A long time ago, I wrote about my little Pomodo Pi project [1] in this magazine. It involved me monitoring my tomato plant's soil humidity and watering the plants when there was a risk of them drying out by automatically opening a solenoid valve when the humidity dropped below a certain value. The Vegetronix sensors I still use for this are high quality and durable, but, unfortunately, they need an extra A/D converter and wiring them involved some tinkering.

Meanwhile, I have bought some humidity Mi Flora plant sensors (Figure 1). Their manufacturer, Xiaomi, envisages sending the data to a smartphone app, but I never bothered installing it; instead I pick up the data directly via Bluetooth.

Figure 1: The Mi Flora sensor, which uses Bluetooth, is a spin-off product from the Chinese smartphone gadget market.

I run the whole thing on a Raspberry Pi, but of course any Linux PC equipped with Bluetooth hardware will do. Software-wise you need the bluez, python3, and python-pexpect packages, which are quickly installed.

Recording the Weather with Bluetooth

The next step is to scan the environment for Bluetooth transmitters. This can usually be done with the command

sudo hcitool scan

but since the sensor is capable of low-energy transmission, I did the scan by typing:

sudo hcitool lescan

The result was a list of MAC addresses and the short names of nearby devices that can speak Bluetooth Low Energy. Hey presto, one of them was my sensor:

[...]
C4:7C:8D:6A:5E:17 Flower care

Now I have the MAC address and want to see if the sensor is sending any data:

sudo gatttool --device=C4:7C:8D:6A:5E:17--char-read -a 0x35

The feedback doesn't need to mean anything to me at first glance:

Characteristic value/descriptor: aa bb cc dd ee ff 99 88 77 66 00 00 0000 00 00 00

The important thing is that there is a response at all, because on GitHub there is a small tool that can interpret the data from the sensor. I cloned it to my current directory, as follows:

git clone https://github.com/open-homeautomation/miflora.git

The package contains a small Python script by the name of demo.py, which reads and displays different measured values from the sensor. Now let's run it:

python3 /home/pi/miflora/demo.py--backend gatttool pollC4:7C:8D:6A:5E:17

Voila! The output looks good:

Getting data from Mi Flora
FW: 3.2.1
Name: Flower care
Temperature: 23.0°C
Moisture: 40
Light: 193
Conductivity: 247
Battery: 100

The humidity is given as a percentage and the luminous intensity in lux. The Conductivity is shown in microsiemens – it indirectly says something about the nutrient content of the soil.

That's the full extent of my electronic green thumb (at the moment, Figure 2), and I'm already looking forward to the grand opening of my salad bar.

Figure 2: 24 hours in the life of a potted plant under Charly's care.