DIY Spectrometer

[Nano sapiens]

A spectometer is used to measure wavelengths of light. It breaks light into it's individual wavelengths and can show the relative intensities of those wavelengths. You can make and use a spectrometer to show you the light makeup of a commercial fixture, a DIY fixture, or any combination thereof. Halide, fluorescent, LED...it's all light that can be measured and examined. Once the spectral output of a light source is known, comparisons can be make to oceanic waters at various depths as well as to other lighting fixtures/arrays.

 

Lab grade spectrometers are an expensive proposition, but can be very accurate. For aquarium purposes (especially coral reef aquariums), it is possible to get 'in the ball park' by creating a simple spectrometer from items that you likely have lying around the house or office.

 

There exists a public source project on the Web (http://publiclab.org/). The site is designed to assist those who are interested in testing various substances in their environment and they happen to have full instructions on how to create a simple spectrometer (http://publiclab.org/wiki/video-spectrometer-construction). There are a number of spectrometer styles available, but I will focus on the 'Foldable Mini-Spectrometer' version designed to be used with a Smartphone (http://publiclab.org/wiki/foldable-spec). Once on the Web page, you'll see that you also buy one of these units from Public Labs, but this is DIY so scroll down on the page to 'Make Your Own'. Simply click the PDF link and print the two pages. I decided to cut out the pattern with text inside from the first page rather than the second page, but either will work. Once you have the pattern cut, then you'll need to overlay onto some type of stiffer material to make a duplicate (I used a common 'Hanging Folder' that manila folder would go into in a filing cabinet).

 

Once you have neatly cut the pattern, just follow the instruction (fold up, fold down, etc.). To make the refractory piece that covers the square, you'll need an old DVD-R (any old DVD will do). Cut into quarters with a sturdy scissors and you'll see that the DVD is actually composed of two sandwiched layers; one is the refractory piece and the other is the reflector/paper backing. Be careful not to smudge the refractory piece and cut a small square slightly larger than the small square in your pattern. Position it as per the instructions and tape carefully into place. You'll also need to make sure that you have cleanly cut a thin slit (~1 mm or ~1/16") in the rear end of the pattern which is where the light will enter.

 

Once the refractory body has been taped together, align the square window to the camera's lens and tape securely (light should not entered this device except through the slit in the end). When complete, you should have something that looks like this:

 

 

Now it's ready to take pics! Having hooked my iPhone up to my Laptop I then visited the PublicLab website and created a Login and Password. Once logged in, use the 'Search' on top and type in 'SpecralWorkBench' and it will take you to where the fun begins (SpecralWorkBench.org). You'll see a bunch of spectral plots that people had already created.

 

The first task is to engage the camera function on the iPhone and point the spectrometer at any lit CFL (compact fluorescent bulb) by getting within a few inches and aligning the spectrometer perpendicular to the bulb. Adjust the spectrometer until you see crisp bands of color and take a bunch of pics. Pick the sharpest and brightest pic and download to the Laptop since you'll be using this as your 'calibration' pic (the online software needs to calibrate off of a standard, which in this case is a CFL bulb). Crop the pic to just include the visible spectrum of light, but allow a bit of the 'black' on either side for those wavelength not quite visible to the human eye.

 

On the previously opened SpecralWorkBench page, click on the blue 'Capture spectra' tab (top, left corner) and click on 'Upload Image'. From this page you can upload the pic, add a name, notes, etc. and then click 'Upload' and it will save to the site. Once you see the spectra graphically displayed, you must click on the 'Calibrate' tab and follow the pop-up instructions exactly.

 

Now you are ready to take a pic of your lighting. My canopy swivels 90 degrees, so it was simple to take a dead on photo at about 8" away from the array. Take a number of pics, focusing on having the spectra as sharp and bright as possible. Once again, pick the best and download to the computer. Crop as described previously. As before, use the blue 'Capture Spectra' tab then 'Upload Image'. This time you should notice that in the 'Use or create a calibration' field that the CFL calibration pic is already there, so the software will automatically calibrate your tank lighting pic using this CFL calibration. Click 'Upload' as you did before and you should see the spectral plot, correctly configured between 400 - 700 nm (mostly) since this correlates with the 'visible spectrum' that our eyes can see.

 

This is what my LED array spectra looks like:

 

 

 

I can then take this spectral plot and compare to graphs of natural sunlight at various depths to see if I have a match (in my case, similar to 10 ft. depth, but with a 'bump' in the 'Green-Yellow' wavelengths):

 

 

 

I wrote up this review from memory and this is not the only way to do this (it is also possible to upload the iPhone camera image directly to the software). If I missed a step or you found an easier way...feel free to post up here for others to benefit.

 

If things get wonky, you can always find help on the SpectralWorkBench' page such as 'Learn' from the top menu which will take you to videos and such describing the process.

 

One handy feature of this software is that you can overlay your spectra onto any other spectra using the 'Compare' tab.

 

If you can post up your spectral plots with a detailed description of the lighting system used, then we'll have a database of various lighting types/systems to analyze and compare to other man-made lighting as well as natural sunlight at various depths in the ocean.

 

Here's an idea for the adventurous (and those with a large enough tank). Take a pic of your lighting through the air as described, then take one underwater (waterproof bag for the iPhone, of course!) at the same distance from the light source. Use the 'Compare' tab to to see what difference light going through water makes on the spectrum actually reaching the corals. Save the compared spectra and then download the pic here so we can all see how much light attenuation and spectral changes occurs through a given distance of aquarium water.

 

Also, consider a donation to the PublicLabs site for supplying this public information and software.

[GHill762]

Very cool.. Gonna have to read this further in depth tomorrow when I'm fully awake.. Cool writeup!

[blasterman]

From what I gather about this technique, you use the camera in your smartphone to input the spectrum to be analyzed, correct?

 

The problem with this is digital camera sensors are not full spectrum. They are composed of RGB filters with very prominent sensitivities. Colors in between those sensitivity peaks are arrived it via perceptual hocus-pocus guesswork and and are not absolute values.

 

This technique results in a very dulled down and normalized spectral graph.

[Nano sapiens]

This technique results in a very dulled down and normalized spectral graph.

 

...and that's why I mentioned that it gets you 'in the ballpark'. This DIY should at least give one an indication of what their lighting spectrum looks like.

 

If I had access to a lab grade spectometer, then I'd be very interested to compare the results from this DIY to the much more sophisticated and expensive instrument.

[nady]

Spectrometer based on web(photo)camera ccd sensor will never be linear in y-axis.

And calibrate the y-characteristic (flux calibration) is very problematic.

[Nano sapiens]

Spectrometer based on web(photo)camera ccd sensor will never be linear in y-axis.

And calibrate the y-characteristic (flux calibration) is very problematic.

 

It should be quite simple to show whether this DIY spectrometer is not accurate enough for our needs by posting up spectrum plots of the same light source taken with professional equipment vs. this simple DIY device. You seem to have some knowledge/experience, so do you have access to a lab-grade spectrometer? Can you provide the evidence?

 

I'll happily delete this post if it turns out that the results show that this DIY device creates unusable, incorrect and inaccurate results.

[nady]

The main problem si the blue part of camera sensor. It is very sensitive and soon overloaded, the rest green and red is approximately applicable. And this is not good because it´s exactly the opposite as we want. Maximum radiation reef lights is situated in the blue range ... .

(And do not delete this post, someone that experience still some cost, money, time ... :-) )

[Nano sapiens]

The main problem si the blue part of camera sensor. It is very sensitive and soon overloaded, the rest green and red is approximately applicable. And this is not good because it´s exactly the opposite as we want. Maximum radiation reef lights is situated in the blue range ... .

(And do not delete this post, someone that experience still some cost, money, time ... :-) )

 

I noticed that the software on the SpectralWorkBench site asks specifically what image capture device is used, so it may take the violet/blue range issue into account.

[DurocShark]

I know I'm necro'ing this thread, but I picked up one of those DIY spectrometers. I got the Desktop Kit as well, but haven't put it to use yet.

 

I was tinkering around with it, and pointed it to the Current USA Marine LED unit I recently got. I have no idea how useful it is, but I was curious.

 

Here's the results from SpectralWorkbench.org.

 

 

That is the result of the following image:

 

 

I think that I need to do this again, but with something to diffuse the LED lights so we don't have several very hot spots. Does anybody think a simple sheet of paper would alter the spectrum enough to make it inaccurate, even with the challenges presented with this device?

[evilc66]

Unless you use a true neutral filter, you will always attenuate some wavelengths of light.

 

One thing I'd be interested in seeing is just the spectral output for a single color, like blue, or just white. If those end up looking somewhat normal, then I can put a little more stock in the capabilities of an inexpensive spectrometer like this.

[DurocShark]

Good idea. My Orbit light has a night mode.

 

I also have individual LEDs I can fire up. IIRC I have a bag of 435nm Crees.

 

This weekend. Give me until this weekend, and I'll have the DIY one and the Desktop one and do some comparisons.

Correction: They're 452nm Cree XT-E Royal Blues.

[evilc66]

Truth be told, I actually have that spectrometer too (I funded the original Kickstarter campaign), but never ended up using it because I have a far better spectrometer I can use.

[Nano sapiens]

I guess the question is can this simple DIY spectrometer at least get you 'in the ball park' so as to provide useful information.

 

Having a side-by-side comparison of output from this DIY vs. output from a more sophisticated instrument would be quite interesting.

[DurocShark]

Unfortunately I don't have access to a proper spectrometer. But here's a spectrum I took of the Philips Warm White fluorescent tube in my office.

 

 

So I think you could get in the ballpark. Which is all I wanted from this.

BTW: The phone used for these images is my Nokia Lumia Icon.

[GHill762]

so is this worth doing?