During a previous article I’ve added CO2 level monitoring to my Home Assistant setup by using the SCD30 NDIR CO2 sensor. Although I haven’t tested a huge amount of air quality sensors I still found the level of accuracy of the SCD30 quite good. But how good is “good”? Let me showcase that by looking at the CO2 levels I’ve recorded through New Year’s Eve.

For starters the SCD30 air quality sensor is installed in the living room at the back of our TV. We started recording CO2 levels at noon (t0). The VASCO D350 ventilation unit is at “low-speed” mode and we had the front door open regularly. During this period only 3 people where inside the house making New Years’s Eve decorations. We notice how the CO2 levels build up until it saturates at around 1000 ppm. Around 17h our first guests arrive (t1). We can easily spot that event since from that moment on CO2 levels start to rise rapidly. After all of our guests had arrived and we all had our first couple of drinks it came to be that we were quite packed (we were with 13 in total). Without even looking at the CO2 levels I decided to ramp up the flow rate of the VASCO ventilation unit (t2). The above charts show that it wasn’t a bad decision to make since at that time the CO2 level had risen up to +2800 ppm. Due to the increased air flow this level dropped back quite a bit and after a while we reached acceptable levels again. However in the “medium-speed” mode the VASCO D350 produces quite a bit of noise in our sleeping rooms because it is installed on the same floor relatively close to them. At 21h15 (t3) my wife decided to put it back in “low-speed” mode since the youngest of our company were put to rest. As confirmed in the above chart the decreased air flow allows for CO2 to build up again. A bit later (t4) we started cooking. We regularly had one of the windows open, but also the kitchen’s hood was on. In effect our living room (which includes our open-space kitchen) is better ventilated and again this is confirmed by the SCD30 since CO2 levels start to drop. After cooking has finished (t5) the windows were kept closed and the kitchen’s hood was also turned off again which leads to increasing CO2 levels for the remaining of the evening. At some moments the CO2 levels even reached unacceptable levels again. Now after midnight has passed you’ll notice a small dip in the chart (t6). It is not some strange kind of artifact but can easily be explained: at that exact moment we went outside for few minutes to watch some of the fireworks around the neighborhood. Also one of the living room’s sliding windows was kept open and as a result CO2 dropped immediately with tens of ppm. Finally at 1h15 (t7) the eldest of the children were ready to catch sleep and all of our guests went home at that moment. This is easily detected by the SCD30: it shows us how the CO2 levels start to drop again. Me and my wife cleaned up a bit and soon after went to bed. At this moment the living room is no longer inhabited so no new CO2 is added. The VASCO D350 has free play and slowly – remember it’s at “low speed” mode – but surely brings our living room air quality back to acceptable levels.

As you can see the CO2 readings from the SCD30 are accurately enough to catch certain events that happened throughout the evening. Combining that data with other data such as the ventilation unit its flow rate we could probable create some software that could guess the amount of people inside the living room. For now I’m not convinced it is accurately enough to guess the exact amount of people because there are too many other variables involved (such as keeping a window open) that are not being monitored.

As a conclusion I’ve learned that when we have people over at our place we should give extra attention to improve the air quality. From the collected sensordata I could easily spot moments where the CO2 value reached unacceptable levels. To automate that process of constantly monitoring the CO2 level and adjusting the ventilation unit its air flow I could look into hooking up the VASCO D350 into Home Assistant. That may be something I try to accomplish later in 2022. For now cheers and best wishes to all of you.