Co2 Monitor

Crystal Clear action run.png
CO² Monitor

Status: stable

Cat co2 sensor.jpg
Beschreibung IoT CO² Sensor
Autor: schinken
PayPal Spenden für Co2 Monitor


Someday a TFA Dostmann CO2-Sensor AirCO2ntrol appeared in our hackerspace - but the CO² reading couldn't be easily acquired through USB or any other connector. There are some implementations using python or nodejs, but they require a raspberry pi or at least a PC nearby to read the values to publish and store them somewhere.

Nowadays ESP8266, NodeMCU or in our case a Wemos D1 mini (which also has a ESP8266 on it) exist and is totally suitable for the job! Sadly, on the TFA Dostmann there's only a USB interface on the outside, and our Microcontroller does not support USB natively. So we opened the sensor and found a "CALIB" interface


Inside the TDA Dostmann CO² sensor, there's a CALIB interface, which consist of four PINs. GND and 5V can easily be determined using a multimeter with continuity test. These PINs are directly connected to the USB port and can be used to power our microcontroller. The other 2 PINs in the middle seem to send some data. After connecting them to a oscilloscope one PIN was obviously just clock, and the other data.

Test readings using SIGROK and a logic analyser can be found on b4ckspace/esp8266-co2monitor/sigrok-captures.


The signals look like they're just I2C, but actually they're not. Using the scope to decode the data failed - the same is true using the Wemos. After wiring up the sensor to the Wemos we used an external interrupt on the clock pin and a digitalRead on the data pin which is basically "bit banging".

The results were promising: Using a simple python script and the documentation from another manufacturer, we were able to verify the checksum. Also implementing the decode of the data was quite simple using the PDF:

case 0x50:
 co2Measurement = (data[IDX_MSB] << 8) | data[IDX_LSB];
case 0x42:
 temperature = ((data[IDX_MSB] << 8) | data[IDX_LSB]) / 16.0 - 273.15;

The last thing to do was connecting to wifi and publishing the data on MQTT. The source code can be found on GitHub and is licensed under the MIT-License.