Cheapest possible "toy" spectrometer? $323

[andy]

Yeah one thing I am interested in doing is pulling together all the variables and make a calculator to get simulated results based on all your factors I listed before in that email. Right now, for example, there's not a way to easily know how much laser power you need to tell your limit of detection (there is also a limit to this based on power being so strong it evaporates liquid samples or burns solids). But you can also re-design your spec to spread the power out more but that's more complicated. I have yet to try SERS but I think if people need <0.1% concentration that's what you need to do.

Just having a camera chip isn't enough you have to have it wired up, lenses to focus the light to it, software to drive it, etc. Stas's one which uses that sensor he sells on https://www.aseq-instruments.com/ and that's probably going to be your best deal but it's roughly the same cost as Luc's (but modular). You can ask Stas to make you one that's 500-650nm or any other wavelength.

[alexose]

Right, that was kind of dumb wording-- What I meant to say was that it's hard to compare a sensor to a finished product in terms of raw performance, because there are indeed so many variables.

I am quite certain that we'll need to design our own modular sensor if we want to get the price floor down as low as possible. I'm more than willing to hack on this, but I still have some learning to do. This youtuber has set some examples to follow: https://youtu.be/VSUVaia8jdE?t=385

One stupid question I have is regarding the SMA fiber connection: Is there any reason not to do it this way? If we start with a fiber-coupled laser, wouldn't this allow us much more flexibility in our design?

[andy]

Yeah maybe, I guess there's a difference to me between practical low cost DIY and high scale. There are a handful of ways to save money if there is a 1,000 unit order. But in terms of a design that someone could build one-offs I think that makes things more tough

You mean why not have the spectrometer SMA fiber coupled instead of directly free spaced like OpenRaman is now? I think the advantage is definitely the modularity and you don't have to build that part of it. There are some good specs, you can customize this part (like change the laser wavelength more easily), and lots of second hand ones of variable quality. You could also theoretically throw it in a freezer more easily, but from my testing just throwing a spectrometer in a ice bag doesn't make a big difference. I think the chip produces more local heat then it can absorb through the case so you have to have the TEC directly on the chip.

The downside is now you have you have to carefully align your light into the fiber. Luc talks about using a single mode ~5um fiber, but in my experimentation my light is far was not well collimated so I struggled to do that. You would have to have your light well collimated going to the focusing lens, then precise alignment and positioning of the focusing lens. Right now I could capture maybe 1% of my light. Though something else was wrong with my setup.

Other then that if you are using 30-50um then the setup is more similar to OpenRaman now with the slit. You lose some light sending things through the fiber but it's not a huge amount from my understanding. So I guess just complexity of having that extra part, but simpler from the perspective of less to build if you are using an off the shelf spec.

[alexose]

Cool, that matches with my understanding as well. The biggest challenge is getting the laser light well-collimated so that it enters the fiber efficiently. I'm going to file this into the "problems to solve in the near future" bucket.

I was pleased to see that Aseq Instruments uses a mostly off-the-shelf Peltier + computer fan combination. These are readily available for less than $50, and can be really effective in the right setting. One neat thing is that we'd get cooling (for the sensor) and heating (for the laser, maybe) from the same small form factor.

In terms of laser alignment, I was just thinking about when I had to align the laser from my cheap K40 laser cutter. There's an excellent guide that's geared towards newbies. Maybe we can gain some inspiration from that process: https://k40.se/k40-laser-lens-mirror/mirror-alignment/

For the K40, the laser is powerful enough to burn a hole in regular printer paper, which makes it easy to visualize your adjustments. For aligning a less powerful laser, maybe we could use photosensitive paper or something.

[alexose]

As hinted, Luc has released his write up on tuning the cheaper Thorlabs laser: https://www.thepulsar.be/article/tuning ... aser-modes

And, wow, this seems like a very encouraging development. Lots to unpack here.

I am particularly interested in the use of a Peltier module to heat the laser, as we'll also get cooling capability that we can use on the sensor.

Also (since I have FDM printing on my mind), I wonder if we can't use some fancy high-end thermally conductive filament to print the assembly? https://tcpoly.com/purchase-ice9-materials/

[Luc]

I did not know about this filament type! It's interesting.

I'd expect that it slows down the TEC due to higher temperature resistance but it shouldn't break the PID stability.