https://www.kickstarter.com/projects/903107259/scio-your-sixth-sense-a-pocket-molecular-sensor-fo
So…is there any chance this is what it claims to be? Seems like a major revolution if it is.
I suppose it could be a hoax, but it seems like it would be possible IF you had a spectrometer that could fit in your hand.
But I’m not sure it’s as revolutionary as you think it is. It would be fun for a day or two, but after a while it would become boring just checking random things to see what they are made of. And how much is it going to cost?
We’ve somehow managed to live just fine for thousands of years without one of these.
Well, it’s not THAT exciting. As far as I know this is the smallest spectrometer out there, but there have been spectrometers small enough to go in your jeans pocket for a couple years at least - that’s how long ago I bought one. And last summer I saw a handheld Raman spectrometer, which is a much more complicated thing. It was closer to the size of a lunch box, but certainly handheld.
I’m only buying one if it is calibrated for gaseous Dikironium!
There was a Raman one on Indiegogo last year: TellSpec: What's in your food? | Indiegogo
Sounds fantastically impossible. It hasn’t passed its delivery date yet so we don’t know.
Companies like Ocean Optics have been making miniature spectrometers for years. The one in the OP measures 73 X 25 X 16.5 mm.
The off-the-shelf Ocean Optics USB2000 and USB4000 measure 89.1 X 63.3 X 34.4. The Ocean Optics STS Mini-spectrometer measures 40 X 42 X 24 mm. I suspect they could make them smaller, but there’s no real need to – their customers haven’t been clamoring for smaller, and they might have to sacrifice some resolution to downsize, which would be bad for their rep.
http://www.oceanoem.com/products/miniature-spectrometers/sts-microspectrometer-2/
The big draw for the kickstart device seems to be that you can use it with your phone. Ocean Optics and similar devices typically plug into a laptop.
When you say ‘spectrometer’ that could mean any number of things including different wavelengths of light (UV-Vis, Infrared). Looking at the video, its pretty clear to me that he’s using Near Infrared. NIR is very commonly used to measure fat, protein, sugar content in foods and identify pharmaceuticals.
THere are several ‘hand-held’ NIR spectrometers currently available, but this is the smallest I’ve seen. While I’m pretty sure this small spectrometer will have some useful applications, it will be limited in that it won’t have the resolution to determine two similar items, and the error on the fat, protein, etc percentages given will probably be much wider than other larger more precise robust instruments.
Note, that he has undoubtably used a ‘training set’ to calibrate the instrument. THis is pretty standard chemometrics for NIR analysis. He has probably loaded a few different tablet spectra in his calibration set, then demonstrated the success with a test tablet that is in his calibration training spectra. Which ever spectra the test spectrum is closest to will be identified on the screen.
I can go into further detail about how he’s using chemometrics and NIR spectrometers if anyone is interested.
(former employee of a spectrometer company)
Would you be able to (briefly) explain the different types of spectrometers in general? If there are NIR ones, are ones in other parts of the spectrum used for different applications? Are there super wideband ones that can detect anything you throw at it?
Also, can a NIR spectrometer identify one plant species from another?
For spectrometers that use the ‘typical’ regions of the electromagnetic range, you have Ultraviolet-Visible (UV-Vis), Near infrared (NIR) and Infrared (sometimes broken further into ‘mid-IR’ and ‘far-IR’). NIR and IR are related to the vibrations molecules go through, while UV-Vis is related to electrons transistioning from various energy states.
IR active species tend to have several sharper peaks that are usually well defined and can be so specific that they can be used as a ‘fingerprint’ to confirm the identity of a particular molecule. NIR usually has more broad peaks with some subtle bumps. UV-Vis is even more broad and really can’t be used to confirm identities so well. IR is used to confirm identity, UV-Vis is used to measure concentrations.
NIR is kind of a odd stepchild between the two. It is often used in agriculture and pharmaceuticals, as well as process manufacturing to confirm the presence, concentration or identity of something, but cannot be used to identify something blindly without some idea of what you might have.
NIR requires that you have lots of standard spectra that you can compare your sample to. Before you use NIR for real, you have to collect a lot of spectra of standard materials (and concentrations) that you already know what they are. Once you have this standard set of spectra, you can compare the unknown spectra to standard set and make a reasonable guess as to the identity of the material. Its still as secondary technique though, and may not give you completely accurate results.
It generally cannot be used to identify one plant species from another. but it depends a lot on the standard set you have and how you use it (or even cheat with it). If you have two plant only to chose from, and one has dried up leaves, and the other is lush, and you only have standard spectra from those two plants in your training set, than I’m confident you can distinguish between those two. But in a ‘real world’ case where you might have 10s of different plants to identify all collected in your standard training set; I expect you will have trouble. The plant leaves are too close to each other and NIR doesn’t have the resolution to do this very effectively.
I can go on…for days…
Thank you! That was enlightening. If you want to keep elaborating, please do 
Very enjoyable stuff to learn about.
One last question, if I may; Why are they separated into different devices? Why isn’t there a single device that measures it across a wider swath of spectrum?
I read through the tellspec kickstarter. They are selling the miniature NIR spectrometer for analyzing your diet. You aim it at your food, and it gathers data. The real selling point is the next step, the data is transferred to an app on your phone/tablet, which will communicate it to a cloud database of foods which your device’s data will be compared against to tell you what’s in the food. The results get reported on your phone in pretty non-expert-readable displays. The idea is you know how much of which kind of fat, how much salt, which nutrients, which chemicals like food dyes, etc are in your food.
Does that sound feasible? would the details be accurate if the cloud database is well calibrated to the device?
Generally they need to be separated into different devices because the sources of the light (either UV-Vis, NIR or IR) need to be different, and the detectors need to be different. Different detectors are needed for the different wavelengths. Plus things like lenses, mirrors and windows are specific to the particular ranges. Glass is fine for visible, but crap for UV for example. The IR source is often the same thing as the ‘glow plug’ igniter in your stove, but that is useless for emitting UV or Visible light.
I won’t go into details of scanning vs. Fourier transform instruments.
There is one instrument I’m aware of that can do NIR, Mid-IR and Far-IR, but it also requires switching sources, beamsplitters, windows, etc. during a run. Its a slick instrument and I highly recommend you spend the $50-$100K to buy it.
The main problem I see with this new NIR is that it undoubtably will have very poor resolution. It won’t give you very sharp peaks (a problem NIR has in the first place) Instead of 3000 to 5000 data points to make a nice graph of peaks with good resolution, it might only give you 250 data points. So there may not be very good differentiation between materials to be extremely useful. It won’t matter if there are a million standard spectra in the cloud if they all essentially look like each other. Though, I think there could be plenty of cases where it is ‘good enough’ especially with a low price point.
small instruments like this are either gimmicks, demonstration of principle or for survey work (it might be this substance and if it is it might be this much of it).
when you get small the optics can and often have to be very crude, course and imprecise.
much of the quality of analysis is sample preparation especially with the broad response in the IR spectrum.
Does NIR have the same problems with water vapor as IR? I’m imagining people using these in crowded restaurants as the steam rises off their food, and using them in the warm, humid air of their kitchens…
From what I understand, TellSpec uses Raman, which is different from other spectrometry in that it only shines one wavelength, then looks for shifts from that wavelength, rather than scanning wavelengths and measuring absorbance.
Water does absorb well in the NIR range but it is less of a problem then IR. Water vapor is not such a big deal, and here is why:
Materials generally don’t absorb light in the NIR range as well as they do in the IR range. So NIR can be considered less sensitive. Put another way, you often need more material to get a good NIR spectrum than you do with IR. You may be thinking this is a disadvantage for NIR…because you need a lot more stuff to get a good spectrum. However, it is often the case where materials absorb TOO MUCH in the IR and you really have to work to dilute them down or spread them into very thin films to get usable spectra. For IR you have to do a bit of sample prep.
For NIR, there is very little sample prep required; just put your stuff on/in the instrument and you’ll get something. Water vapor usually isn’t present in high enough quantities to interfere with a NIR spectrum; whereas just a little will be a problem in the IR range. Aqueous samples in IR need to be a few microns thick and no more, otherwise the water will totally cause problems. NIR will allow you to analyze aqueous samples up to 1 mm or maybe even 2 mm thick.
Note that in this spectrometer case, they are looking at reflected light, and most of my discussion up here is related to light transmitted through samples.
Another great advantage of NIR is that you can shunt it through fiber optic cables which allows you to use probes remotely, and stuck in various locations several meters from your instrument.
I finally just glanced at their website. Technology – Tellspec
From there I was able to confirm that it is a miniature NIR spectrometer, not Raman
From their website:
I can still go on…and on…and on…
You’re right, they changed technologies in one of the updates and I didn’t keep up. Now they claim to be using DLP (http://pando.com/2014/04/04/revealed-healbe-isnt-indiegogos-first-giant-medical-scam/). Doesn’t seem any more feasible, though.
I think I could agree with the assessment of that link. With NIR though the technology is a lot easier than Raman, and they may have a bit more success in building such a small (real) device. I still doubt it will have the quality and resolution needed to be as useful as they claim.