How Do You Adjust LED Color Temperature?

We've discussed the need for someone to come up with an inexpensive way to drive LEDs from the power line before (We Need Cheap LED Drivers). And to make them dimmable. Smoothly and reliably. And be inexpensive and make things look right.

If you read the blogs on All LED Lighting regularly, you know that the important issue regarding these (or any) light sources has to do with color temperature and color rendering/CRI. What color is the light, and how does it make stuff look compared to some ideal source?

It's quite straightforward to make an LED lamp with three-color emitters. Then by adjusting the drive current, you can get the full color spectrum. But this adjustment becomes completely arbitrary unless there is some attached calibration device or method.

To tell what the color temperature is, we need an integrated three-color measurement system on a chip (SoC). Making phototransistors or photodiodes is nothing new — they are readily available. All that's needed is a closed loop system to produce a very tight tolerance accuracy spec. Oh — and three separate, good-quality color filters.

Put all this on a chip and — voilà! — you can accurately produce a light source that's the right color. This will address concerns regarding the color temperature of the light source in general. More importantly, it will permit continuous adjustments: As the LEDs age, their output drops. To deal with these age-related variations, you can tweak the drive current based on a control signal (well, actually three control signals) from this color temperature controller IC.

Note that the issue of the CRI is not completely addressed by this solution. But it is a good starting place.

Have you worked on the issues of color temperature and CRI? What solutions did you come up with?

Related posts:

— Brad Albing, Editor-in-Chief, Planet Analog and Integration Nation Circle me on Google+

10 comments on “How Do You Adjust LED Color Temperature?

  1. etnapowers
    December 17, 2013

    The LED Temperature can be characterized, by tuning up the controller of the driving current in order to be sure to have the proper brightness. A semiautomatic circuit to acquire the brightness and / or definition corresponding to a certain current load value would be useful to this scope.

  2. RedDerek
    December 17, 2013

    I was looking at something like this, mind exercise at the time, years ago when I was trying to figure out, “Is the color on my screen true or not?” The thought process was to image a spot on the screen and compare the color hues to a standard – the instrument would need to be calibrated. Then it could be used to apply to printers as well to ensure proper ink/toner is used to get proper colors on the print.

    The concept can apply to almost anything we use to generate an image, or light intensity.

  3. etnapowers
    December 17, 2013

    @RedDerek: interesting point, I think that it is possible to have a LED feedback provided by the output of a sensor which senses the brightness of the LED. In this way the system could be autoregulating.

  4. RedDerek
    December 17, 2013

    The other concept was to have the feedback also allow for brightness adjustment for aged LEDs; since LEDs become dimmer as they are used – especially at high power.

  5. Victor Lorenzo
    December 17, 2013

    @RedDerek, “Is the color on my screen true or not? ” This question not only comes from people looking at the screen. Well, this is embarrassing, I was working (designing and programming) a smartcards personalization machine in a previous occupation and the guy from the laminating machine brought a box full of cards. I needed them for test purposes in the machine I was working on. I wondered how they managed to do it that way, but almost all cards were exhibiting a unique set of colours, amazing! I told the guy, just joking with him, “we don't need to 'personalize' them, they all have a unique 'personality'”. It was due to a failure in the printing machine that was only detectable after lamination (the ink colors use to get altered a bit by the laminating machine heat).

    For avoiding that situation there exist color readers with very precise color sensors (24 bits+ resolution) that come calibrated to panthone color palettes.

  6. eafpres
    December 17, 2013

    Hi Brad–I believe that, mathematically and for physics types, your use of pure R, G, B sensors with narrow filters should allow calculation of the CCT value if you assume you are mixing RGB and they are spectrally very narrow.

    However, I think this would fail in many real instances since there may be significant off band emission.  To calculate CCT of sources with more complex spectra requires measuring the spectra, methinks.  

    I wrote a review of a recent entry in the rapidly falling price cateogry of portable spectrophotometers used for lighting measuremnts.  It is on All LED Lighting here.

    Another problem is if your LED Luminaire uses phosphors.  The resulting emitted light may emit over a wide spectrum, again futzing with your measurement.

    My final remark is that what you propose for adjusting the combination of pure RGB LEDs is good; however in the long-run it is likely that won't be the dominant solution for LED Luminaires used for lighting; that is why there is so much work on phosphors, as well as OLEDs and other stuff.  

  7. Victor Lorenzo
    December 17, 2013

    @Albin, “Note that the issue of the CRI is not completely addressed by this solution. But it is a good starting place

    I agree with you, it is. What do you thing about combining the chip with a light pipe? The pipe could be constructed so only a fraction of the light comes to the sensors, just enough to be able to measure the CRI values and close the loop. The filters could be embedded as part of the pipes or as part of the sensors window.

  8. jvn01
    December 23, 2013

    The other important factor is that of efficiency – although in theory it should be possible to create 'white' light of the desired colour temperature by mixing RGB sources (even after allowing for the influence of non-pure spectra, etc) in reality the brightness level/efficiency achieved by mixing RGB is unlikely to be sufficient for conventional lighting purposes; certainly not when compared to existing white-LED technology. Hence the use of phosphors along with blue LEDs in conventional 'white' power LEDs to bring the lumen level up to the required value, with the inevitable spectral compromises that incurs… 

  9. Brad Albing
    December 31, 2013

    @Victor – the technique with the light pipe could work. But we'd need to do it in an inexpensive manner. It's important to keep the cost down in this application.

  10. Brad Albing
    December 31, 2013

    @eafpres – I overlooked the issue of phosphors – I was thiking in terms of RGB LEDs. You're right — in a lamp that uses phosphors, the spectral output is broad with blue spikes.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.