Several spectrometer-measurements of CRI of LEDs are available online such as for LED panels and for various LED bulbs and they show how the CRI value can hide serious issues with colour rendition. Reds and cyans are often the weakest colors (sometimes they are not lit at all, they would appear grey).

Would it be possible to couple a white LED bulb or strip with a properly dimmed red LED bulb/strip to compensate? Cyans would still be weak, but the issue is mostly with reds (and cyans are less common than reds in everyday life, think about skin, lips, blood, ...) and it could improve.

  • 1
    I would think that adding a single wavelength would reduce the CRI of a high CRI LED, since the existing LED's spectrum was balanced to give that CRI. But you could still end up with a spectrum that's more pleasing to the eye, which is what you're probably looking for rather than just trying to optimize the CRI number. – Johnny Jun 2 '16 at 15:44
  • 1
    This question more belongs on electronics.se. Keep in mind that without phosphors, an LED will emit only on one single frequency, so you'll get a vertical "notch" which may be no better, or even worse, than an uneven curve – Harper - Reinstate Monica Jun 2 '16 at 17:14
  • This idea sounds logical... but you may want a few "reds"... between 640nm and 700nm. In other words maybe, LEDs at 640nm, 660nm, and 680nm. – Ben Welborn Jun 2 '16 at 17:16
  • Yes, I'm also worried about Cyan Valley. – Harper - Reinstate Monica Jun 2 '16 at 22:49

OSRAM published an exhaustive document explaining the limitations of CRI and the case with deep red. It also describes the mixing of white LEDs with amber LEDs to obtain a high-CRI (including R9) light with high emission efficiency.



No. Using one source with good CRI is better than combining several poor CRIs. RGB leds are the most exaggerated example: when set to appear white, color rendition is HIDEOUS. That's because their emission peaks don't quite match with human eye sensitivity peaks. And you can never get a perfect match, because there are several types of pigments around our gene pool and even if you match one person's sensitivity, another one may disagree.

You're much better off trying to find one good LED (most likely a multi-phosphor one) or even foregoing LEDs at all and getting a decent FL. Simply speaking, of all home light sources, LEDs are the worst at color rendition.

BTW, you can also try monochromatic light of some nice, warm color. Our brains are quite good at white balance (anyone remember the white/blue dress?), and 100% red or amber light will look actually less upsetting than imperfect white.


Adding in red LEDs (usually in the range of 660-680nm wavelength) at the appropriate brightness will improve color rendering when matched with white LEDs with poor reds. However, the resulting combination will have lower electrical efficiency than just using white LEDs with a better phosphor.

  • Are red LEDs less efficient? controlled in current, not with a resistor to match the input voltage of the white LEDs, shouldn't they be equivalent? Also, if you check the light of high-CRI white LEDs with a spectrometer (my second link), you can see that they STILL lack proper reds. Only one product (the Arcipelago) has a uniformly high CRI with acceptable reds. – FarO Jun 2 '16 at 18:54
  • 1
    @OlafM yeah I'm not comfortable that discretes are less electrically efficient. Phosphors are a huge efficiency loss. Of course cheaper phosphors are an even larger efficiency loss, so maybe he's right in the whole. In any case it is definitely less financially efficient. For a one-off you'll need expensive spectrum analyzers, and for production you'll be driven crazy by the variations in cheap emitters. Cheaper in the end to use quality well-binned emitters. – Harper - Reinstate Monica Jun 2 '16 at 20:24
  • 1
    @OlafM - Yeah, the standard 8-color CRI doesn't test reds effectively. Many high quality LEDs also advertise "R9" values, which is a deep red test case. >50 is an indicator of pretty good red performance (which you can see in your link both the CREE and Hollywood T8s exceed). If that's not sufficient for very color sensitive work, you're probably better off just using sunlight or incandescents – Zhentar Jun 2 '16 at 22:06
  • 1
    @Harper - the problem isn't the LEDs, but rather our eyes; detection of deep reds (>650nm) is quite poor so it takes quite a lot of red light. The broad spectrum of phosphor can emit more of the red light in spectra that we can perceive decently well. – Zhentar Jun 2 '16 at 22:22

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.