# Thread: Spectral Graph - Quantitative Tint Comparison!

1. ## Spectral Graph - Quantitative Tint Comparison!

This is something I'm pretty excited about, I'm working on a crude spectrometer that can quantify a comparison of tint! A spectrometer basically takes light and splits it up into the different wavelengths it's composed of, showing you the full spectrum. This lets me see the difference between tints, because I can measure just how much of each wavelength is in the light. A warm tint will have more on the red end of the spectrum, and a cool tint will have more on the blue end.

This is a quick comparison I made to demonstrate this concept, between a Cree XM-L cool white and warm white LED (cool from a V11R and warm from an E03):

White: Cool White XM-L
Red: Warm White XM-L
Yellow: Difference

As you can see, the cool has much more blue, and the warm has much more on the red end.

There is definitely a lot of value to comparative beam shots to show tint differences, and this won't replace those, but this will take out a lot of the subjectivity and variance from things like camera settings, ambient light, monitor settings, etc. This has me pretty excited!

I like it...

3. ## Re: Spectral Graph - Quantitative Tint Comparison!

I like it too! Very good idea and very interesting not only for tint but for colour rendering as well! Now the only thing missing from your reviews is a luminance and candlepower measurement!

4. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by dimak
I like it too! Very good idea and very interesting not only for tint but for colour rendering as well! Now the only thing missing from your reviews is a luminance and candlepower measurement!
Yes, color rendering is definitely something I've had my mind on here! The math is that would go into calculating a CRI from a spectral graph like this looks to be pretty intense, but luckily I like math . As far as my reviews go, I could go ahead and convert my measurements to lumens, but I don't feel confident that I have enough data points to make a reliable conversion factor. Currently my accuracy is in the range ≈8%, which isn't good enough by my standards, so I'm not going to publish it yet. As for candlepower, I think lux seems to be what most people are using now (candlepower seems to have gone out of style), and I'm working on a setup to get precise and accurate lux measurements.

5. ## Re: Spectral Graph - Quantitative Tint Comparison!

+

Interesting

~ please include tint codes or K degrees,... because warm can mean many things

6. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by orbital
+

Interesting

~ please include tint codes or K degrees,... because warm can mean many things
You're right, it certainly can. I'll see if I can look those up. I know I've seen the tint code for the warm Xeno E03, but I'm not sure if Sunwayman has published the tint they use in the V11R. Right now, this is mainly just meant to show off what I can do so far, and get suggestions on possible uses for this kind of data. I'm already thinking of color temperature, tint, and color rendering.

7. ## Re: Spectral Graph - Quantitative Tint Comparison!

What measuring instruments are you using?

8. ## Re: Spectral Graph - Quantitative Tint Comparison!

ah, well.

I'll hop into this thread with a FYI:

3000K High CRI Cree, ________________6500K CW Cree.

Ocean Optics Spectrometer.

Craig

9. Originally Posted by 0dBm
What measuring instruments are you using?
The measuring instrument is an HD webcam, with a DVD for a diffraction grating. Like I said, a crude setup

But it's great for guys like me who don't have an extra \$10k laying around

Sent from my mobile device. Please excuse brevity and typos.

10. Originally Posted by csshih

ah, well.

I'll hop into this thread with a FYI:

3000K High CRI Cree, ________________6500K CW Cree.

Ocean Optics Spectrometer.

Craig

Sent from my mobile device. Please excuse brevity and typos.

11. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by orbital
+

Interesting

~ please include tint codes or K degrees,... because warm can mean many things
From Xeno's website:

1/.Emitter: CREE High-Power LED,XM-L series. 3-CT optional:
XM-L 1B T6,Color Temperature 5000~8300K,CoolWhite,MaxOutput430lms OR
XM-L 4C T5,Color Temperature 3700~5000K,NeutralWhite,MaxOutput400lms OR
XM-L 7B T4,Color Temperature 2600~3700K,WarmWhite,MaxOutput350lms

So the warm is somewhere between 3600K and 3700K. I'll see if I can get something more specific.

12. ## Re: Spectral Graph - Quantitative Tint Comparison!

Ok, according to the Xeno marketplace thread, the warm tint is 3300 K.
http://www.cpfmarketplace.com/mp/sho...=1#post2658684

I've sent a PM to Sunwayman to find out the tint bin of the V11R.

13. ## Re: Spectral Graph - Quantitative Tint Comparison!

Very interesting...I'm still an advocate of warm lights, the only reason I haven't purchased a thrower is because most of them don't have warm versions.

14. ## Re: Spectral Graph - Quantitative Tint Comparison!

Interesting results and a nice idea. But as we can see, your graphs are pretty different from the graphs given in the data sheets, and pretty different from the ones given by csshih too (measured with a real spectrometer). For example, three distinctively separate spikes instead of two. Are you sure about the accuracy? What kind of setup did you use for the webcam? A white box/sphere?

The graphs are clearly different, but they represent the two extremes (3300K and presumably 6000K+). Due to that, I would have expected even bigger difference (especially in the 450nm spike). I believe we might need more accuracy and precision before we could reliably compare, say, two manufacturers with the same CCT.

Anyway, a nice project. Looking forward to see more measurements.

15. ## Re: Spectral Graph - Quantitative Tint Comparison!

jcw122 ^

Throwers are a different breed.
You truly get more punch w/ throwers using cool tints,,, 20% easy.

They just tear through the night sky
>>> This is the only time you'll hear me say anything positive about cool tints.

16. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Bigmac_79
The measuring instrument is an HD webcam, with a DVD for a diffraction grating.
I wonder if that's where the third peak is coming from? There are twice as many green receptors as red or blue. That's why half of all dead pixels on cameras that aren't black are green.

17. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by jcw122
Very interesting...I'm still an advocate of warm lights, the only reason I haven't purchased a thrower is because most of them don't have warm versions.
Originally Posted by orbital
jcw122 ^

Throwers are a different breed.
You truly get more punch w/ throwers using cool tints,,, 20% easy.

They just tear through the night sky
>>> This is the only time you'll hear me say anything positive about cool tints.
Agreed that cool tints have more output, though I've heard people say that warm tints don't catch on fog/dust particles in the air as much. I haven't tried a real comparison to test this. I'd like to see a warm thrower, just to see what it's like.

18. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by AnAppleSnail
I wonder if that's where the third peak is coming from? There are twice as many green receptors as red or blue. That's why half of all dead pixels on cameras that aren't black are green.
Well, the spikes are in the right places, but if the program is at least reasonably good, it shouldn't make that kind of mistakes...

19. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Esko
Interesting results and a nice idea. But as we can see, your graphs are pretty different from the graphs given in the data sheets, and pretty different from the ones given by csshih too (measured with a real spectrometer). For example, three distinctively separate spikes instead of two. Are you sure about the accuracy? What kind of setup did you use for the webcam? A white box/sphere?

The graphs are clearly different, but they represent the two extremes (3300K and presumably 6000K+). Due to that, I would have expected even bigger difference (especially in the 450nm spike). I believe we might need more accuracy and precision before we could reliably compare, say, two manufacturers with the same CCT.

Anyway, a nice project. Looking forward to see more measurements.
In short, no, I have no estimate of the accuracy. As I said at first, this is a crude spectrometer. I just put it together yesterday, and was too excited to wait to show it off . The webcam is currently inside a black box. A white box would reflect the light around too much, when I want the scattered light to be absorbed and not interfere with the new light coming in.

I've done a little fine tuning and now I'm seeing bigger differences than before, but still not as big as I might expect. I think it may be partially due to the nature of measuring the emitters from inside a reflector. I've noticed on several lights that the warmer tints tend to get thrown to the middle of the beam, while the cooler tints tend to get thrown to the outside (the reason for sometimes seeing a blue or pink ring around the edge of a beam). All my measurements were done towards the middle of the beam. Also, while the V11R is definitely cool, it doesn't seem to be extremely cool. It is still pure white, no hit of green or pink or blue (except a little blue around the edge as I mentioned). You can see what I'm talking about in this picture from my review:

Also we have to keep in mind that a certain color temperature refers to a broad range of possible tints. This picture illustrates what I mean:

As you can see, an emitter could be rated at 6000K and be either very green or very pink, which would show up very differently on a spectral graph.

Originally Posted by AnAppleSnail
I wonder if that's where the third peak is coming from? There are twice as many green receptors as red or blue. That's why half of all dead pixels on cameras that aren't black are green.
I had thought of that, and I'm thinking of some way to try to accurately measure the spectral response of my webcam to account for any errors there. However, my webcam in general does a pretty good job of showing things how I see them, so it's response must be pretty close to that of the human eye.

Esko, thanks for that link to the info on camera sensors, I'll read that as soon as I get the chance.

Thanks for all the feedback, you are all being very helpful!

20. ## Re: Spectral Graph - Quantitative Tint Comparison!

Well, you were talking about a 8% accuracy in post #4... I was thinking it might be a white box, because with multiple reflections the possible tint differences in beam would be pretty much eliminated (mixed).

21. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Bigmac_79
no hit of green or pink or blue (except a little blue around the edge as I mentioned). You can see what I'm talking about in this picture from my review:
ahh, I'm pretty sure that's the AR coating reflecting.

Craig

22. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Esko
Well, you were talking about a 8% accuracy in post #4... I was thinking it might be a white box, because with multiple reflections the possible tint differences in beam would be pretty much eliminated (mixed).
Oh, sorry I wasn't specific. The 8% accuracy is for the device I'm using to measure the output levels of lights, which is a completely separate device from the spectrometer I made to make these spectral graphs. I should have been more clear on that.

Yes, for measuring the output level, a white box is what I have, and definitely the best way to go, because you want the light to reflect around as much as possible and even out the intensity of light throughout the box, so that measuring the light intensity at one point gives a good representation of the overall amount of light put out.

For a spectrometer, I need a black box. Light enters through a small slit and shines on a DVD, which has many tiny grooves that act as a diffraction grating. The diffraction grating separates the incoming slight into the component wavelengths, and sends that spectrum into my camera. However, a diffraction grating like this one both transmits and reflects the spectrum at many different angles. I've got my webcam positioned to catch one of the two main ones, but the others are sent back out into the box. Any of that light that bounces back into the camera will throw off my reading, because I only want to measure the spectrum for the original beam. So, I've lined the box in black to make sure the extra light is absorbed, and won't come back to interfere. Even with the box lined in black, I've found that shining a very bright beam in can still overload my box and I get images of reflected spectra overlaid on top of the original. Thankfully that's only a problem with very intense beams (about 500 lumens within one foot of the slit will do it), but I still want to improve my design so I'm looking into possible coatings that will be cheap but absorb more light than black paper.

Your point about evening out the tint differences in the beam is a good one. Because I'm also planning a new output measuring device, I considered building it so that I could attach the spectrometer to it for that very reason. However, I'd need to be very sure that whatever coating I choose for the output measuring device is completely white, so that it doesn't change the tint of the beam at all by absorbing some wavelengths more than others.

I hope that clears things up, but feel free to let me know if it didn't. I apologize again for being ambiguous earlier. Two separate devices

23. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by csshih
ahh, I'm pretty sure that's the AR coating reflecting.

Craig
I hadn't thought of that. I'll have to try removing the lens and see if that blue is still there. Amusing to think this might be a reflection from an "anti-reflective" coating

24. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Bigmac_79
I hadn't thought of that. I'll have to try removing the lens and see if that blue is still there. Amusing to think this might be a reflection from an "anti-reflective" coating
hehe, yeah. the coating is mainly used for reflections perpendicular to the coating - if you view it look at the lenses, skewed, you often see the colors, depending on how thick the coating is.
..I think.

Craig

25. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Bigmac_79
For a spectrometer, I need a black box. Light enters through a small slit and shines on a DVD, which has many tiny grooves that act as a diffraction grating. The diffraction grating separates the incoming slight into the component wavelengths, and sends that spectrum into my camera.
Ok. I thought you had a program (on DVD) that makes the measurements based on white light. Something related to the photography programs (with auto white balance and such) or programs used in meteorology (I believe).

My Verbatim DVDs look pretty violet, Verbatim CDs cyan and movie DVDs have some tone on yellow/gold. Wouldn't be surpriced if the DVD was a source of significant error. Maybe a cheap prism would be better?

26. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Esko
Ok. I thought you had a program (on DVD) that makes the measurements based on white light. Something related to the photography programs (with auto white balance and such) or programs used in meteorology (I believe).

My Verbatim DVDs look pretty violet, Verbatim CDs cyan and movie DVDs have some tone on yellow/gold. Wouldn't be surpriced if the DVD was a source of significant error. Maybe a cheap prism would be better?
The program that I'm using is pretty simple, it basically just takes the spectrum seen by the webcam (light coming through the slit and diffracting through the DVD) and analyzes the intensity at each point.

It's true, DVD's have a certain color tint to them, but most of that comes from the thin reflective coating applied to one piece. A DVD actually has two discs put together, one with a reflective coating and one clear. If you have a spare DVD you can pry it apart with a knife to see what I mean. To my eye, the clear part of the DVD seems to be pretty color neutral. For this, I pried the DVD apart and used a small piece of the clear half.

However, I do think a prism would probably lead to an increase in the quality of my readings. A diffraction grating is limited in it's precision by the density of the grating, and also the actual shape of the grating (right triangles are best). DVD's are pretty good at this, because they are built to use the properties of diffraction to store and access the data written. But with a prism, the spreading of the spectrum is more continuous, not based on small lines. I just don't know if my webcam is sensitive enough to detect the difference though, because to my eye the spectrum through the DVD looks pretty continuous, and I can't see any lines on it.

The other advantage of a prism is that for most wavelengths it has significantly better efficiency. A diffraction grating will reflect and transmit a spectrum in both directions at several angles, the two first order transmitted ones being the strongest, and subsequent ones having decreasing intensity. I'm measuring one of the two strongest ones, but there is still a significant amount of light sent to the others. Using a prism, there would only be a single spectrum made, with nearly the full intensity of the incoming light (if it was a good clear prism with anti-reflective coatings). But again, in my setup, increased efficiency isn't really necessary, because with as bright as our flashlights are, I'm already maxing out my sensor when I put the lights too close to the slit.

All that being said, I would still enjoy testing it with a prism, because until I do this is all just theory . Let me know if you know where a guy can get a good prism for cheap. Maybe there is one in some broken household device I have lying around?

27. ## Re: Spectral Graph - Quantitative Tint Comparison!

This is very interesting, are you going to make a list of lights and their spectrum graph?

28. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by Cataract
This is very interesting, are you going to make a list of lights and their spectrum graph?
My plan, when I am satisfied the device is as good as I can get it for now, is to begin adding a spectral graph to the reviews that I do. As I have time, I'll go back and do spectral graphs for the lights I have reviewed in the past. If there seems to be a significant desire for it, I'll probably make a separate thread dedicated to listing all the spectral graphs I do, so that they could be all in one place for reference.

29. ## Re: Spectral Graph - Quantitative Tint Comparison!

Cool. Very interesting stuff...

30. ## Re: Spectral Graph - Quantitative Tint Comparison!

Originally Posted by csshih
I'll hop into this thread with a FYI:

3000K High CRI Cree, ________________6500K CW Cree.
Do you know which specific chip this "3000K high CRI" is?

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•