Climate scientist hoping to learn some instrument design.

[ChrisATX]

Hi everyone, 

I'm a physical chemist by training, but I mostly work on the paleoclimate and meteoric fractionation of water isotopes in regions where there is “big weather”. 

A significant limitation for such work has traditionally been the need to bring water samples back to the lab and measure their stable isotope ratios on a big, bulky, magnetic sector field mass spectrometer. In the last decade or so, some companies took off on an admirable (hey, I'm a physical chemist, after all, and that means spectroscopy > mass spectrometry  ) tangent, using telecommunications grade, fiber-coupled, distributed feedback laser diodes and high finesse optical cavities to exploit cavity ringdown spectroscopy of water vapor isotopologues. Such systems can measure water stable isotope ratios to a similar, or better, precision than their mass spectrometric counterparts. More importantly, for the purposes of my field, they are portable. They are not light or necessarily small (imagine a large, comically heavy version of a late-90's desktop computer and you've got the idea) but you can take them nearly anywhere in the world and measure these isotope ratios. Unfortunately, these systems are insanely expensive to purchase (~$100k), and are prone to non-user-serviceable failures that can run up to quarter of the cost of the instrument to have repaired. In order to utilize the technology appropriately to gain new understanding of weather and climate systems, we need "nets" of such instruments at different places and times. Even scaled down, less precise versions would suffice. But here's the rub: these companies have been ignoring these desires and abundant, loud, repeated requests in favor of more diverse product lines and (unnecessary) improvements in precision. 

So the need is there to build a far less expensive, open-source-esque tool for this type of spectroscopy, pulling from the rapidly expanding telecommunication fiber optic and laser technology spheres. Digilent seems to have provided a key component for toying with such a wild/dumb idea by producing such an incredibly affordable oscilloscope/spectrum analyzer platform that can be run with a credit card-sized computer. 

I am no engineer. I will have to learn all of this from scratch as I go. Hopefully this rabbit hole will be fun and interesting, and maybe even yield some forward progress, if not helpful caveats, can be chiseled out of the chaos. 

I hope this intro finds you all well in wild times, 

Chris

 

[xc6lx45]

Hi,

very quick thought: you don't necessarily need a spectrum "analyzer" for spectrum analysis. Your problem sounds like having to digitize the signal, make it available for offline processing. So the question I'd look at is, how can I get the signal from some measurement frontend (which you seem to have, something fiber-optic that outputs an electrical signal at the other end) into a PC. Then write software to process it.

Hint: Eventually you may realize that the calibration for your test system is as difficult a design problem as designing the test system itself (maybe less so if you can base everything on frequency).