Very low cost spectrometer for water analysis

[PeterL]

Hi,

I Just found this project:
http://www.kickstarter.com/projects/1758749515/specshot

I find this an interesting project, but it needs some attention because it's going a bit too slow...,

I have no idea whether this sensor could work in practice. Maybe someone here could comment on that?

[Terabyte2007]

This is a very cool project, but it's application is limited to water samples. I recently purchased this Spectrometer which comes with 1 full software license. It's website markets to the education sector, but astronomers use this device and software worldwide. I use it in the lab for identifying all kinds of different lights and even some laser work. It's extremely useful for a wide range of spectroscopy. Here is the link to the website. Let me know what you all think!

http://www.fieldtestedsystems.com/

[alm]

Seems quite similar to this project, but less versatile (only useful for liquid samples) and with a simpler mechanical design and built-in light source. I would be skeptical about their ability to detect all the compounds they name. BPA for example absorbs in the UV region, while their product only covers visible light and near IR. Never mind their ability to get any sort of quantitative results without standards.

[mikeselectricstuff]

No mention of calibration - method of scanning also sounds dubious...

[alm]

Note that they didn't publish any spectra with vertical or horizontal scale. I wonder if this reflects the lack of wavelength accuracy, since I concur with Mike that it's hard to see how this arrangement can ensure wavelength accuracy without calibration. I believe the publiclab one is calibrated using peaks in CFL emission spectra from to the phosphors typically used in those lamps. Would it be possible for this device to at least do a one point calibration using the blue peak emitted by the white LED?

[mikeselectricstuff]

I don't think wavelength al is a major deal - there are sources like cheap lasers & LEDs that should give you a few absolute reference points. I suppose if you at least know that the scan is linear, you can probably correct against known, ubiquitous lines that are always likely to be there, but without calibrated amplitude I'm not sure how useful the results would be.

[alm]

Sure, you could calibrate with laser lines, but are they going to do this during manufacturing? Is the memory foam arrangement reproducible enough to hold this calibration? Is a user going to be able to get the laser light through the slit and diffraction grating under the right angle?

In a normal absorbance spectrum you'd measure optical density, which is a function of the ratio between light intensity with the sample in place over the light intensity with a blank sample. I'm not sure if it's a good idea to use an (almost) empty chamber as blank because of the different refraction of light due to all the glass/air or water/air interfaces. To determine how optical density relates to concentration you'd use known concentrations of whatever pollutant you're trying to measure as reference.

That would turn it into a chemistry experiment as opposed to a 'insert water bottle and see if the light turns red' type of device for the average Joe, however. Their software does not appear to expose the raw spectra to the user (though they note you can get access to them), but seems designed to take the raw data, apply some magic analysis sauce, and then report "DANGER Will Robinson! We found an unknown concentration of calcium in your drinking water!".

[PeterL]

This is a very cool project, but it's application is limited to water samples. ...... Here is the link to the website. Let me know what you all think!

http://www.fieldtestedsystems.com/

Well in my case I would use it on waste-water, polluted with different forms of nitrogen (NH3, NO3 and NO2). We do know how to measure these, but we also want to know what else is there. This RSpec seems interresting, Thanks!

No mention of calibration - method of scanning also sounds dubious...

and

.... Would it be possible for this device to at least do a one point calibration using the blue peak emitted by the white LED?

He does mention calibration with an empty chamber in the comments. "The SS works best when it is calibrated between each use to avoid noise. To do this it is simply run with an empty chamber (in calibration mode), then quickly run again with the nozzle submerged - total time less than 20 seconds. The calibrated data tested against itself in this fashion (empty both times) is negative for 99% noise."
So it seems that they use indeed the profile of the LED for this.

I agree on the scanning method, I found this dubious as well...

...apply some magic analysis sauce, and then report "DANGER Will Robinson! We found an unknown concentration of calcium in your drinking water!".

I actually think they don't have their magic analysis sauce yet, but that they think that the data of a few hundred active devices will somehow bring them the recepy for this sauce.
They claim the project(or at least the sensor) is going to be open source, which for me is enough. The raw data is my main interrest for now. And since they start by complaining about governments about not providing any data, they probably want to do better than just a 'DANGER' warning, but who knows..

Seems quite similar to this project

Thanks, looks very interesting to me!

[alm]

Well in my case I would use it on waste-water, polluted with different forms of nitrogen (NH3, NO3 and NO2). We do know how to measure these, but we also want to know what else is there. This RSpec seems interresting, Thanks!

What do their absorption spectra look like? I can't see these absorbing anywhere in the visible-near IR band without adding reagents (like transition metals).

[PeterL]

I really don't know what their spectra look like. In fact I know nothing about spectrometers, except for the basic principle.  This project looks  like a way to learn more about it. (which I'm allready doing ).

We allready measure bound nitrogen, using commercial laboratorium equipment (from Eutech). And this does indeed use reagents.

But it would be interresting to know what else we can find in our waste water.

As for the wtrDrop website: this becomes pretty lousy if they can't detect any form of nitrogen....