Built on Facts

Light Bulbs, Improved.

A quick and simple way to roughly check the calibration of a spectrometer is to point it at the ceiling. Fluorescent lights put out a particular spectrum, and by comparing the colors the spectrometer senses to the colors you know the light emits, you can see if your spectrometer is accurate to a first approximation. I did this very thing yesterday, with the following result:

i-1a1b4477f1e4f0cc79a1ce12538faa2e-spectrum.png

This is about what we expect to see. Fluorescent lighting consists of a relatively discrete set of colors compared to the broad Planck’s law emission of a hot incandescent bulb. The central peak in the florescent spectrum is in the yellow region, the sharp peak to the right is in the orange, the spike near the other end is in the blue, and so forth corresponding to the colors of a rainbow with red -> blue running from right to left. But there are slices of the spectrum that aren’t well-represented. This means that some colors aren’t rendered very well to the eye and accounts for the somewhat “unnatural” feel of fluorescent lighting.

This was a problem for the widespread adoption of CFL bulbs in the home. People tend to prefer the broad spectrum of incandescent bulbs, which looks more like a smooth and vaguely bell-shaped curve centered perhaps around 680-700 nm. Modern CFLs have a more complex internal chemistry which makes their spectrum look more like an incandescent bulb than a traditional fluorescent light.

Personally I’ve never been too impressed with CFLs as a technology. They’re nice and I use them, but their electronics are more complicated and fragile than a bulb can justify, they take time to warm up, and I’m not a fan of the (admittedly slight) mercury risk. As such I’m excited about the rapid progress in LED lighting. LEDs are light-emitting diodes, and until recently their problem was much the same as fluorescent lighting: they only emit in a small slice of the spectrum, and even if you combine red, green, and blue LEDs into one fixture the color rendering is still not so good. All of these problems are rapidly falling to technological advance, in particular the organic-chemistry based OLEDs and comparatively broad-spectrum quantum dot LEDs.

Though unfortunately OLEDs are rather unstable at the moment, in general LEDs are very electronically simple and astonishingly robust. In terms of efficiency and lifetime they could be practically as much of an improvement over CFLs as CFLs were from incandescents. Not only that, but they’re potentially much more versatile in their colors, they’re safe and easy to dispose of, and as a bonus they don’t have the warm-up issues that tends to plague CFLs. Bring ‘em on.

Comments

  1. #1 Johnny Vector
    November 25, 2009

    I’m eagerly awaiting LED lamps bright enough for stage lighting. Imagine all the brightness with only 10% of the power (and heat)! Every community theatre performing in a space with inadequate power feeds will suddenly have brightness limited only by budget and not by the facility breakers! Plus, blackouts so instant that it makes your eyes bounce!

    Probably a few years off, though.

  2. #2 Michael F. Martin
    November 25, 2009

    I was also very excited about the prospects for OLEDs when I was grad student… in 2000. But we’re much further along now than we were then.

    Since then, I’ve gotten more interested in solving the non-technical problems that can keep new technology like OLEDs from reaching consumers.

  3. #3 Eric Lund
    November 25, 2009

    I’ve been using CFLs at home since I bought the house 11 years ago. They have served well, and I haven’t noticed the warmup issues Matt mentions–they are on within a fraction of a second. (Matt may have been thinking of older generation fluorescent lights, which did have warmup issues). But OLED technology would be even better, especially if they could be made suitable for outdoor use (CFLs are a bit larger than incandescent bulbs of similar brightness, which is a problem for installing them in outdoor light enclosures).

  4. #4 Chase Grafton
    November 25, 2009

    Matt, I really enjoy your blog. I’ve read everything from the original site to today’s post. Have you considered doing worked problems at the beginning graduate level? And what happened to the (text)book reviews?

  5. #5 Paul Murray
    November 25, 2009

    An effect of the weird fluorescent light spectrum is that colors change oddly, compared to daylight, as the peaks in the light source hit or fail to hit the absorption spectrum of whatever you are looking at.

    Paintings look very different, to a bee.

    BTW – I saw Massive Attack a couple of years back, and I’m pretty sure their stage show involved a background that was LEDs. It was plenty bright – although probably not bright enough for general lighting.

  6. #6 Matt Springer
    November 25, 2009

    Honestly I feel I’ve been a bit delinquent in the recent lack of worked problems at any level. They just take a lot of time to do, and that’s been in short supply over the last few months. Grad level problems are especially difficult because it’s tough to thread the needle between “comprehensible to non-specialists” and “nontrivial to specialists”. I will try to be more efficient in bringing that kind of post (both graduate and non) back more frequently.

    Book reviews I can bring back too. Of late I’ve become a big fan of Fetter and Walecka’s classical mechanics textbook, and I’ve been meaning to write that one up for a while.

  7. #7 _Arthur
    November 25, 2009

    I remember when a printer sent samples of a paints color chart before printing 10,000 copies, and when we checked individual colors against the paint samples, THEY DIDN’T MATCH, although the printer had used optic tools to obtain a perfect match.

    We phoned the printer, and he told us to go outside, and compare again. We did. Under sunlight, the color chart matched the paint samples perfectly.

  8. #8 CCPhysicist
    November 25, 2009

    Could you point your spectrometer at a CFL and an incandescent? That would be really useful as a teaching tool. It would also show (within the limits of your spectrometer) how much incandescent energy is in the IR.

    I know what you mean about the warmup issue. The CFL bulbs we use come on quickly (there is a short delay) but have essentially their full brightness right away. My parents have some that remain very dim for many seconds and don’t reach full brightness for as much as a minute or more.

  9. #9 Uncle Al
    November 26, 2009

    they’re safe and easy to dispose of Gallium, indium, arsenic? Enviro-whiner Luddites will go ballistic, moving their lips as they slowly read the contents. The epoxy encapsulation contains bisphenol-A, potentially transforming babies into oozing putrescence. Save Our Children from parts-per-trillion Enviro-contamination!

  10. #10 MRW
    November 26, 2009

    “It would also show (within the limits of your spectrometer) how much incandescent energy is in the IR.”
    Those limits are very important. Both the diffraction grating and the CCD detector will skew the results significantly. If Matt’s spectrometer is the brand/model that I think it is, using it to look at the spectrum of an incandescent light would lead you to the false conclusion that most of the emitted light is visible.

  11. #11 Andre
    November 26, 2009

    The mercury issue is one that really irks me. The only reason most people complain about mercury is because they don’t like the CFL bulbs a priori and are just using mercury as an excuse.

    LCD tvs and monitors contain mercury in them, but we don’t see the same health concern from people. The reason? People want cool new LCD tvs, but don’t want to be told that they must buy a specific lightbulb. Plus we know people break these screens all the time by throwing their Wiimotes into them (according to the internets).

    If people find they don’t like LED lights for some reason, they will complain about the chemicals in them too (arsenic in particular).

  12. #12 CCPhysicist
    November 27, 2009

    @10: I assumed as much, but since the spectrum showed a peak beyond 700 nm, one would expect that there is an efficiency calibration curve that goes at least that far out — and you can sharply cut off the spectrum at whatever point you think the uncertainty in the corrected spectrum lacks significance.

    You also need to compare it to an efficiency curve for our eyes. One thing you see in that spectrum above is that most of the energy is in the middle region where our eyes make the greatest use of it.

  13. #13 MRW
    November 28, 2009

    Ocean optics, as far as I know, doesn’t provide the actual numbers for their efficiency curves, but there are fairly rough scaled and incomplete graphs for both the gratings and the CCDs. I’m making some assumptions about Matt’s setup, but I’ve made made a combined efficiency plot before for what I’m guessing is very close to what he has. In my system, the efficiency was greatest in the green and fell off fairly rapidly in either direction. That makes any comparison of intensities without a calibration not very useful.

    With a bit of work, you could come up with an approximate calibration to correct everything. As I said, I’ve done that for a similar system. At some point, though, I wonder if it loses some pedagogical value. If the point is to teach about detector and grating efficiencies, it would be useful. If the point is to demonstrate that a lot of light is lost in the NIR, I suspect the point will be lost on many students, and they’ll be as likely to remember the uncorrected plot, with it’s clear peak at ~550 nm, as they will the nearly flat line that would probably result from a correction.

  14. #14 MRW
    November 28, 2009

    Edit: Nearly flat is a significant overstatement, but flatter than the uncorrected curve and with no memorable peak.

  15. #15 CCPhysicist
    November 29, 2009

    Every spectrometer we used in nuclear physics experiments had an efficiency calibration done if you wanted to know anything more than energies (e.g. branching ratios), so these remarks are surprising to me. You never look at intensities in your experiments?

    But I really don’t understand the implication that you can’t tell the difference between a thermal source with a color temperature circa 3000 K, a CFL, and a long tube fluorescent — although I can imagine that it might be hard to tell the difference between an uncoated bulb and a “cool white” coated bulb’s spectrum.

  16. #16 Lilian
    November 30, 2009

    Stage lighting skills are quite important and sometimes the key to a show’s success. StageSpot carries a wide range of theatrical lighting from manufactures like Altman, ETC, Lycian, Martin, and Color Kinetics, just to name a few.

  17. #17 MRW
    December 1, 2009

    I really don’t understand the implication that you can’t tell the difference between a thermal source with a color temperature circa 3000 K, a CFL, and a long tube fluorescent

    I apologize if I implied that; I didn’t mean to. You’d certainly be able to tell the difference between fluorescent and incandescent lights. That would be straightforward. With enough effort, you’d even be able to get an idea of how much energy is wasted in the very near IR. All I mean to say is that the effort required to do the latter might be more than it is worth as a teaching tool.

    On the broader question, I’ve rarely been interested in calibrated intensities. I’ve often been interested in how the intensity at a particular wavelength changes or in how the ratio of the intensities of two wavelengths changes. The times when calibrated intensities have been important, I’ve used a more sophisticated spectrometer. (Except one time that I needed only a rough calibration.)

    At the time that Ocean Optics was using the software that it looks like Matt has, they didn’t even have an efficiency curve for their CCDs. The best they could do was point you to a spec sheet that they said was probably similar. Now, at least, they can point you to the spec sheet for the exact detector, but I still wouldn’t trust the calibration for anything beyond a rough correction.

  18. #18 MRW
    December 1, 2009

    As far as just demonstrating the difference between different types of lights in the visible, I’ve always used hand-held spectroscopes, which have the advantage of being a few hundred times cheaper.

  19. #19 Lyle
    December 2, 2009

    Actually since an incandescent is pretty close to a black body radiator, taking its temp and going the the stefan Boltzman equation gives a peak around 10,000 angstroms or near IR. So you could do that with math alone, unless the experiment was a demonstration of the black body law.

  20. #20 Leia
    April 22, 2010

    I wonder if they will come up with a way of creating a better light bulb with using deadly chemicals. (: I really don’t know much about light bulbs cause im only 15, yet im doing research on it with a project I have been constructed to do.

  21. #21 calibration equipment
    February 18, 2011

    LED’s are everywhere now but they still have a long way to go. It is great to know that we are fastly approaching the time when most bulbs will be replaced by LED’s or OLED’s. The biggest improvement (as far as convenience) would be in the stage lighting sector. If you have ever played a full 30-45 minute set on stage as a musician or stage tech, you know how unforgivably hot it gets. As a former percussionist, after the first 3 minutes of playing I would be drenched in sweat just from the intense heat of the lighting venues use. Now the interesting part of this comment is how calibration processes can be used to help alleviate this problem and further advance the spread and adaptation of LED & OLED lights in the public environment. However as with any calibration process you must make sure that the techniques which are being utilized are as simple and easy to comprehend as possible for the integrity and reproduction of the data acquired.

  22. #22 Power ixir
    February 19, 2011

    I assumed as much, but since the spectrum showed a peak beyond 700 nm, one would expect that there is an efficiency calibration curve that goes at least that far out — and you can sharply cut off the spectrum at whatever point you think the uncertainty in the corrected spectrum lacks significance.

    You also need to compare it to an efficiency curve for our eyes. One thing you see in that spectrum above is that most of the energy is in the middle region where our eyes make the greatest use of it.