Here is my homemade Sunlike 3V flashlight. I got an existing cheap 1W lead acid battery powered flashlight with 80 CRI LED. I just thought of replacing it with Sunlike 6500K to have a high CRI flashlight.
Photometric data from my OPPLE G2 is 7612K @ -0.0084 dUV. The thing that confuses me is that a -0.0084 dUV Sunlike 6500K looks exactly like Actual Daylight from the window compared to a -0.005 or -0.006 Sunlike 6500K . A Sunlike 6500K at -0.005 to -0.006 dUV looks slightly “Yellow Green” compared to Actual Daylight.
20mm MCPCB is compatible with the reflector so all I need to do is just direct replacement. I reflowed two Sunlike 6500K side by side since this is a low power LED at 0.2W nominal and 0.5W maximum)
When you compare against daylight, what illuminated surface are you comparing on, and what is the composition of daylight (direct sunlight, overcast sunlight, blue sky, etc.)? I’m trying to figure out why the extremely low duv seems natural in comparison.
It is quite a royal pain to figure out an optical system that eliminates the multi emitter artifacts while maintaining efficient transmission.
the illuminated surface i am comparing is a white t-shirt. i used white t-shirt this time because both real sunlight and Sunlike have ability to generate blue-fluorescence from white fabric. if I am comparing Bridgelux Thrive to daylight, i would use white styrofoam to exclude blue fluorescence.
Regarding the daylight source, it is a indirect sunlight (overcast) coming from the window. it is mostly skylight with some slight reflection from concrete or vegetation. The actual daylight dUV in my home ranges from +0.005 up to +0.01. Even at +0.01 I really can’t see green tint from real sunlight, only white, bluish white or maybe to some extent bluish-greenish white.
the thing that makes comparing a headache is because of the “chromatic adaption” or “auto white balance” function of the human eye. So most of the time, I would get blind from the real tint of the light source and see it as neutral. The only way not to get blind sided is to compare two light sources side-by-side at the same time, and viewing them simultaneously using a light separator of some sort.
I think I got an explanation as to why a Sunlike 6500K with an extremely low duv (-0.0084) seems more closer to daylight compared to a Sunlike 6500K at -0.005. The thing that makes the Sunlike 6500K dUV -0.005 look a bit off from real daylight, is its “greenish tint”, but if you increase the “intensity” of light, this greenish tint sort of gets minimized. The Sunlike 6500K with -0.0084 has lower greenish tint and it actually looks like white with sort of light orange-yellow tint which to my eye looks more neutral and closer to daylight.
From my personal observation, Actual daylight 6500K is whiter and bluer than Sunlike at 7,200K to 7,400K at -0.0084 dUV (Sunlike actually looks more Yellow compared to daylight despite the higher CCT)
also Actual 5700K Sunlight at +0.01 dUV looks neutral-bluish white compared to 5600K Sunlike at -0.005 to -0.008 ( Sunlike actually looks Light Orange compared to real Sunlight at 5700K ).
Bridgelux Thrive at 6200K with -0.002 to -005 looks like Actual Sunlight at 5400K at +0001 to +0.01 dUV
(1) I am worried that the SunLike induces fluorescence to nowhere near the same amount compared to overcast daylight. Even compared to, say, the Optisolis, the SunLike has a lower violet spike. Furthermore, I wear photochromic lenses, which readily turn dark in daylight but does so much slower when illuminated by SunLike 5700K at point-blank range.
(2) It is possible that daylight is an ill-defined reference. Things like background (vegetation and concrete as you pointed out), cloud type, air quality, time of day, etc., can all significantly impact not just the duv but other aspects of the spectrum. Very cold daylight has a great proportion of its emission in the violet+UV range, which induces a significant change in spectrum due to fluorescence.
If you are lucky, you might be able to find some toilet paper or other white surface (check with UV light) that does not fluoresce.
I did a re-test today and I followed your suggestion using toiled paper but we don’t have one, so i used Kitchen paper towel. I tested it first to make sure there is No UV fluorescence by exposing it to Marktechopto 400nm LED. In my area, Daylight never goes 0.000 dUV it is usually +0.005 to +0.01 on average. occasionally, +0.002 and as high as +0.02 but it never goes zero or negative. I can usually get 6500K and higher around late 5:00PM or earlier than 6:00AM. Lightmaster G2 Pointed directly on the sky would get 6500K at +0.005. I’m not interested on lighting below 6000K because of poor color contrast, although it is more pleasing to the eye.
So here are my observation:
Sunlike 6500K ( 7612K at -0.0084 dUV ) - using a side by side reference daylight, to my eye, this looks like a daylight source at 5500K with +0.005 dUV. It is actually “orange-yellow” compared to an actual 7600K at +0.001 to +0.01 dUV which is bluer and whiter.
Sunlike 6500K at ( 7423K at -0.0060 dUV) - side by side comparison with actual daylight, to my eye, it actually looks like 5500K at +0.02 dUV daylight (yes we have an actual window light at +0.02 dUV) . This one looks “yellow-green” compared to an actual 7400K at +0.001 to +0.01 dUV daylight in which again is whiter and bluer.
From my observation, there is a difference in perception compared to what is being displayed by the instrument (lightmaster G2).
For Example: If we compared two light sources A and B. Light source B would appear yellow or orange compared to Light source A, even if B has higher CCT at 7400K. CCT alone can’t be the one that defines the yellow-blue property of a light source, dUV also affects the yellow-blue color of an LED. Likewise, a higher dUV would also make an opposite effect, shifting the color of the LED to be blue-green.
Light source A: 6500K at +0.001 dUV
Light source B: 7400K at -0.008 dUV
There is definitely something wrong with the equipment, since the SunLike seems 2000K warmer than the measuring device indicates. This is a point that I have previously overlooked–I remember seeing the same device testing a 519A and reading a CRI of 100, so its numbers are likely far from reliable for even higher CRI sources than 519A.
I would imagine the error to be particularly egregious toward the upper end of CCT, since as CCT goes up, it takes less perturbation (in fact, an arbitrarily small perturbation) to a spectrum (measured by Euclidean distance on the CIE 1931 diagram) to produce the same change in CCT.
either it’s the Lightmaster G2 tha’s wrong or it is my eye. The G2 suggest that the Sunlike 6500K is at 7000+K CCT and -006 to -0.008 dUV but my eye suggests that it is at 5400K - 5500K at +0.02 dUV .
today, i have an overcast rainy sky at 6200K +0.0027dUV. Sunlike 6500K measured by G2 at 7000+K with dUV -0.006 to -0.0084 still looks slightly yellowish greensih compared to overcast sky on white fabric target. But if I used a Kithen paper towel target (no uv fluorescence), overcast daylight and Sunlike looks very close it’s hard to tell which one is overcast daylight and which one is Sunlike.
I wonder if there are ways to detect whether/how the device fails to register the correct CCT/duv. Like starting with an incan or sunlight (reference spectrum) and adding blue to it and see how the CCT/duv/CRI changes. Or mixing daylight and SunLike and see if the combined spectrum has CCT/duv that measures between the two extremes.
Kitchen paper (no fluorescence) is a better target for these comparisons. As mentioned before, overcast daylight emits a lot more UV proportionally than SunLike-type LEDs (which were engineered to minimize UV to avoid damage in art galleries). I could have a room that is barely lit by daylight, and all the glow in the dark items would be glowing obviously.
Another interesting test is to get readings for direct sunlight/daylight, and then get the reading for daylight bounced off white fabric. This way you could tell how much fluorescence throws off the spectrum.
So now I have a Sekonic C-800 spectrometer. The dUV reading of the Lightmaster G2 is different from Sekonic Spectrometer. The Lightmaster G2 CCT reading is off by 400K and the duV is off by around 0.007.
You are also right when you said to me that i should “trust my eye” when I said that the Sunlike looks yellowish-green while the Lightmaster G2 is saying it is -0.006 to -0.0084 ( magenta ). The Sekonic spectrometer confirms what my eye says, that the Sunlikes were never that magenta at all.
Not surprised, these Opple devices are now known to be super inaccurate. I’ve been suspicious about them ever since they gave 519A 5700K a 100.0 CRI rating when the emitter first came out, completely failing to detect the spectral irregularities.
I think a trained eye, plus a few reference lights, can be much better at estimating CCT/duv than the Opple.
I learned a lot about light, color and human eye perception by playing around with a Spectrometer. The spectrometer also invalidated some of my previous thoughts, for years i thought 6500K was really bluish white and very white. It turns out, I was just exposed to LOW CRI 6500K for a very very long time that it became my standard for whiteness. In reality, 6500K is only very white if it is LOW CRI 6500K (due to high amounts of blue and low amounts of red, yellow, cyan). In reality, a HIGH CRI 6500K and DAYLIGHT 6500K is “slightly yellow green white” once your eye adapts into it.
So if we put side by side LOW CRI 6500K and HIGH CRI 6500K LEDS, this is my observation:
LOW CRI 6500K - appears whiter than high cri 6500K this is due to BLUE PEAK at 450nm and desaturated red, green and cyan.
HIGH CRI 6500K ( Sunlike, Bridgelux) - would appear yellowish green compared to LOW CRI 6500K, in other words, HIGH CRI 6500K would appear to have lower CCT compared to LOW CRI 6500K.
HIGH CRI 6500K ( blue pumped ) - i only own one which has negative dUV so it appears as “magenta tinted”. but they would always appear “magenta” tinted even if they have + dUV because of the 450nm blue peak. In reality, 450nm is not really blue when seen by a human eye, it is sort of blue-violet.
I first noticed this in 2003, when i first handled 6500K High CRI Bridgelux Thrives. I was expecting to be “bright, clean, bluish white” like low CRI 6500K but they turned out to be “yellowish green”. I thought it was a manufacturing defect. I didn’t have spectrometer back then, and had to rely only OPPLE which says that the CCT is around 6500K with negative dUV. Then i got the Sunlikes later, which has the same appearance, still “yellowish green” compared to LOW CCT 6500K. With the arrival of the Sekonic C-800, i was able to compare that the Bridgelux and Sunlike 6500K were actually very close to actual daylight 6500K which is really “yellowish green” and NOT “bright, clean, bluish white”.
Since I wanted a “bright, clean, bluish white” look of LOW CRI 6500K. I now have to combine other LEDs to HIGH CRI 6500K just to raise the CCT by several hundreds or even thousands then keep the dUV to about +0.0025 to +0.0030. I even went as far as 12,000K ( very good bright, clean white but horrible on skin tone hehe.). As for me, HIGH CRI 6500K (D65 white point) is just too yellowish-green for my taste. I am a bit obsessive in achieving perfect white. Old SONY Television white points even go as far sa D93 at 9300K.
Yeah that is very interesting. Whenever I compare a high and low CRI emitter at the same CCT, the high CRI always appears more “creamy” rather than a pure white. Extremely high CRI sources can often look greenish because (i) the blue spike is cut off, which increases duv, and (ii) the cyan dip is filled in, which also makes things a bit greener.
The fact that blue spike makes things look extra white has been known for some time, that’s how some printer paper achieves reflectance above 100%. Sunlight on printer paper looks extra cool and white, due to the additional broad-spectrum blue emission from UV fluorescence.
That’s an interesting phenomenon. I notice that if I see a large quantity of 450nm, my eyes register is as blue; however, as you turn the brightness down, the blue becomes more and more violet to the eye. I believe this is due to the Bezold-Brucke effect, which I recently learned through writing this paper digest, which may be of independent interest to you.
That is quite an unusual preference! Looks like you want some sort of daylight mixture that assigns sunlight a lower weight, and skylight a higher weight. I don’t know of any broad-spectrum blue LEDs that accomplish this. How did you get your hands on 12,000K?
YES, i prefer Skylight and in my area i have already mapped out my preferred CCT. it’s an indirect light at 8:00AM on a clear blue sky coming from above without direct sun , which on the Sekonic C-800 registers at 12,000K and dUV of 0.003 and 14,000K with dUV of 0.004.
I was able to achieve 12,000K by adding violet and blue LED with Sunlike but it reduces R9 to around 80-85. so additional LEDs are needed to shape the spectrum without sacrificing R9 in exchange for higher CCT.
. A Sunlike 6500K at -0.005 to -0.006 dUV looks slightly “Yellow Green” compared to Actual Daylight.
.