Skip to content

Latest commit

 

History

History
167 lines (113 loc) · 7.99 KB

README.md

File metadata and controls

167 lines (113 loc) · 7.99 KB
Raw

reefer

tl;dr Yeah, but what does it look like?

screenshot

This started out as a project to control my reef aquarium, hence the name. It then morphed into a simple Python script to check the indoor temp/humidity and compare it with the outdoors using OpenWeather API. Update: It now monitors the temperature of my aquarium!

Raspberry Pi

I moved the project from a FT232H to a Raspberry Pi Zero W and added a DS18B20 waterproof temperature sensor for an aquarium. With plenty of GPIO on the Pi and relays in hand, I plan on controlling pumps/lights/etc in the future.

Hardware

BME680

https://pinout.xyz/pinout/i2c

  • VCC to 3v3 (pin 1)
  • GND to ground (any)
  • SCL to GPIO 3 [I2C1 SCL Clock] (pin 5)
  • SDA to GPIO 2 [I2C1 SDA Data] (pin 3)
  • SDO to nothing
  • CS to ground (any)

During testing I kept having failures of the sensor until I discovered an unofficial fix that contradicts the official documentation a bit. CS must be run to a ground pin. All of a sudden the sensor was rock solid stable for weeks after that.

Si7021

https://pinout.xyz/pinout/i2c

  • VIN to 3v3 (pin 1)
  • GND to ground (pin 9)
  • SCL to GPIO 3 [I2C1 SCL Clock] (pin 5)
  • SDA to GPIO 2 [I2C1 SDA Data] (pin 3)

This sensor always seems to work no matter what and is very accurate. I am impressed with the performance. I am currently using it outdoors on a Pico W, so the wiring is different than shown above.

DS18B20

https://pinout.xyz/pinout/1_wire

  • Red Wire to 5v (pin 2)
  • Blue Wire to ground (pin 6 or 9)
  • Yellow Wire to GPIO 4 [Data] (pin 7)
  • 4k7 pullup resistor between 5V and Data (yellow wire)

According to my testing, Google searches, and the DS18B20 datasheet 3v3 is not enough voltage to pull up the data line with the commonly included 4k7 resistor. I measured ~1.6 volts with the 4k7 resistor in place and the datasheet says it wants a minimum of 3.0 volts. I was having massive dropouts of the sensor until I switched everything from the 3v3 rail to the 5v rail. I don't know how anybody else got it working on 3v3 and then wrote the guides floating around the web (I'm looking at you too Adafruit!). You could, of course, substitute an appropriate resistor and still use 3v3 based on some multimeter measuring, calculating, etc. But why not give that sucker some juice! It's been rock solid since I changed to the 5v rail + 4k7 pullup.

Dependencies

You can satisfy pretty much all dependencies with these commands on a fresh Pi:

$ sudo apt install git nginx rrdtool python3-rrdtool python3-pip
$ sudo pip install adafruit-circuitpython-si7021
$ sudo pip install git+https://github.com/nicmcd/vcgencmd.git

I2C and 1-Wire interfaces must be turned on in $ sudo raspi-config

RRD

Create RRD databases

$ rrdtool create temperatures.rrd --step 60 DS:outdoor:GAUGE:120:-20:55 DS:indoor:GAUGE:120:0:55 DS:tank:GAUGE:120:0:55 DS:pi:GAUGE:120:0:100 DS:picow:GAUGE:120:0:100 DS:outdoor_dew:GAUGE:120:-80:55 DS:indoor_dew:GAUGE:120:-80:55 RRA:LAST:0.5:1:2880
$ rrdtool create humidities.rrd --step 60 DS:outdoor:GAUGE:120:0:100 DS:indoor:GAUGE:120:0:100 RRA:LAST:0.5:1:2880
$ rrdtool create pressures.rrd --step 60 DS:indoor:GAUGE:120:800:1100 RRA:LAST:0.5:1:2880
$ rrdtool create gas.rrd --step 60 DS:indoor:GAUGE:120:50:200000 RRA:LAST:0.5:1:2880

This will create databases with a 60 second interval, 120 second heartbeat timeout, between -20 and 55 degrees Celsius for the outdoor sensor, between 0 and 55 degrees Celsius for the indoor sensors, between -80 and 55 degrees Celsius dewpoints, 0-100 degrees Celsius for the pi CPU sensor, 0-100% relative humidity, 800-1100 hPa pressure, and 50-200,000 ohms gas resistance with 48 hours of data before rolling over.

More information here: https://michael.bouvy.net/post/graph-data-rrdtool-sensors-arduino

HTML

index.html

Very simple HTML to display the graphs. You will likely want a webserver for this. I use nginx. It refreshes the page every 30 seconds in an attempt to reduce misses while the graphs are being generated. This doesn't work very well, so I would love to implement a server-side generated trigger for refresh, more like a "push" based system. Unfortunately, when I first looked into implementing this it looked way too complicated for my motivation at the time. Looking for better solutions.

Relay

There are 3 python files to control a relay on the Pi GPIO pins. You can use this as a timer for lights, pumps, etc.

relayOn.py Simply turns on the relay

relayOff.py Simply turns off the relay

relay.py Toggles the relay based on a schedule

I used relay.py for a while, but there's no need to have a constantly running python daemon just to flip a GPIO pin twice a day. I switched to a crontab schedule using the first two mentioned python scripts.

Use the example below to turn a light on at 7:00 AM and off at 7:00 PM:

$ crontab -e
# m h  dom mon dow   command
0 7 * * * python /home/pi/relayOn.py
0 19 * * * python /home/pi/relayOff.py

More Advanced Relay Control

$ pip install flask
$ python relayControl.py

This will launch a webserver that hosts a very simple webpage with some buttons that control a relay. This depends on relayOn.py and relayOff.py, which you can edit to your liking and of course expand everything to control more relays.

Handling Unforseen Errors

I have tried to continually improve error handling inside the main reefer.py script, however there are still errors that can pop up. I am usually running the script using $ python reefer.py & or some variation thereof to background the process and/or handle stdout/stderr logging (in addition to the built-in logging), but I am still getting various crashes which are not obvious looking at the graphs (they just stop on the last data point). So there is no way of knowing the main process crashed besides waiting a while and seeing the timestamp at the extreme right of the graphs is old.

The API's I am poking prove to be a continuous source of beard-pulling as I try to debug what is failing. I honestly did not expect polling of aviation METAR data from a government website to be so unreliable for example.

systemd to the rescue?

$ mkdir -p .config/systemd/user
$ cp reefer.service .config/systemd/user
$ systemctl --user start reefer
$ systemctl --user enable reefer
$ sudo loginctl enable-linger pi

This should automatically restart the script if it fails in any fashion and also log it properly. Note that this service is running as the unprivileged 'pi' user. The way I have my files setup in the pi home dir and nginx makes this work for my particular setup. You could do this all as root in /etc/systemd/system if you want to do it that way and manage permissions accordingly.

The reefer.service systemd unit file is very simple and included in the repo:

[Unit]
Description=Reefer Service
After=network.target

[Service]
Type=simple
ExecStart=/usr/bin/python /home/pi/reefer.py
Restart=always

[Install]
WantedBy=multi-user.target

I realize this is turning more into a glorified weather station app, but that's where my interests have led me!

TODO

  • Move generated files to tmpfs as to not kill the sdcard with all the writes.
  • Document tmpfs and symbolic link to files
  • Maybe switch from nginx to built-in Python webserver to further reduce system load, configuration, writes to sdcard, etc.
  • Remove DS18B20 references and documentation
  • Add BME680 documentation
  • Add Pi Pico W + Si7021 sattelite sensor code and documentation
  • Squash indoor/outdoor pressures to "local"
  • Output data to Wunderground
  • Break out more things in main loop to functions and use main loop for flow control only (aka refactoring?)
  • Add High/Low stats
  • Log rotation using logging.handlers.RotatingFileHandler
  • Break out graphing args to variables, squash common args to reusable variable
  • Switch to built in Python module
  • Break secrets out to external file (lat, lon, API key, etc)
  • Fix readings from barometer and gas resistance sensor on first loop iteration