2
\$\begingroup\$

I need to swipe between two white shades from one to another color temperature, let's say from 2700K to 6000K. Can it be done by simply mixing the intensities of two LEDs of respectively 2700 and 6000K? So that half of 2700 and half of 6000 will lead to a full (2700K+6000K)/2 of CCT? I thought about this because I saw many table lamps that are somehow using this technique, or at least they seems to do this, but this does not imply that it is right.

I have also started to think about using a weighted average by adding intermediate temperatures by means of using more than 2 LED types, but if the starting point of this averaging technique is totally wrong, it doesn't make sense.

I also see some formulae from the CIE graph, in which you can find the quantity of each RGB color to obtain certain CCT and viceversa, but this is another story and I think that must not be misleaded with generating a CCT equivalent light from two types of white, which is the subject of this question.

\$\endgroup\$
  • 1
    \$\begingroup\$ Just recently I stumbled over this article, it may help: indiecinemaacademy.com/… \$\endgroup\$ – PlasmaHH Mar 1 '17 at 15:35
  • \$\begingroup\$ Switch a constant current source between the LEDs using PWM. \$\endgroup\$ – skvery Mar 1 '17 at 16:04
  • \$\begingroup\$ in short You will have no colour consistency with this approach and they won't blend unless you are so far away it looks like 1 spot. it will look like of colour off-white with no consistency \$\endgroup\$ – Sunnyskyguy EE75 Apr 22 '17 at 11:29
3
\$\begingroup\$

The problem with CCT has been you have to buy binned LEDs to get the correct mix. But if you use two drivers, white and Red LEDs, then binning is not required.

Use high power cool white and mix with medium power Red.

Use a switching regular for the high current white LEDs. Use a linear for the Red LEDs that require less current.

TI makes a chip just for CCT. Whether you use this chip or not, there is still some good CCT information.

One output of the device is a non-synchronous buck controller which is used to regulate current of high power white LEDs. The other output of the device is a linear regulator controller which is used to regulate current of medium power red LEDs. The TI TPS92660 is used for correlated color temperature (CCT) applications by mixing white LEDs with red LEDs.

LINK: Datasheet TI TPS92660 CCT Driver

TPS92660 Schematic

LINK: Manual for TPS92660 Evaluation Board

TPS92660 Eval Board


RECOMMENDATION: LEDs WITH SMOOTH COLOR MIXING

What I like about these LEDs is the wide viewing angle of 175°.

They are targeted for entertainment lighting applications.

They are manufactured with secondary optics for great color mixing

The white can be pushed with up to 1225mA. The red 700mA.

LINK: LUXEON C Color Line Datasheet

These white are spec'd at 85°C, to compare with an LED spec'd at 25° C, add 10% to these.
enter image description here

There are some additional colors with blue and red that are very unique. These are targeted for horticulture but some are the same as the Color C Line but spec'd with PAR (Photosynthetically Active Radiation). You can add Red with a little bit of Blue. Various rations of red and blue are available.

LINK: LUXEON SunPlus Series Datasheet

The Color C LEDs are tiny but pack a punch. Some of the brightest most efficient LEDs available. Notice they have a lower forward voltage (less watts), spec's at 85, and wide viewing angle. The binned by luminous flux has some much brighter than the "Typical"

In my opinion the most important spec is the Thermal Resistance junction to case. The Red are 2.8° C/W

enter image description here

enter image description here

\$\endgroup\$
  • \$\begingroup\$ good info ty... I thought everyone binned by XY coordinates but I see here some are peak and others dominant wavelength for pure colours. \$\endgroup\$ – Sunnyskyguy EE75 Mar 2 '17 at 14:55
  • \$\begingroup\$ I agree Tjc is critical in addition to application design Tca often much worse. Cree's latest Xlamp XHP50.2 is 1.2 °C/W in a 5050 chip \$\endgroup\$ – Sunnyskyguy EE75 Mar 2 '17 at 15:15
5
\$\begingroup\$

White LED's are typically created with 2 phosphors that absorb some of the intense narrow blue spectrum and convert to many Gaussian distributed longer wavelengths. The phosphor thickness is in microns is difficult to control compared to the huge thickness of the Blue LED. So originally they could only make cool or warm. Now they can control the process to get ranges in between like ANSI white or Neutral White or 4500'K or 5000'K

But this doesn't tell the whole story. It only shows what the apparent colour is to our eyes and not the observed reflectance on pastel colours which defines the CRI index. This because White LEDs are missing a lot of the spectrum.

There is another NIST standard, not adopted widely, called CQS which shows the colour quality for intense saturated reflectance. This makes LED's better than FL tubes so the strong FL lobby industry did not want this standard to be adopted.

In short if you only wanted to cool Warm LEDs, you only need to add sufficient Blue LED Light so the peak blue and Warm phosphors are about the same intensity to get 4000~4500K.

But you can't easily make a cool LED to a warm LED since there are no LEDs with broad spectrum but you can approximate it with a calibrated instrument using RGY LEDs.

Samsung Samsung LED

Cree Cree LED

\$\endgroup\$
  • \$\begingroup\$ Note I think supplier made an error in spectrum of 5000'k and showed 6000'k for this as they are the same and 5000'K should be more balanced,. I have over 20k surplus 5mm 4000~4500K LEDs already in this range if anyone wants to buy them. \$\endgroup\$ – Sunnyskyguy EE75 Mar 1 '17 at 16:18
  • \$\begingroup\$ Most white LED's cannot exceed 82 CRI due to limitations of 2 phosphors , even if perfectly balanced. However tri-phosphor Fluorescent tubes "True Daylight" can reach CRI of 88. \$\endgroup\$ – Sunnyskyguy EE75 Mar 1 '17 at 16:30
  • \$\begingroup\$ There are many CRI 90 LEDs available. Phosphors have come along, the book Light-Emitting Diodes 2nd Edition by E. Fred Schubert covers the topic very well. \$\endgroup\$ – Misunderstood Mar 2 '17 at 6:07
  • \$\begingroup\$ I see that now with "remote" phosphor bulbs. Yet still CRI is a poor metric because CRI is not an indication of light colour, but a measurement of a light source's ability to accurately illuminate different pastel colors and not saturated colours. Skin tones and Reds usually suffer. \$\endgroup\$ – Sunnyskyguy EE75 Mar 2 '17 at 14:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.