BME688 Web View

Web view for BME688 air quality sensor.

Reference: electro-dan.co.uk

Screenshots

Hardware

Any modern (RAM >= 512 MB) Raspberry Pi should work fine. Below is a picture of my setup with a Raspberry Pi Zero 2 W enclosed in a Pibow Zero 2 W case with a Pimoroni BME688 sensor with a female right angle header soldered on sticking out.

Setup

1. Database

These steps are for MariaDB.

1. Create user db_user with write privileges for logger and add password to logger/password.txt.
2. Create user db_reader with read-only privileges for view and add password to view/password.txt.
3. Create database bme688_telemetry.
4. Create a table inside bme688_telemetry called time_series:

CREATE TABLE time_series (
timestamp DATETIME NOT NULL,
iaq DECIMAL(5, 2),
iaq_accuracy TINYINT(1) UNSIGNED,
static_iaq DECIMAL(5, 2),
static_iaq_accuracy TINYINT(1) UNSIGNED,
co2_equivalent DECIMAL(6, 2),
co2_accuracy TINYINT(1) UNSIGNED,
breath_voc_equivalent DECIMAL(5, 1),
breath_voc_accuracy TINYINT(1) UNSIGNED,
raw_temperature DECIMAL(4, 2),
raw_pressure DECIMAL(8, 2),
raw_humidity DECIMAL(5, 2),
raw_gas DECIMAL(8, 2),
stabilization_status TINYINT,
run_in_status TINYINT, 
temperature DECIMAL(4, 2),
humidity DECIMAL(5, 2),
gas_percentage DECIMAL(4, 1),
gas_percentage_accuracy TINYINT(1) UNSIGNED
);

5. Enable scheduler:

SET @@GLOBAL.event_scheduler = ON;

6. Create event that hourly cleans up measurements that are one day old. If you want to see measurements on a longer time scale, just change the INTERVAL value.

CREATE EVENT AutoDeleteExpiredRows
ON SCHEDULE 
EVERY 1 HOUR
DO
BEGIN
DELETE LOW_PRIORITY FROM bme688_telemetry.time_series WHERE timestamp < DATE_SUB(NOW(), INTERVAL 1 DAY);
END

2. Logger

1. Create virtual environment in folder logger:

python -m venv env

2. Activate virtual environment:

source env/bin/activate

3. Install MySQLdb and BME68x Python packages. You might also want to run burn_in.py and provide your own state_bme688.txt file for more accurate measurements.
4. Deactivate virtual environment:

deactivate

5. Copy service file:

sudo cp bme688_logger.service /etc/systemd/system/

6. Edit service file to match your directory structure:

sudo nano /etc/systemd/system/bme688_logger.service

7. Reload services and enable and start logger:

sudo systemctl daemon-reload
sudo systemctl enable bme688_logger
sudo systemctl start bme688_logger

The logger service will now write measurements to the database at XX:00, XX:15, XX:30 and XX:45.

3. Web view

1. Create virtual environment in folder view:

python -m venv env

2. Activate virtual environment:

source env/bin/activate

3. Install packages:

pip install -r requirements.txt

4. Deactivate virtual environment:

deactivate

5. Copy service file:

sudo cp bme688_view.service /etc/systemd/system/

6. Edit service file to match your directory structure:

sudo nano /etc/systemd/system/bme688_view.service

7. Reload services and enable and start web view:

sudo systemctl daemon-reload
sudo systemctl enable bme688_view
sudo systemctl start bme688_view

You should now be able to see your BME688 at your Raspberry Pi's port 8050.

Notes

You can see more data in the dropdown menu by uncommenting lines from col_to_label declaration in bme688_view.py.

Dash, by default, gets some resources from external CDNs. We have our own server, so relying on other servers is not necessary. To fix this, I added the required .css files to the assets folder and added these two lines of code:

app.css.config.serve_locally = True
app.scripts.config.serve_locally = True

Speaking of .css files, if you don't like the current theme (SLATE), you can download other themes here. Then update this part of bme688_view.py:

SLATE = "assets/slate/bootstrap.min.css"
app = Dash(external_stylesheets=[SLATE])

You can change the look of the graph, by changing the template parameter here:

fig = px.line(data_frame=df, x="timestamp", y=val, labels=col_to_label, template="plotly_dark")

It also seems like the database connection will eventually timeout and reloading the page will at first show a broken graph. Adding pool_pre_ping=True to the create_engine call fixed this.