A simple webpage designed to display information read from the AMS Franklin Lightning Sensor.
The website makes use of the RaspberryPi-AS3935 library.
In my test setup I connected my breakout board to the Pi as shown
| AS3935 Pin | Raspberry Pi Pin |
|---|---|
| 4 (GND) | 25 (Ground) |
| 5 (VDD) | 1 (3v3 Power) |
| 10 (IRQ) | 11 (GPIO 17) |
| 11 (I2CL) | 5 (SCL) |
| 13 (I2CD) | 3 (SDA) |
To run this webserver you'll need to have the correct kernel modules loaded. Adafruit has a nice
tutorial
set up, though depending on the breakout board that you use, you may
not see anything show up when you run i2cdetect.
To directly access the hardware, you'll need to install the python-smbus package available in
the repositories on Raspbian.
Other dependencies can be installed using sudo pip install -r requirements.txt. sudo is required
unless you use a virtualenv and write udev rules so that the I2C bus is accessible by an
unprivilged user.
Copy the settings.cfg.sample file to settings.cfg and edit the appropriate
portions to match your setup. Each setting has comments to help.
After everything is done, you can start the server by running sudo python lightning_web.py. If
everything starts without any errors, you can visit http://pi_address:port/ to interact
with the sensor (the default port is 5000).
An upstart/systemd job to start the server on boot will likely come later - we're just not there yet (feel free to submit a PR).
Running the server this way does not scale well beyond a few connections. If you need to support more users, I'd suggest looking in to gunicorn, and possibly nginx (configuration information at https://flask-socketio.readthedocs.org/en/latest/).
I've tested running the site with gunicorn using the command
sudo gunicorn \
--worker-class socketio.sgunicorn.GeventSocketIOWorker \
--bind :5000 \
lightning_web:app \
--log-file -
Just ignore the exception it throws early on for now (it appears to be a bug in gunicorn).
If you still have scaling issues (how many people care about lightning at your house anyway?), then things need to be re-designed. At this point you should probably be pushing the data to real server(s) that can re-distribute it as needed.