I am doing my master thesis using Nerves

[nkokkos]

Hi everyone,

I decided to do my master thesis using nerves. Basically, the end product is a IoT device that connects to the local wifi and provides readings from an electrochemical C0 sensor through a web interface.

The CO sensor is an a professional sensor from Figaro, therefore I had to create an analog circuit on perfboard to capture the signal and then feed it into a breakout board I2C like asd1115 adc. I also use the breakout board bme680 since I need the temperature to do re-calibration of the sensor using a look up table. I also use the nerves key to do some sort of data push to an IoT platform. I plan to use the nerves cloud too. And finally I plan to use homex to push the data to a home assistant installation.

Most of the things used in the thesis/project have been covered here in the forum, plus in some nerves books that have been published. Through my research, no elixir nerves has been used before in thesis project in our department, so I suggested this proposal and my advisor gave me the ok.

In a nutshell, this is what this project does:

Uses a rasberry pi zero wireless, that connects to the perfboard where the CO sensor and the breakout boards are installed. Gets all the readings, provides all the readings through a poncho project using a web interface, pushes the data to an IoT platform and a local home assistant setup.

The most intriguing part was building the analog circuit interfacing the CO sensor since I had to go back to my undergraduate days and read all about operational amplifiers and low pass filters and all that.

I have started writing the code/firmware and the more I read and learn about nerves, the more amazed I become. Frank and all the people involved with nerves and the hex libraries have done a remarkable job and I thank you all. The documentation is so good.

This thesis is not rocket science but anyway I plan to release everything (source code and perfboard design) as a tutorial series, time willing.

Thank you for nerves!

[lawik]

Delightful! Please keep us updated in this thread and share pictures

[ssd]

Sounds Interesting! We’re working on an IoT cloud platform called Fostrom built using Elixir, and we even have an Elixir SDK that works with Nerves out of the box. If you find it useful, I’m happy to help you set it up for pushing the sensor data to the cloud.

Good luck!

[nkokkos]

Thank you. I will try it!

[nkokkos]

Hi, I am uploading some pictures from the progress I have done so far.

The i2c breakout boards used are the bme680, the nerves key, and a cheap ads1115 adc,

The idea was to take an electrochemical sensor like the figaro tgs5042, which outputs a tiny current and superimpose it using an analog circuit based on an application note, and then pass the voltage produced to one of the adc inputs of the ads1115. I also use the bme680 to measure the temperature, since the temperature is used for a correction factor while calculating the CO ppm range.

I started with the idea to lay everything on a perfboard to reduce EMI but it turned out that was silly. So I kept the i2c boards on a breadboard and the analog part on a perfboard. Got really frustrated with the dupont cables and their connections.

I am currently running on the very first rpi0, one core cpu, sampling the inputs of the ads1115, doing median filtering to remove emi noise and pushing data to liveview using the classic poncho scheme. Code is almost complete and I was impressed that it worked. I am currently doing experiments now with measurements and I will provide a link to github in short time.

On the breadboard, you can see that I had to use coupling capacitors before the inputs. Coupling capacitors are used also on the perfboard to stabilize the voltage source

[nkokkos]

Hi, I tried it and if I remember well it won’t compile for my target, which is the very first rpi0, of 2017. That’s the board I am using.