6 ms·

Public Lab DIY Spectrometry Kit

135 points by stevewilhelm 14y ago

daniel_reetz 14y ago
I applaud the approach, but it seems like a challenge to achieve useful accuracy across many devices. I am a little skeptical about the 3nm claims. First, the (bandpass) Bayer color filter arrays on cameras limit their color resolution pretty sharply. It might be a better (if less elegant) approach to use the ambient light sensor and scan it across the spectrum generated by the diffraction grating. Also, the resolution of a spectrometer is in part determined by entry slit width. Old timers and DIY'ers use abutted razorblades to make a clean, straight slit. Problem here is that there is a trade-off. You need a wide slit to get light in for these insensitive webcam sensors. But then you lose spectral resolution because your slit is wide. Alternately, you narrow the slit to get more spectral resolution, but then your image is dominated by noise. I'm sure these guys have found the sweet spot. Second, the spectral response of the color filter arrays varies both from camera to camera and even within same-model sensors (I personally own over 40 of one model of Canon Powershot, and the differences in color response are fairly extreme). This could in principle be at least partly calibrated out, if you had each device take a picture of a calibration standard. Third, the image signal processors are closer and closer to the sensor these days. It may be impossible to get data which hasn't at least been somewhat mucked with, meaning some people will have a useful spectrometer and some won't. This goes doubly for phones with bad imaging pipelines, where the image processing is out of your control. Lots of spatial and spectral processing going on. It will be interesting to see, for example, how the auto white balance and sharpening algos deal with having zero natural image content in the captured data. Aside from the within-sensor variation/image processing problems it would definitely be usable for relative data - I see that they used a relative approach on their UV detection of a bluing dye in detergent. Nice. All that said, I run a somewhat-similar project using cheap cameras to do work previously only reserved for hard-core SLRs and scanning line sensors, and I've found that over and over again, something is vastly better than nothing and often "good enough" is better than expensive, complicated, and pain in the ass. I've also dealt with a lot of people laying out a bunch of complicated technical "Here's why it won't work" talk and have learned to counter with my own "But look, it does work" answer. Blah blah: I think it's important for the capture software captured as much metadata about the image sensor as possible. Their software looks great and a database of spectra would also be great. It would be AWESOME if the Shazam-style auto identification worked at all. Good luck, I'll probably be supporting the project.
- sp332 14y ago
  >First, the (bandpass) Bayer color filter arrays on cameras limit their color resolution pretty sharply. It might be a better (if less elegant) approach to use the ambient light sensor and scan it across the spectrum generated by the diffraction grating. The diffraction grating spreads the different frequencies physically across the sensor. So the camera sensor doesn't need any color information, you only need position and brightness information.
  - daniel_reetz 14y ago
    Right, but the different physical locations on the sensor have different spectral sensitivities due to the bayer CFA. Ideally, you'd have panchromatic pixels with no bayer pattern.
    - sp332 14y ago
      Oh I hadn't thought of that. I assume some calibration would be required for each device anyway, I wonder how hard it would be to detect and correct for this effect?
      - daniel_reetz 14y ago
        People have done it for a variety of purposes (natural image statistics, astronomy, etc). I can post references later. Problem is that it almost universally requires you to photograph a standard light source or a standard reflectance target under a standard illuminator. Or a monochromator. So you can calibrate your cheapy device but you need an expensive/scarce standard. It's an interesting problem and I look forward to seeing how they solve it.
        eternauta3k 14y ago
        > requires you to photograph a standard light source How about a black body? Tungsten lamp perhaps?
zdw 14y ago
I'm no optical scientist, but seeing as the diffraction grating being used is recycled from a DVD and the sensor is a consumer webcam, I wonder how accurate/reproducible the results would be. A conventional diffraction grating is pretty inexpensive ($2.50): http://www.rainbowsymphonystore.com/difgratfilsh.html http://www.rainbowsymphonystore.com/difgratfilsh.html
- tomkinstinch 14y ago
  Gratings like this (or a CD) will work, but they introduce plenty of artifacts. Scratches and material nonuniformities can scatter light in a way you don't want, and reduce the (already limited) amount of light hitting the right spot on the sensor.
nsp 14y ago
rze
nsp 14y ago
.-
tomkinstinch 14y ago
I attempted this before, even on Kickstarter: http://www.openspectrometer.com/ http://www.openspectrometer.com/ http://www.kickstarter.com/projects/makerhaus/open-spectrometer?ref=live http://www.kickstarter.com/projects/makerhaus/open-spectrome... Unfortunately we did not hit our fundraising target, and it's now a dead project. I'd love to see this new Kickstart project succeed, especially the online spectral database. Commercial databases of spectral signatures are expensive (thousands of dollars), and they shouldn't be. Part of our problem, I think, was that we were attempting the project while taking classes full time, and working. We were planning to release two different designs at different price points. One would have used inexpensive ruled transmission grating, while a more expensive unit would have used a concave holographic aberration-corrected reflective grating. The latter would have had better system efficiency, less "smear" across pixels, and better linearity. Each would have made use of a linear CCD array with good sensitivity (the same chip used in the Ocean Optics units). Concave gratings add a couple hundred bucks to the price, but are well worth it if you need to do UV. A challenge we didn't get to solve was how to make something DIY that could be used to detect UV. It's difficult to sputter a uniform phosphor coating or remove a sensor window without specialized equipment.
- forgottenpaswrd 14y ago
  "Part of our problem, I think, was that we were attempting the project while taking classes full time, and working." I think your project was fantastic. I see the main problem is the price. Most of the normal people can't justify expending $800 on an spectrometer. They don't even know what they could use it for. People that could justify it because they need it will buy tested equipment anyway, they have money but not time. So you need to fill the gap. The new project is making it super cheap so they could get traction like Reprap machines. At first they are terrible, but with enough people and huge market improvement is so fast like with the PC that at some point in the future it could get better than commercial closed systems. Start small, if you can't detect UV it does not matter, remember it is 100x better than Newton ever had.
  - nmz787 14y ago
    Thanks for the kind words! The $800 price tag was just a kickstarter only price, since we were still largely in the development phase we weren't sure of final pricing. Now it looks like a UV capable scientific instrument with a chinese grating would cost around $400, if you want the same optics but U.S. made it would be around $900.
defdac 14y ago
When I do this with a Project STAR spectrometer and my cameras and point it to a candle I get two distinct dips in the spectral distribution. It should be even like a blackbody. I have made a parser for the RGB-picture to turn it to a B&W power distribution and I thought I'd might do a simple software filter to make it completely even, like a calibration of sorts. Like so: http://www.flickr.com/photos/defdac/7874867702/in/photostream/lightbox/ http://www.flickr.com/photos/defdac/7874867702/in/photostrea... I use the power distributions coupled with lumen and wattage-specification as input for my home made spectrally based global illumination renderer to calculate PAR in aquariums. Do you think this will work?
- defdac 14y ago
  The IR-filter most cameras have is a real problem too. I have a webcam with the IR-filter removed.
Jun8 14y ago
What are some of the cool experiments one can do with a spectrometer, other than the ones mentioned here (I especially liked the wine analysis application)?
- tomkinstinch 14y ago
  If you can transmit light through a sample curvette to the spectrometer, you could use it to measure optical density (i.e. the amount of stuff in the way that blocks or scatters light). You could use that to measure the concentration of particulate in engine oil, or the strength of coffee. If you can detect at 405nm, you could measure blood coagulation time/extent (separate blood cells from plasma, measure attenuation at 405nm though plasma). You could cludge together a pulse oximeter if you can detect at 660nm and NIR (905, 910, 940nm) and have light sources to emit at these wavelengths. Absorbtion at these wavelengths changes as hemoglobin picks up and loses oxygen. The ratio at 660nm and one of the other wavelengths could be used to show blood oxygenation. If you were into photography or cinema, you could measure the emission spectrum of a light source and fit a blackbody curve to it to find its color temperature--useful for white balancing. If you wanted to match paint colors, you could use a spectrometer to do (along with some additional calculation and knowledge about your pigments). If you perform a flame test, you could use a spectrometer to learn about elemental composition.
chm 14y ago
Best of luck with the project. I think the really good idea is the open-source spectral library. You need to educate your readers a bit more, though. They need to be aware of the kind of spectroscopy the machine does, i.e UV-Vis, and what information it provides. I might try DIYing it, but what would be really interesting is an IR spectrometer, under 250$. Tough.

135 points by stevewilhelm 14y ago

I applaud the approach, but it seems like a challenge to achieve useful accuracy across many devices. I am a little skeptical about the 3nm claims. First, the (bandpass) Bayer color filter arrays on cameras limit their color resolution pretty sharply. It might be a better (if less elegant) approach to use the ambient light sensor and scan it across the spectrum generated by the diffraction grating. Also, the resolution of a spectrometer is in part determined by entry slit width. Old timers and DIY'ers use abutted razorblades to make a clean, straight slit. Problem here is that there is a trade-off. You need a wide slit to get light in for these insensitive webcam sensors. But then you lose spectral resolution because your slit is wide. Alternately, you narrow the slit to get more spectral resolution, but then your image is dominated by noise. I'm sure these guys have found the sweet spot. Second, the spectral response of the color filter arrays varies both from camera to camera and even within same-model sensors (I personally own over 40 of one model of Canon Powershot, and the differences in color response are fairly extreme). This could in principle be at least partly calibrated out, if you had each device take a picture of a calibration standard. Third, the image signal processors are closer and closer to the sensor these days. It may be impossible to get data which hasn't at least been somewhat mucked with, meaning some people will have a useful spectrometer and some won't. This goes doubly for phones with bad imaging pipelines, where the image processing is out of your control. Lots of spatial and spectral processing going on. It will be interesting to see, for example, how the auto white balance and sharpening algos deal with having zero natural image content in the captured data. Aside from the within-sensor variation/image processing problems it would definitely be usable for relative data - I see that they used a relative approach on their UV detection of a bluing dye in detergent. Nice. All that said, I run a somewhat-similar project using cheap cameras to do work previously only reserved for hard-core SLRs and scanning line sensors, and I've found that over and over again, something is vastly better than nothing and often "good enough" is better than expensive, complicated, and pain in the ass. I've also dealt with a lot of people laying out a bunch of complicated technical "Here's why it won't work" talk and have learned to counter with my own "But look, it does work" answer. Blah blah: I think it's important for the capture software captured as much metadata about the image sensor as possible. Their software looks great and a database of spectra would also be great. It would be AWESOME if the Shazam-style auto identification worked at all. Good luck, I'll probably be supporting the project.

sp332 14y ago
>First, the (bandpass) Bayer color filter arrays on cameras limit their color resolution pretty sharply. It might be a better (if less elegant) approach to use the ambient light sensor and scan it across the spectrum generated by the diffraction grating. The diffraction grating spreads the different frequencies physically across the sensor. So the camera sensor doesn't need any color information, you only need position and brightness information.
- daniel_reetz 14y ago
  Right, but the different physical locations on the sensor have different spectral sensitivities due to the bayer CFA. Ideally, you'd have panchromatic pixels with no bayer pattern.
  - sp332 14y ago
    Oh I hadn't thought of that. I assume some calibration would be required for each device anyway, I wonder how hard it would be to detect and correct for this effect?
    - daniel_reetz 14y ago
      People have done it for a variety of purposes (natural image statistics, astronomy, etc). I can post references later. Problem is that it almost universally requires you to photograph a standard light source or a standard reflectance target under a standard illuminator. Or a monochromator. So you can calibrate your cheapy device but you need an expensive/scarce standard. It's an interesting problem and I look forward to seeing how they solve it.
      - eternauta3k 14y ago
        > requires you to photograph a standard light source How about a black body? Tungsten lamp perhaps?

zdw 14y ago

I'm no optical scientist, but seeing as the diffraction grating being used is recycled from a DVD and the sensor is a consumer webcam, I wonder how accurate/reproducible the results would be. A conventional diffraction grating is pretty inexpensive ($2.50): http://www.rainbowsymphonystore.com/difgratfilsh.html http://www.rainbowsymphonystore.com/difgratfilsh.html

tomkinstinch 14y ago
Gratings like this (or a CD) will work, but they introduce plenty of artifacts. Scratches and material nonuniformities can scatter light in a way you don't want, and reduce the (already limited) amount of light hitting the right spot on the sensor.

nsp 14y ago

rze

nsp 14y ago

tomkinstinch 14y ago

I attempted this before, even on Kickstarter: http://www.openspectrometer.com/ http://www.openspectrometer.com/ http://www.kickstarter.com/projects/makerhaus/open-spectrometer?ref=live http://www.kickstarter.com/projects/makerhaus/open-spectrome... Unfortunately we did not hit our fundraising target, and it's now a dead project. I'd love to see this new Kickstart project succeed, especially the online spectral database. Commercial databases of spectral signatures are expensive (thousands of dollars), and they shouldn't be. Part of our problem, I think, was that we were attempting the project while taking classes full time, and working. We were planning to release two different designs at different price points. One would have used inexpensive ruled transmission grating, while a more expensive unit would have used a concave holographic aberration-corrected reflective grating. The latter would have had better system efficiency, less "smear" across pixels, and better linearity. Each would have made use of a linear CCD array with good sensitivity (the same chip used in the Ocean Optics units). Concave gratings add a couple hundred bucks to the price, but are well worth it if you need to do UV. A challenge we didn't get to solve was how to make something DIY that could be used to detect UV. It's difficult to sputter a uniform phosphor coating or remove a sensor window without specialized equipment.

forgottenpaswrd 14y ago
"Part of our problem, I think, was that we were attempting the project while taking classes full time, and working." I think your project was fantastic. I see the main problem is the price. Most of the normal people can't justify expending $800 on an spectrometer. They don't even know what they could use it for. People that could justify it because they need it will buy tested equipment anyway, they have money but not time. So you need to fill the gap. The new project is making it super cheap so they could get traction like Reprap machines. At first they are terrible, but with enough people and huge market improvement is so fast like with the PC that at some point in the future it could get better than commercial closed systems. Start small, if you can't detect UV it does not matter, remember it is 100x better than Newton ever had.
- nmz787 14y ago
  Thanks for the kind words! The $800 price tag was just a kickstarter only price, since we were still largely in the development phase we weren't sure of final pricing. Now it looks like a UV capable scientific instrument with a chinese grating would cost around $400, if you want the same optics but U.S. made it would be around $900.

defdac 14y ago

When I do this with a Project STAR spectrometer and my cameras and point it to a candle I get two distinct dips in the spectral distribution. It should be even like a blackbody. I have made a parser for the RGB-picture to turn it to a B&W power distribution and I thought I'd might do a simple software filter to make it completely even, like a calibration of sorts. Like so: http://www.flickr.com/photos/defdac/7874867702/in/photostream/lightbox/ http://www.flickr.com/photos/defdac/7874867702/in/photostrea... I use the power distributions coupled with lumen and wattage-specification as input for my home made spectrally based global illumination renderer to calculate PAR in aquariums. Do you think this will work?

defdac 14y ago
The IR-filter most cameras have is a real problem too. I have a webcam with the IR-filter removed.

Jun8 14y ago

What are some of the cool experiments one can do with a spectrometer, other than the ones mentioned here (I especially liked the wine analysis application)?

tomkinstinch 14y ago
If you can transmit light through a sample curvette to the spectrometer, you could use it to measure optical density (i.e. the amount of stuff in the way that blocks or scatters light). You could use that to measure the concentration of particulate in engine oil, or the strength of coffee. If you can detect at 405nm, you could measure blood coagulation time/extent (separate blood cells from plasma, measure attenuation at 405nm though plasma). You could cludge together a pulse oximeter if you can detect at 660nm and NIR (905, 910, 940nm) and have light sources to emit at these wavelengths. Absorbtion at these wavelengths changes as hemoglobin picks up and loses oxygen. The ratio at 660nm and one of the other wavelengths could be used to show blood oxygenation. If you were into photography or cinema, you could measure the emission spectrum of a light source and fit a blackbody curve to it to find its color temperature--useful for white balancing. If you wanted to match paint colors, you could use a spectrometer to do (along with some additional calculation and knowledge about your pigments). If you perform a flame test, you could use a spectrometer to learn about elemental composition.

chm 14y ago

Best of luck with the project. I think the really good idea is the open-source spectral library. You need to educate your readers a bit more, though. They need to be aware of the kind of spectroscopy the machine does, i.e UV-Vis, and what information it provides. I might try DIYing it, but what would be really interesting is an IR spectrometer, under 250$. Tough.