Review of light quality from 98 CRI Chinese brand

[JoakimFlorence]

I recently purchased a Chinese brand LED A19 (E26 size) retrofit bulb, because I was very curious about the light quality. The product was advertised as having 98 CRI. I decided to get the 3000K version. The online product specifications had a spectral graph image, which suggested that this was not a violet emitter based LED.

I do know that 98 CRI LEDs that are blue emitter based do exist, but are very uncommon. Or at least they were extremely uncommon 5 years ago. I would think it would be extremely unusual to find them actually being used in a retrofit LED bulb. Soraa has sold violet emitter based 97 CRI LEDs, but that is another thing. It's easier to achieve this range of CRI with violet emitters, which are going to focus more on full completion of the indigo-blue-cyan part of the spectrum, allowing the other areas to not be quite as crucial.

The first thing, I opened up the packaging and saw a tiny little brochure in the box. It has an image of a spectral graph of their "daylight" product, which I presume must refer to the 5000K light color option, which I did not buy. The graph obviously depicted a spectrum from a violet emitter based LED.
So I immediately did a test on the 3000K light, a crude test, using a CD as a diffraction grating. Judging by the appearance of the spectrum I'm very confident it is normal blue emitter based rather than violet. I could see a fairly distinct gap between the royal blue and the cyan lines. Not fully dark, but there was definitely a big decrease in light brightness in that area.
So apparently the 5000K option uses a violet emitter while I guess that 2700K and 3000K options use a blue emitter. I'd guess because they are trying to aim for higher efficiency and more light output.

Some might be very skeptical about the claims of 98 CRI, coming from an off-brand product coming from China. Everyone knows all the BS made up things that are so normal in advertisement descriptions for Chinese products. But the little brochure in the package went on and on about CRI and got quite technical. It looks less like they were trying to make their product look good to the gullible public and more like the brochure was written by engineers who have an excellent knowledge of the intricacies of CRI and believe in their product. So I am inclined to believe that the designers actually believe their product is 98 CRI. That still does not necessarily mean it actually is though.

I am pretty familiar with very high CRI LEDs and have several other lights to compare it too, so I decided to try to see what the approximate CRI seems likely to be just judging by how the light looks. While I can't say for sure, I am rather confident the light appears to be at least 96 or 97 CRI. It could perhaps be 98, I do not know. It is definitely 95 or above, I have no doubt about that.

I decided to try to compare the light quality to another 95 CRI downlight I have (3000K). Now in all fairness, I believe that "95 CRI" rating of this downlight is actually just the maximum CRI range, so it's possible it could actually be more like 94.

So here's my assessment of trying to compare the claimed 98 CRI bulb to the 95 CRI downlight.

I can't be entirely certain about this, but I think that the "98 CRI" bulb feels a little more like halogen light than the 95 CRI. I kept being skeptical and doubting myself, so I compared them again and again, in different ways, trying to verify that there actually seemed to be some difference.

The quality of the light from the 95 CRI just doesn't quite feel like exactly fully like real halogen. But the 98 CRI pushes the direction further in the direction of halogen. Maybe not precisely like halogen but it feels like it could be a mix between real halogen and the 95 CRI LED light.

(Violet emitter LEDs have exactly the same sort of feeling, in this particular way which is hard to describe, like halogen or natural daylight)

I tried reading some fine print text under the light and came away with the impression that the black letter words were a little bit more clear under the 98 CRI than the 95 CRI. I find them to be most clear under real halogen/incandescent. (It has to do with the average blue wavelength being a little shorter with LED compared to an incandescent filament, which the eye cannot focus as easily, I believe)

You don't have to believe me. These visual differences are very subtle and I believe the majority of you reading this would not be able to notice it.

But this type of observation does make me more inclined to believe that this light is, or could be, higher than 95 CRI.

The 98 CRI light seems to sort of have the tint of a neodymium glass "Reveal" bulb. It strangely seems to have that feeling. Compared both to the 95 CRI LED, and to halogen.
What I mean is it's just a tiny bit magenta-leaning in tint, rather than slightly greenish-leaning, and the light quality completely lacks any of the faux "yellowish" feel that virtually all [blue emitter] white LEDs tend to have.
I would not personally necessarily consider this a good thing, but I imagine it would be considered a significant plus to many of you.

Next I tried to judge color rendering quality.

Unfortunately, despite supposedly or possibly being 98 CRI, this light does not really make red colors look the most vibrant and warm. Oh certainly, compared with 90 CRI LED light it does, but not compared to the 95 CRI LED.

The specifications provided in the brochure claim an R9 (red saturation) value of 94. While this is considered very high for an LED, it does fall short and is almost a little "moderate" for what we might expect from a super-high CRI white light.

Why did they not go higher? No doubt because there are significant trade-offs in efficiency that start appearing when the R9 gets super high. (assuming the red light is produced from the phosphor)

Apparently they managed to achieve what was, or what they thought was 98 CRI overall, and thought the R9 was high enough that trying to increase it even more wouldn't really matter enough.
I suspect there are also some difficulties getting a good balance to achieve 98 CRI overall for a blue emitter white LED if the R9 starts going to high, since the excessive red would create an imbalance with the deficit of cyan in the spectrum.

But even though red colors don't seem to be quite as brightly and vibrantly saturated, I think the red colors might be a very slightly deeper hue under the 98 CRI, very slightly a tiny bit less orange than under the 95 CRI.

The thing is though, I can't think of many situations where this would really be any benefit, outside of perhaps some fine art situation where precise rendering of the reds are more important and for some reason it might be desired to get a tiny bit more of a truer ruby red hue, or differentiate between red and orange colors just a little bit more. Even then, the benefit would be very incremental.
Or if perhaps if someone had a fake spray-tan and was trying not to emphasise the orange hue on their skin. Many people, I think, have the opposite problem, and appear better under light that makes their skin appear a little more orange and lively saturated.

I don't think I can say this 98 CRI has "superior" color rendering to 95 CRI, even if the light does technically have a higher CRI level.

Even though the red saturation from the supposed 98 CRI appears almost a little inferior to 95 CRI, I do not believe that necessarily demonstrates that it is not 98 CRI (or higher than 95).

I've done plenty of experimentation, composing a spectrum with separate red emitters of all sorts of different wavelengths, both real observed and theoretical experimentation, and what I've found is that more orange-red wavelengths, like 520 or 535nm give more vibrant saturation to skin with a healthy-looking orange glow, whereas when too much of the total light is deeper 550 or 660nm red, skin tones start looking very pale sickly pinkish, and paradoxically lose saturation and start looking more greyish, not what one would expect. I don't need to go into detail of exactly why that is, but the point is that creating a spectrum with a CRI that starts going much higher than maybe 95-97 with more greater coverage into the deep red (usually seen as desirable) might not give the color rendering appearance that you want. So I do not believe this should be taken as an indicator that the CRI cannot be as high as claimed.

It could indeed be 98 CRI. Nothing I have seen disproves that. It could also be possible the real CRI might be more like 96 or 97. I can't say for certain.
It might not necessarily be easy for the engineers to tell whether the light really reaches 98 or is 97, precisely. Especially in China, where companies are going to be less technologically sophisticated and avoiding excessively expensive test methods. I don't think the company is intentionally lying but I doubt they have the exacting test methods of a German company like Osram.

(I do have 96 CRI 4000K Osram emitters to compare in light quality, but of course deeper red coverage in the spectrum is a little less essential to achieve a high CRI value at 4000K compared to 3000K)

It can also be mentioned that for the company's "98 CRI" light quality option, the company only offers 40W and 60W equivalents, nothing higher. The specifications for the 40W equivalent say it uses 5W and 450 lumens.

There was no flicker detectable (not visibly, I did the crude pencil waving back and forth test), and the bulbs appear to be as high quality constructed as mainstream big brands.

edit: I decided to try comparing again to better evaluate.
Comparing the 95 CRI to a standard CRI 3000K LED, the light from the normal LED has an aura that is a lot more "yellowish", whereas the 95 CRI very noticeably feels more "orange", "warm", and "full", by comparison. This isn't really the overall color tint of the light but something about how the light "feels".
Comparing the 98 CRI to 95 CRI, the light from the 98 CRI makes the 95 CRI light look "yellowish" by comparison. While the light from the 98 CRI feels more "orange" and "pink". It may sound strange but I would draw an analogy to the feel of candlelight. (Even though we're of course talking about 3000K, much higher than the color temperature of candle flame)
The color rendering ability of the 98 CRI definitely seems noticeably better in comparison to the 95 CRI, comparing side by side. That being said, if you were just looking at things under the 95 CRI, I don't think you would notice any defecit of color, even for an individual very discerning.
I also did notice that red colors looked a little bit more of a ruby red color under the 98 CRI. It wasn't obvious, and overall red colors look just fine under the 95 CRI, but it's just the 98 CRI enriches them just a tiny bit. The average person probably wouldn't even notice it.

 

[jtr1962]

Good, detailed review. Efficiency seems reasonable at ~90 lm/W. I think the real clue that this is CRI 98, instead of 95, is this observation you made:

"But even though red colors don't seem to be quite as brightly and vibrantly saturated, I think the red colors might be a very slightly deeper hue under the 98 CRI, very slightly a tiny bit less orange than under the 95 CRI."

In the end, light which is higher CRI might not necessarily be more pleasing. It's all a matter of what you're looking at and what you prefer.

I bought some CRI 98 lighting boards a while back. Haven't really had time to play with them much. Mine are 5000K. As you said, deep red coverage is less essential to achieving higher CRI as CCT goes up. My boards are using Nichia Optisolis, which is based on a regular blue LED, not violet.

I think ultra high CRI LEDs probably make more sense at higher CCTs. There's less point trying to duplicate a blackbody if the CCT is so low that the light is going to have a strong yellow bias no matter what. You get maximum color gamut going with high CRI and high CCT. Besides, if you want 100 CRI, low CCT light you already incandescents.

Last thing to mention is almost nobody can tell the difference between CRI 95 and CRI 98. Indeed, for the majority the advantage of anything better than CRI 90 is lost. They just won't notice or care about the subtle differences. The most bang for the buck with LEDs is making sure they don't flicker. A rather large percentage of the population will be negatively affected by that.

 

[JoakimFlorence]

Indeed, for the majority the advantage of anything better than CRI 90 is lost.

It might not make a significant difference to most people, indeed I doubt more than half the population can easily discern the difference (at least consciously), but it makes a difference for me personally. There is a big difference between 90 and 95 CRI. I feel like in the home, 90 is just barely adequate and kind of subpar, 92 might start just being okay and acceptable.
I even find I notice the difference between 90-91CRI bulbs and the better 93CRI bulbs.

I think ultra high CRI LEDs probably make more sense at higher CCTs. There's less point trying to duplicate a blackbody if the CCT is so low that the light is going to have a strong yellow bias no matter what. You get maximum color gamut going with high CRI and high CCT. Besides, if you want 100 CRI, low CCT light you already incandescents.

It depends what CRI range you are talking about, I think. If you're talking about CRI higher than 95, then yes, aiming for a higher CRI probably would have more importance at higher CCT. But if you're talking about CRI lower than 90, then in my opinion CRI has more importance at lower CCT. Because 2700K is really going to look awful and yellowish at 80 CRI, and it will almost be like looking at colors under the light of a sodium vapor lamp, everything seem very yellowish. You really need some higher CRI in this situation.

In the end, light which is higher CRI might not necessarily be more pleasing. It's all a matter of what you're looking at and what you prefer.

This might sound confusing but I would like to make the quick emphasis that I do not believe that is necessarily the case. It is just the case in this particular situation.

With a blue emitter design, I think they might be just a little constrained how much total red light that can be added, while still keeping the neutral white color tint of the overall light, and trying to maintain a balance with the rest of the wavelengths (yellow-green-cyan) so the CRI does not suffer in other areas. There is going to require a trade-off. This probably does not even become an issue until above maybe 93, 94, 95 or 96 CRI (I don't know precisely). The trade-off is do you want full red saturation, even if it might be just a little too orange, or do you want full red hue, so the true reds look more pure in color. For almost any use, so long above whatever this CRI threshold is, the saturation becomes more important, if we had to pick between the two.
(Not to make this even more complicated but as one goes from 80 to 91 CRI, saturation is what primarily increases first at the forefront of importance, then only after that you start worrying more about R9 hue. But then above a certain R9 value maybe we start having to worry more about whether it's going to require throwing off saturation)
You could argue it still fundamentally would hold true for violet emitter phosphors, that open up any possible wavelength coverage, but I think there is less of a trade-off in those situations. For example, I don't think anyone would argue that 3000K true black body spectrum does not do a fantastically suitable color rendering job and could be improved.

There are of course some artificial possible means of allowing more color contrast in 2700K CCT white light, which would not necessarily follow the ideal trying to stick to the shape of black body curve or trying to achieve optimal CRI. But I'm not aware of any LED product that has ever done this, certainly not with the normal use of phosphor formulation. One could argue neodymium glass Reveal bulbs sort of do this to some small degree.
That's probably another long discussion though, a separate topic. I only brought that up because you seemed to be alluding to the concept in your claim that only aiming for high CRI might not really be the most ideal, even though it does not really have to do with exactly what we are talking about.

I think the real clue that this is CRI 98, instead of 95, is this observation you made:

"But even though red colors don't seem to be quite as brightly and vibrantly saturated, I think the red colors might be a very slightly deeper hue under the 98 CRI, very slightly a tiny bit less orange than under the 95 CRI."

The difference is very slight though. That part is almost barely noticeable to me. I really have to strain my perception to be able to notice it.

I mean for normal use, whatever they lost in saturation and "warmth", only the tiniest advantage was gained in improving red hue.
In normal circumstances there are few red things that are even actually pure red enough that it would matter.

If we're starting at 90 CRI, the R9 value matters, but I'm starting to think that, at 3000K, once you are already at maybe 95 or 96 CRI, increasing R9 does not really matter so much and only results in the tiniest incremental improvements.
(This is assuming the light I evaluated really was 98 CRI of course)

 

[jtr1962]

It might not make a significant difference to most people, indeed I doubt more than half the population can easily discern the difference (at least consciously), but it makes a difference for me personally. There is a big difference between 90 and 95 CRI. I feel like in the home, 90 is just barely adequate and kind of subpar, 92 might start just being okay and acceptable.
I even find I notice the difference between 90-91CRI bulbs and the better 93CRI bulbs.

Same for me. Definitely notice the difference going from 90 to 95. Not a major difference, mostly in the saturation of reds.

It depends what CRI range you are talking about, I think. If you're talking about CRI higher than 95, then yes, aiming for a higher CRI probably would have more importance at higher CCT. But if you're talking about CRI lower than 90, then in my opinion CRI has more importance at lower CCT. Because 2700K is really going to look awful and yellowish at 80 CRI, and it will almost be like looking at colors under the light of a sodium vapor lamp, everything seem very yellowish. You really need some higher CRI in this situation.

True. I noticed this also with CFLs. Those were rarely above low 80s CRI. The 5000K ones looked reasonable. The 2700K and 3000K just looked really off, more yellow than like a blackbody at that CCT. The effect is less pronounced with LEDs. Still 80 CRI looks far better at higher CCTs. That said, I personally think 90 CRI should be the bare minimum for most interior lighting, with the excpetion of perhaps utility spaces. At this point the cost in efficiency is pretty minimal.

This might sound confusing but I would like to make the quick emphasis that I do not believe that is necessarily the case. It is just the case in this particular situation.

Yes. It's probably highly dependent upon the CCT AND what you're looking at under the light.

With a blue emitter design, I think they might be just a little constrained how much total red light that can be added, while still keeping the neutral white color tint of the overall light, and trying to maintain a balance with the rest of the wavelengths (yellow-green-cyan) so the CRI does not suffer in other areas. There is going to require a trade-off. This probably does not even become an issue until above maybe 93, 94, 95 or 96 CRI (I don't know precisely). The trade-off is do you want full red saturation, even if it might be just a little too orange, or do you want full red hue, so the true reds look more pure in color. For almost any use, so long above whatever this CRI threshold is, the saturation becomes more important, if we had to pick between the two.

I think one of the constraints is also keep the spectrum close to the Planckian locus. The same tradeoffs apply to fluorescent tubes which are excited by UV. I remember we had some CRI 90 tubes which had a slight pink tinge to them. That was mostly noticeable when looking at the tubes themselves, not the objects lit by them.

There are of course some artificial possible means of allowing more color contrast in 2700K CCT white light, which would not necessarily follow the ideal trying to stick to the shape of black body curve or trying to achieve optimal CRI. But I'm not aware of any LED product that has ever done this, certainly not with the normal use of phosphor formulation. One could argue neodymium glass Reveal bulbs sort of do this to some small degree.

I've played around with combining red, green, amber, and blue LEDs to make white light. By the metric of CRI, the results are often lousy, like CRI 70. However, colors look very saturated. In some uses this might be more important than absolute color accuracy.

I mean for normal use, whatever they lost in saturation and "warmth", only the tiniest advantage was gained in improving red hue.
In normal circumstances there are few red things that are even actually pure red enough that it would matter.

Sort of what I was getting at. At some point there are not many real world gains going beyond a certain point for CRI. I might say for interior lighting that point is roughly around 95 CRI. For critical color matching tasks though you could make a case going to 98 or 99 CRI.

If we're starting at 90 CRI, the R9 value matters, but I'm starting to think that, at 3000K, once you are already at maybe 95, 96, or 97 CRI, increasing R9 does not really matter so much and only results in the tiniest incremental improvements.
(This is assuming the light I evaluated really was 98 CRI of course).

Yep. And it seems lately 95 CRI is becoming more and more readily available for those who want something better than 90 CRI.

 

[JoakimFlorence]

I've played around with combining red, green, amber, and blue LEDs to make white light. By the metric of CRI, the results are often lousy, like CRI 70. However, colors look very saturated. In some uses this might be more important than absolute color accuracy.

Totally off topic but I have some data about that which may be of interest you.
RGB can only attain a CRI value of 67-81, although RGBA can technically achieve a CRI up to 92.
Theoretical limit for the efficacy of red, green, blue and white LEDs
The CRI values in this case are not really the most accurate indicator of its color rendering ability.

85CRI white with a very small amount of R+G added can theoretically attain a CRI of 95 to 96. Wavelength choice of red matters; 650nm red is the optimal wavelength for modifying the lower range of color temperatures, but 615-635nm red for the higher end of color temperatures. The white LED used should ideally not be more than 1000K away from the desired color temperature.