hi Andy,

This looks very interesting!  I've been working with (or at?) temperature loggers for a while, and the interchangeable thermistor idea here is the most promising where accuracy at low cost is concerned.  It's important to note it is the accuracy of the thermistor that is 0.05 degC; the accuracy overall of the logger is going to be less.  I've not been able to find more technical details such as part numbers or schematics, so it's hard to get a handle on that.  Concerning cost,  I do know however that similar thermistors run on the order of $10 in onesies, which makes me think the $9 BOM cost is attainable at qty 1k or 10k.

Thanks,

-harold

Oh, that looks like such an interesting project. I like that the cratering of hardware costs is enabling projects like this to happen. I have some comments on the build which are definitely not a criticism since I love seeing applications of technology that serve an actually useful purpose.

If the goal of the hardware is to measure temperature at a low price point, I would offer some suggestions on possible ways to bring the price down or improve the design while bringing down the cost:

• Instead of a preform PET tube, it might be an idea to use old PET drink bottles. The preform bottles would either need to be included with the design kit or be found easily by anyone wanting to test out the design. This would mean manufacturing your own tubes and caps which would be a pain. Instead, I would recommend using empty PET drink bottles. These could likely be hardened to be watertight to X meters and would both repurpose the plastic for something useful and remove the need to manufacture custom plastic. PET drink bottles are usually specified to withstand fairly high pressures due to holding carbonated drinks and are definitely waterproof. @htarold  would actually be the expert on the feasibility of this approach. 
• Re-design the board to integrate everything. [Note: I realize it might be difficult to do these changes, so if you're in contact with the people on this project, I'm okay to advise or contribute time to help implement them] In re-designing the board, you could create a form factor that fits into the above-mentioned PET bottles saving on those as a system cost. It's also possible to custom design the power supply to optimize the power and maintain stability. The board being used is designed to run at 5V, but you can actually run that microcontroller down to1.8V reducing the power by a factor of almost 10 (power consumption is proportional to voltage^2).  It might also be worthwhile to consider using a precision voltage reference rather than the power supply as voltage reference. Even if you have a precision sensor, if your voltage reference isn't at least as precise, that would introduce uncertainty. I believe @htarold was referring to this in the above post. And finally you can integrate the sensor and all associated components onto a single board saving the cost and labor of the prototyping perfboard needed underneath the Nano.
• From the pictures in the paper, it looks like the sensor is located inside the plastic tube. If that's the case, what it's sensing is the air inside the tube, and the unspoken assumption is that the ocean temperature will reach equilibrium with the tube walls which will reach equilibrium with the air temperature so the sensor will reflect the ocean temperature. I think it's a valid assumption, but air, especially non-moving air, is a poor thermal conductor so it would take a long time to detect changes in ocean temperature. One possible idea would be to use a sensor that is actually directly inserted into the water. This would sample the actual ocean temperature in real time and would respond more quickly to any changes, ie: if the temperature changes at night vs day, etc. Thermistors can actually be purchased encased in stainless steel tubes. Or for a specialty precision thermistor like this, it's also possible to purchase a stainless steel temperature sensor tube housing and manually insert the sensor. An image of an encased thermistor is below.
• It might be possible to provide the components in kit form and lightly disguised as a soldering learning kit as well. That way, it's possible to have "soldering workshops" where people learn how to solder by assembling these kits. Attendees learn a new skill while protecting the environment, the orgs get temperature dataloggers, save the cost of assembly, and perhaps even make a bit of money for soda and chips if attendees pay to learn how to solder :)

If the Opuhala project wants, we can help procure components wholesale through our purchasers. 
Anyways, it sounds like a great project, if they need any advice or help, we're happy to contribute to improve reef health. 

Akiba, FreakLabs

Hi Akiba,

I made a mistake, I did manage to find some +/-0.05 degC thermistors for USD4 on Digikey, so the USD9 figure can be feasible in low quantitites.  I actually bought a bunch of the thermistors you linked to; the part variation is quite high so calibration is needed to get good results.  I think this is one area where we can truly say Cheap, Accurate, Simple: pick 2.  Edward Mallon whom the authors also cite https://thecavepearlproject.org/ has blogged extensively on this.

Personally I would try to avoid analoque devices because I don't want to deal with having to amplify it, which is where the errors creep in.  Maxim make an ADC that directly compares resistance ratios, and HX711 simplifies bridge measurements.  I'm not sure how this logger does it; the Arduino's ADC isn't very good as you said.  The MAX30205 human body temperature sensor may be worth looking in to for this application.  The measurement range is small but probably adequate for this application, and the accuracy is well documented.

The sensor time constant is a bit high but should be ok for for this low-rate application, and kudos to the authors for characterising it.  There are applications where the time constant needs to be <1s, like for ocean turbulence structure measurement, but that is more physical oceanography and only indirectly wildlife conservation.  The internal sensor simplifies the logger and this is a big advantage in this instance, one I have taken advantage of in the past.

I have considered PET bottles in the past and it is very attractive.  They do very well with internal pressure (7 bar I think!) but with external pressure they are prone to buckling.  One approach may be to re-mould the bottle to be smaller so the walls are thicker.  This should be easy because the bottle shrinks with the application of heat anyway.

I like cheap hardware too!  Plus it goes well with citizen science in a field where investigation is labour intensive and it lends itself well to outreach efforts.

Hey guys,

please count me in. I am software engineer and would love to get some well structured knowledge about conservation, environment, ecology, climate change, biodiversity loss and alike. I'll be more than happy to create learning group with you.

Hi everyone,

I've taken several courses from the Edx platform and a few from Nat Geo.

I highly recommend Edx courses. They cover material with enough detail that I've always learned something but never been overwhelmed. Many of them provide additional material for more in-depth reading.

I can't speak to any of the other platforms on the list, but I intend to check them out.

Hi Andrew,

I watched your vid and finally I've got a decent idea of what you're trying to do!  Took a while but what can I say!

I absolutely agree that the machine vision techniques are overkill and that you're on to something here.  It got me thinking about these red line lasers, which project a straight line onto a surface.  These are used in place of chalk lines in construction sometimes.  If you were to project a red laser line onto a tree branch, and focus an image of a length of that line onto a detector (LED), you could pick out the flicker in intensity as ants cross the line.

If you want to get fancy, you can use your optical mouse as the sensor, or a regular camera such as OV7660 for a couple of bucks.  You'd bolt the laser to the camera such that no matter what surface you projected the line onto, the line would show up only as pixels on row 240 (say) of the image. Now it's a matter of watching the pixel values fluctuate.  You can use the usual signal processing techniques to filter out the wrong frequencies.  Use 2 laser lines on 1 camera and you can get ant travel direction too.

It sounds like you are referring to what is often called the "Cocktail Party" problem.  There has been quite a bit of research in this area, but it is a very hard problem.  I would start with the general literature on this topic as there isn't much in the conservation space. Check the hearing aid literature....picking out someone talking in a crowd of voices and sounds.  Humans are pretty amazing at this...machines not so much.  You might consider following up on some recent research at MIT:

https://blogs.nvidia.com/blog/2018/08/28/music-youtube-cocktail-party-problem-ai-artificial-intelligence-deep-learning/

For what its worth, it's a fascinating problem and has a lot of conservation applications.  As Nick mentioned, the biggest challenge is getting enough usable soundscapes.