Raspberry Pi Environmental Data Logger

This project sets up a data logging system for my raspberry pi to monitor temperature, relative humidity, and CO2 levels. It presumes you have an available postgres database to log to. It also presumes that your Raspberry Pi has an internet connection.

Hardware Setup

Adafruit's i2c sensors are very well set up with excelent python integration. For this project, I used

• SCD-30 real CO2 reading which handles CO2 PPM, % relative humidity, and temperature
• DS3231 Real Time Clock (RTC) and temperature source #2
• TMP117 temperature source #3
• MCP9808 temperature source #4
• Stemma QT SHIM for easy connection of the sensors

I opted to use PoE to power my Raspberry Pi since I needed to get a network connection to it anyhow. I'm cheap so I used Adafruit's USB C PoE Splitter (but you can find them on other online stores too) rather than using a PoE hat. This also simplified making the "case" for my setup.

I designed and 3D printed a "lid" and mount panel that I could use plastic standoffs to attach everyting with. I used M2 standoffs so that I'd have a bit more tollerance in the hole layouts. I went with this set of standoffs which includes various lengths of standoffs in addition to M2 plastics screws and M2 nuts.

I used some Dual Lock velcro to affix the Pi and PoE splitter to a beam in my crawl space. I used a low-voltage in-wall mounting box/bracket and a keystone-compatible face plate with some keystone Cat6 female-to-female coupler jacks to pass ethernet through my wall from my network closet into my crawl space.

Installed, it looks like this:

Database Setup

1. Make sure an instance of postgres running somewhere
2. If a new database is needed, set that up
3. Use the contents in the ddl.sql file to set up a table, modify as needed

Note: I used heroku's postgres hosting. They have a free tier that supports 10,000 rows and 20 connections. This system only uses 1 or 2 connections and at 15 minute logging frequency, that's 96 records per day or around 100 days of logging. One caviat with using Heroku is that they ocasionally perform database "maintaince" that results in the hostname, database name, user name, and password being changed. They email before this happens but it can still be a bit of a hastle. In the future, I think I'll either install postgres on the Pi, or once I get my container cluster running, host a general-purpose postgres database there instead.

Software Setup

Fill out the logging_config_template.ini file with the pertinent info. A copy of this file is used both on the pi and locally for reporting.

Logger

1. Set up a Raspberry Pi, probably with raspbian or debian or a similar OS
2. Make sure there's a running ssh server on the pi
3. Install python 3 if it isn't already installed
4. Copy the requirements-pi_logger.txt, env_logger.py, and filled out logging_config.ini to the pi and make sure the .ini and .py files are in the same directory
5. Install the requirements file with python -m pip install -r requirements-pi_logger.txt
6. Set up the .py file to be run automatically when the pi launches, perhaps by
   • Adding a cron entry
     1. Launch cron editor: crontab -u pi -e
     2. Add a line like @reboot python3 /home/pi/env_logger.py
     3. Save and exit the editor
     4. Reboot
   • Adding it to rc.local
     • Untested, but probably with su pi -c 'python3 /home/pi/env_logger.py &'
   • Adding a script to the init.d directory
   • SYSTEMD

Reporting

1. Make sure you have a python 3 environment
2. Install library requirements with python -m pip install -r requirements-local_reports.txt
3. Copy the filled out logging_config.ini file to where you store the plot.ipynb file
4. Launch jupyter lab (or use something like VSC) from the same directory as the ploting notebook
5. Run the notebook to see the plots, optionally edit the query or save the image

Outputs should look something like

Note: I suspect that the erratic data shown in this demo is due to the proximity of my Pi to my HVAC ducts. I'm planning to either move the Pi away from the ducts or distribute the temperature sensors around the crawl space rather than having them all mounted together.