Public Lab Research note

Reflected sunlight and a spectrometer improving proposal

by viechdokter | | 1,116 views | 9 comments |

Read more:

I went into the garden today and - believe it or not - the sun was shining. Again. So instead of photographing ants or chasing the dog around the garden I took my spectrometer out and photographed plants and spectra. Here is the first:


A nice white spring flower. And here is its spectrum:


Very bright but not overexposed this time. As mentioned before, when it gets overexposed in green and red the yellow will come out much stronger. But its a pretty smooth spectrum over all colours with a slight dip at 590 nm.

The next "pictures" I took of green leafs of grass:

green_grass_small.jpg sunlight_reflected_off_green_grass.jpg

Great. We have proven that the main colour in a green grass spectrum indeed is green. Probably not a surprise to most of the scientific world, but as long as no one measures it everyone just has to guess or believe it... We have a broader range of blues but a higher peak of red than of blue. Perhaps there were a few yellowish leafs. Thats the problem with the setup. You can't know for sure what exactly you are taking the spectrum of when you just point and shoot somewhere.

And now for a nice yellow flower:


Isn't it pretty? We should get a lot of yellow from that one, shouldn't we?


First thing you might notice is the flickering "waterfall" image. There was some wind that let the flower tumble a bit. But for the curve I used one of the brighter lines in the middle. The second thing you should notice is that there is almost no yellow at all! Where green and red curves should overlap to give us some nice yellow the curves are strictly separated! We have a very strong blue and green section but only a little red and still it adds up to a strong yellow in the photo. Perhaps I did not properly aim at the yellow part of the flower? Okay, lets take another spectrum. This time with a stillstanding flower and aimed right at their yellow heart:


That's more like it! Green and red dominating, adding to yellow. Seems it was a targeting problem.

Improvement proposal:

To avoid that we would need a kind of "sights", preferably a direct photo through the slit and a "two picture monitoring image" on the computer screen. One image of what the spectrometer "sees" and one image of what it measures, in real-time, of course. And it shouldn't raise the price of the spectrometer too much. A second web cam inside? A prism to part the light? Let's challenge the tech wizzards here!

Okay, last but not least a nice blue flower:


In fact it is a bit violet, so in the spectrum we see the dominant blue and a good amount red adding up. Interesting here is the blue "plateau" on the red hand side of the spectrum. Again: shouldn't the diffraction grating send the blues to the left and the reds to the right?



The daffodil spectra are great lessons. It's very unusual to see much yellow in the photos of diffraction patterns from Public Lab spectrometers. That's probably because consumer cameras don't have yellow sensors, only R, G, and B. To figure out how much yellow is hitting a pixel, the camera's computer uses the brightness of the red and green sensors and applies a rule to make up a number for how much yellow there is. When there is lots of yellow (like in a daffodil) the rule gets it right, but it seems that the yellow part of spectra (around 600 nm) of white light sources is often represented by a low area. That dip is right where the sensitivities of the red and green sensors overlap. The rule must be a little stingy when it comes to assigning yellow to pixels.

In general, using a Public Lab spectrometer to learn about the color of reflected light requires great care. In an ideal Public Lab laboratory, the light source for these observations would have a spectrum like this:


The material used to make the diffraction grating would have a transmittance spectrum like this:


The lens of the camera would have a transmittance spectrum like this:


And the camera sensor would have sensitivity like this:


The actual curves for all of those things are not even close to straight lines. To interpret the reflectance results from a Public Lab spectrometer, you have to know how each of the actual curves deviate from the perfect curves. Then you also must know how the color image of the diffraction pattern is being processed to make a jpeg photo or video frame. All I know is that's a lot of things to know.


Hi Chris, you said "consumer cameras don't have yellow sensors". All the modern cameras I know just have the RGB sensors. I wonder if we would get a much more correct spectrum if we wouldn't use any colour sensors at all but only intensity sensors at the different places where the photons arrive. Photon counters, so to speak. A much finer grating or prism to get a wider spread-out spectrum and a range of photon counters spread over the whole range. Is that how they do it in precision spectrometry?

BTW, I wondered why I did not see any hint of Fraunhofer-lines in the sun spectra I have taken, but I guess that low-cost setup is just not precise (and widespread) enough to see them. I think the PLab spectrometer really has its strengths when it comes to emission spectra like in environment pollution testing, not so much when its about absorption spectra. Right?

Is this a question? Click here to post it to the Questions page.

There are lots of examples of DIY spectrometers based on linear CCD detectors which are just single rows of pixels with no color filters over them. Here are a couple of notes at Public Lab:

The linear CCDs cost less than US$10 now. This design completely eliminates the variable sensitivity with wavelength caused by the consumer camera dependence on color filter arrays (Bayer filters). If you want to compare the intensities of different wavelengths in a spectrum, that is a much better approach than using a consumer camera.

Detecting Fraunhofer lines just requires good wavelength resolution. Using a narrow slit and a point and shoot camera instead of a webcam makes a few of the Fraunhofer lines pretty conspicuous:

If it is well adjusted, a webcam spectrometer can also detect some Fraunhofer lines. They show up in some sky spectra at Spectral Workbench.


Yes, those linear diode array spectrometers were exactly what I had in mind. Unfortunately I am not so good at electronics and programming but that seems to be the way to much more precise readings without depending on coloured sensors. And it needs a 1000 lines/mm grating which shouldn't be too hard to get. I looked at those setups and there is only one thing that could improve them: if you could have a slightly circular bent diode array, so that all the rays have exactly the same way from grating to sensor. In one plan they used a bent mirror but then there is the same problem on the other side of the mirror - to get a circular wavefront become a linear one. But all in all it is very interesting to think about these things and the maths behind them. Thanx for the links!

your photographs are beautiful! Actually, you don't need a prism inside the spectrometer, that's what the DVD piece does, the slit only allows a certain quantity of photons into the detector and as they hit the DVD grating they are diffracted into their constituent parts...hence, a spectrum!

The RGB channels make all colors, they are called the primary colors, secondary colors lie between the primary ones, a good example of this is here:

What's your dog's name? I ask because I just love animals, I have 3 cats.

have a good one my friend!

Dave H

Is this a question? Click here to post it to the Questions page.

Hi Chris,

I just had a look at the "ideal spectra" above. The ones with the 100% straight lines. Made me think about how light is created in the light source. Usually, natural radiation (from the stars for instance) is a kind of black body radiation. Something is "heated up", it's electrons get raised to a higher band and when they fall down they emit radiation. The highest peak of this radiation is related to the temperature of the black body. Here is a spectrum of our sun. (public domain)


The only way to get a straight curve (not necessarily 100%) would be to add loads of different curves from different sources with different temperatures. Something like that (just a crude drawing, don't mind the colours, they are just to distinguish between curves):


Should be interesting to think about how to construct such a light source physically. Anyone keen on a Nobel Prize???

(One way could be to use only one single curve and "shift" it constantly towards red or blue...)

Is this a question? Click here to post it to the Questions page.

Hi Dave,

yes, of course the DVD grating spreads the light but what I meant was a splitting of the light beam into two parts, where one part could be used for the spectrum and one part for the "sight" or targeting monitor. Just looked it up: its called "beam splitter":


Oh, and here are my two little hobbits. They are called Frodo (the cat) and Sam (the dog):


Ah, ok that makes more sense now, a beam splitter, interesting how you want to use that, sounds cool.

What a pair of adorable little rascals!! That made my day! Thanks!

Dave H

Login to comment.

Public Lab is open for anyone and will always be free. By signing up you'll join a diverse group of community researchers and tap into a lot of grassroots expertise.

Sign up