at my desk now, I illuminate the room and work surface with about 30W of CFL-based lighting. It’s a small room and the light sources are generally illuminating most surfaces through indirect means, the average illuminance is about 150 lx. At that level, the curve shows the ‘sweet spot’ to be about 2700K to 3500K. (hard to be exact, but it’s certainly well under 4000K). So, I don’t see evidence that my preference is outside the norm, as suggested. I use 3000K CFLs and it looks good to me. Standard incandescent looked good to me too, but I’m a cheapskate, so I tolerate the worse CRI of the CFLs.

Maybe your preference for higher CCT lamps is mainly contingent on a preference for higher illuminance levels?

I have two rooms where we have Philips CMH 3000K lamps for task lighting. Both use the PAR style lamps shining straight up to a white ceiling, and the light is very pleasing – In the daytime, where it is supplemented by natural daylight, it’s a tad pinkish, at night without a daylight comparison – it’s white. Very white. Not sure I recommend CMH lamps for everybody, though – the warm-up and restrike time is a problem that many people wouldn’t tolerate.

I checked DOE’s energy star site today, and see that a new spec for CFLs is in draft stages. Lamps of 2700, 3000, 3500, 4100, etc are being listed as acceptable, with the requirement for ‘daylight’ on the label being dropped. Good news for all lovers of choice...

http://www.energystar.gov/index.cfm?...ions.cfls_spec

[jtr1962]

Originally Posted by brickbat

The reference CCT is stated to be 5457K, and anything differing from that can not have a perfect FSI. Interesting concept.

True. I didn't say it was a perfect measure, but I think it's a lot better than CRI which we both agree can be misleading. Based on some of the work I've done with RGB LEDs I'm starting to think we need a whole new measure, perhaps measure the difference of a light source from the combined response of the three types of color receptors in our eyes.

For example, at my desk now, I illuminate the room and work surface with about 30W of CFL-based lighting. It’s a small room and the light sources are generally illuminating most surfaces through indirect means, the average illuminance is about 150 lx. At that level, the curve shows the ‘sweet spot’ to be about 2700K to 3500K. (hard to be exact, but it’s certainly well under 4000K).

Yep, it's a little hard to read since it's a logarithmic scale but at 150 lux the range runs from ~2600K to ~3600K, with the average being 3100K. 2700K then is within the range but on one end so you're close to an outlier if that's what you prefer. 3000K on the other hand seems pretty close to the norm at those light levels.

Maybe your preference for higher CCT lamps is mainly contingent on a preference for higher illuminance levels?

That's part of it. To me 150 lux is cave-like, and I'm of the opinion that most people light their homes too dimly simply because they use incandescents and it would consume far too much power to light them decently. After all, to light at the same levels as even my bedroom, never mind my workroom, you would need anywhere from 600 to 1000 watts of incandescent, depending upon what size bulbs you use. On the other hand, when you light with fluorescent, especially the more efficient T8 linear tubes, it becomes perfect feasible to light a room to 1000 lux or better. 500 lux is the lower end of what I consider acceptable, 1000 to 1500 is what I usually use. I would actually light to 10000 lux or more but even with flourescents the power consumption would be ridiculous. At 500 lux the acceptable range is starts at 3000K and goes off the chart, but it looks like 5000K is average. At 1500 lux probably 5500K is average.

Notice how the acceptable range widens at higher light levels. I thought that was interesting. One thing is that even at low light levels, I still don't like low color temperatures despite the curve. For example, at 10 lux the "pleasing" range begins and ends well under 2000K, in other words candlelight. I positively cannot stand a room lit with candlelight. On the other hand, a room very dimly lit with LEDs seems quite natural, sort of like being out at night under the stars. I wonder then if Kruithof missed something at very low light levels or if I just have a strong aversion to any light much differing from natural light (either sunlight or starlight). Also interesting how the scotopic response at low light levels is actually biased towards higher CCTs (i.e. the peak shifts from yellow-green to cyan).

I have two rooms where we have Philips CMH 3000K lamps for task lighting.

Excellent choice. Are those readily available, and do they have self-ballasted screw in versions? I'd like to try CMH but maybe 4000K. The light quality according to both FSI and CRI seems to be superior to even the best full-spectrum fluorescents.

I checked DOE’s energy star site today, and see that a new spec for CFLs is in draft stages. Lamps of 2700, 3000, 3500, 4100, etc are being listed as acceptable, with the requirement for ‘daylight’ on the label being dropped. Good news for all lovers of choice...

Finally, choice! It's about time. I guess I'll get my wish when LED replacements come out and be able to get the CCT I want instead of 2700K being the only readily available choice. It should be interesting to see where the general public's preferences lie once choice is available. I'd hazard a guess that 3500K and 4100K will both be big sellers, 5000K will sell somewhat well but in fewer numbers. 3000K might sell about the same as 5000K, and the worst sellers will be 2700K and 6500K. Also, if LED makes lighting at higher levels economically feasible then I expect more people will do so, and that will shift the most popular CCT upwards.

[jtr1962]

Originally Posted by Wim Hertog

How is the color rendering of those high CRI lamina LEDs? Any beamshots?

I never bought any because the taxi light project was very cost sensitive and the 4300K part gave less light and cost more. If I do buy one or two for kicks later on I'll be sure to post a beamshot.

[NewBie]

Originally Posted by jtr1962

Interesting link about the LED backlighting, NewBie. When I was playing with options for my taxi light project last year, one thing I briefly considered instead of white LEDs was RGB LEDs. I abandoned the idea as unworkable because of the difficulty obtaining a consistent white without some sort of active feedback. However, I did play with mixing RGB LEDs and manually adjusting them to get a decent white. I also compared the quality of light of this source to regular white LEDs. My first impression was "wow, RGB really makes the colors jump out". This was surprising to me, because the 3 narrowband emitters probably left very large gaps in the spectrum yet the resulting light hardly seemed deficient at all. I liked it better than regular white LEDs, and perhaps even more than my full-spectrum fluorescents.

Maybe this is telling us that we shouldn't necessarily be obsessed with obtaining a "blackbody" spectrum so as to duplicate sunlight exactly. It might be that all we need are three or four properly chosen narrowband emitters. The resultant spectrum may have humps and valleys, but it won't have the severe spikes of many fluorescents (which should definitely be avoided).

IMHO, the RGB spectrally pure sources really merit alot of thought and very serious consideration. Especially if the wavelengths chosen are carefully matched to the individual cone responses. Even further improvements can be had, just by picking the right wavelengths for the LEDs, instead of just ordering whatever red, green, blue LEDs happen to arrive at the doorstep...

Yeah, you betcha, the colors seriously stand out with RGB, nearly reaching out and grabbing your eyes, much more than white LED, multiple bin whites, fluorescent, HID. You are taking advantage of the eye response with RGB....

[brickbat]

Originally Posted by jtr1962

we need a whole new measure, perhaps measure the difference of a light source from the combined response of the three types of color receptors in our eyes.

The scheme being used to bin LEDs seems reasonable - essentially grouping sources into regions on the CIE chromaticity chart. probably too complex for the average Joe, though, so I doubt we'll see it on consumer products like CFLs.

Originally Posted by jtr1962

Excellent choice. Are those readily available, and do they have self-ballasted screw in versions? I'd like to try CMH but maybe 4000K. The light quality according to both FSI and CRI seems to be superior to even the best full-spectrum fluorescents.

You can find them (CMH lamps and ballasts) at most commercial lighting distributors, but a better bet, if you're not in a hurry, is to troll eBay for a while. (search tip - look for "mastercol", that way you'll pick up both spellings of the Philips trademark) and electronic ballasts show up there from time to time. Mine are from Aromat, and I'd recommend them. AFAIK, there are no self-ballasted CMH lamps yet. Nor does any manufacturer make a 4000K lamp in a PAR bulb, but the ED17 bulbs are common and I think the Philips Mastercolor 4000K has a very high CRI - well over 90.

[NewBie]

I've seen press releases on RGB MR-16, press releases on RGB in the standard 60 Watt bulb format, and plenty of controllable RGB light strips, and many are available for purchase now. An example:
http://www.para.com.tw/pdf/J&K-EP/K1...-A14XXB5-1.pdf

There is already a RGB CFL package with controllable brightness and color temperature, I just can't find the press release. It doesn't have a really high lumen output, but its just a matter of time until folks get these high efficiency parts that are now available, worked into the package. And there are plently of manufacturers that have 70-100 lm/W LEDs that are working their way to the production stage...

I'm just waiting for this stuff to become a common item you can buy at Wal-Mart...

[pedalinbob]

This is an incredible conversation...GREAT info!

I replaced 3 bulbs in the basement with spirals...I think they were GE. Though they take a minute to brightne up, the color is pretty nice. Almost every basement light is now fluorescent.

I wanted to leave on a light at night in the kitchen, and the bulb had to be small. I purchased a Lights of America candlabra-looking fluorescent bulb. 7 watts...2700k and CRI 82.

Ouch. It basically hurts my eyes, and makes everything look strange.

Based upon the info provided, I will look for something between 3500k and 6000k, with a CRI as high as possible.

Bob

[James S]

Just to resurrect this thread for a moment. I've been experimenting I've got some 3000k phillips lights, that are significantly different from the 2700 lights that I used to buy from commercial electric or GE. And I think I really like them, at least where it's bright enough. In lower light situations my personal choice is the 2700. For example, in the living room in the evening when we're not in there and just pass through on the way to other places I have a single 60 watt equivalent in there. When it's a 2700k bulb it just looks right, a dim 3000k bulb in that position looks horribly strange to me.

However, in the array in the kitchen where I have 8 cans with 100 watt equivalents in them, the newer 3000k lights look much better. I've experimented with higher valued bulbs, but I just dont like them. 3000 looks white to me, but 4000 starts to look blue. I dont like it.

Just my personal preference Yours may vary, so experiment. It doesn't cost much to order several small bulbs to see what you think of the different color temperatures when you're going to order a case of them anyway. Better to waste $50 than to waste $500!

[snakebite]

if anyone sees some 5000k 13w 4 pin g24 cfl with high cri i need some.
no major mfr makes any i can find.
i use 950 tubes in my home workshop and i can now read resistor color codes without errors and dont have to go outside.
i have a lot of 12v cfl ballasts in converted fixtures that use these 13w quad tube 4 pin tubes and i woult like everything to match.

[Wim Hertog]

Originally Posted by snakebite

if anyone sees some 5000k 13w 4 pin g24 cfl with high cri i need some.
no major mfr makes any i can find.
i use 950 tubes in my home workshop and i can now read resistor color codes without errors and dont have to go outside.
i have a lot of 12v cfl ballasts in converted fixtures that use these 13w quad tube 4 pin tubes and i woult like everything to match.

I don't think there is a manufacturer who produces those....the problem is the low demand. Here in Europe, there is a quite large demand for 11W and 25W high CRI, 5000K+ CFL's so that's what we manufacture. We'd love to be able to design and mass produce all those other speciality lamps, but the demand is just too low.

Philips and Osram have the 13W G24 in 827, 830 and 840 only.

[jtr1962]

Originally Posted by Wim Hertog

Here in Europe, there is a quite large demand for 11W and 25W high CRI, 5000K+ CFL's so that's what we manufacture. We'd love to be able to design and mass produce all those other speciality lamps, but the demand is just too low.

Just out of curiosity, is high CRI 5000K starting to become the preferred lighting for both residential and commercial in Europe? It sure seems that way if there's a large demand for the CFLs you mentioned. I know for a fact that 5000K caught on big in Japan. I'm just wondering if its popularity is spreading West. To me anyway it seems to be just about perfect.