Leveraging Tasmota for Indoor Aerosol Science

[Grumium]

Hello everyone,

I’d like to briefly introduce my current Tasmota project, which was developed as part of the EU Horizon project EDIAQI (Evidence Driven Indoor Air Quality Improvement) at Leibniz Institute for Tropospheric Research (TROPOS) :

AQBIE – Air Quality Beacon & Immission Evaluator

15 of these devices have been recently installed by me and my colleages in households in Zagreb, Croatia. The goal is to collect high-resolution data on the personal exposure of children to particulate matter and air pollutants in their homes, to better understand sources of indoor pollution and personal ventilation habits. Another important aspect is raising parents’ awareness of indoor air quality.

Why Tasmota?

Commercial devices are often expensive, inflexible, closed-source, noisy, and do not measure particle size distribution, which is crucial for deriving scientifically meaningful data. I have been working with Tasmota for quite some time and was therefore convinced that we could address our current research questions using Tasmota as a solid base, and without reinventing the wheel. In the end, it was still a lot of work. One of our key goals was to create an easy-to-use device that integrates well into children's rooms, while also enabling us to receive live data via MQTT and perform remote software updates if needed. Development started back in February and included everything from hardware and software development to 3D-printed enclosures and custom adapter boards for the various sensors and MCU. All devices are currently up and running, and the measurements will continue until the end of November.

Technology

Our platform is based on the M5Stack Core2 v1.1 and uses sensors from Alphasense, Sensirion, and Bosch Sensortec, among others. In the course of the project I developed new drivers for the Core2 v1.1, the SEN66, BMV080, and Alphasense OPC-N3. Only the combination of all these sensors provides a complete picture of particle load in the 0.3–10 µm range. A single sensor, as found in many commercial devices, is insufficient. For the web interface, we built upon Tasmoview (by @wjohn007) to provide users with a modern UI and easy access to live data via the web interface:

Several widgets for the front display were newly designed using LVGL and Berry, including an AQI scale, a radar plot, a setup page, hidden touch buttons to leave more room for the widgets, the UI can be customized (dark mode/bright mode, dim time, ...), etc.

AQBIE is intended as a platform for further IoT sensor and air quality projects at TROPOS, and we have no plans to commercialize it.

Many thanks to @sfromis , @s-hadinger , @Jason2866 , and of course @arendst for your work on Tasmota and for patiently answering questions on Discord and here.

Let me know how you like it.

[Jason2866]

Thx for sharing this great project. Your experience about commercial devices and the used sensors are the same i made.
Good commercial devices are really expensive and all available consumer devices are more or less useless.

[Grumium]

Hey @Jason2866,
Thank you! From an intercomparison campaign in our reference laboratory, we saw that most companies’ expertise lies primarily in gas measurements or weather stations and not in aerosol particle physics. Particulate matter measurements are not their core business, but rather a sideline, which often leads to poor performance and calibration. Insufficient attention is given to impaction losses and sedimentation effects in inlets, and most companies show little interest in equipping their devices with optical particle counters. That's why we decided to combine recent nephelometric sensors with a relatively low-cost optical particle counter and we hope that we will learn something.

[sfromis]

I suppose that a lot of the reason for using sensors of "modest quality" is when the intended audience is for home users, where "cheap stuff" is a priority, and caring more about high/low instead of some accurate number. Like when I measure CO² at home, and can't honestly say that it is important to me if the number is 1100 vs 1150 ppm. And I've chosen not even trying to deal with various ideas about a number for "air quality", without a clear definition.

[Grumium]

Hi @sfromis,

I completely understand your point. Affordability is often the main priority.
For people who don’t know that 1000 ppm CO₂ is already a high value, or that 50 µg/m³ PM10 is the EU daily limit, an Air Quality Index is still a useful tool for assessing air quality. We have adopted the following color scale because it is fairly well established and is based on the U.S. EPA 2006 standard: https://atmotube.com/blog/indoor-air-quality-index-iaqi.
There are publications, such as: https://pubs.aip.org/aip/acp/article-abstract/1808/1/020043/795703/Development-of-indoor-environmental-index-Air and the Atmotube algorithm is out there on Github: https://atmotube.com/blog/what-is-air-quality-score-aqs
So there are, in fact, some standards and definitions. Of course, you can also create your own scale, as long as you transparently explain how it is constructed, that’s perfectly fine. Naturally, there are other issues to consider, such as the fact that the TVOC value is actually not very meaningful as it doesn't say anything about health risks, and there are no legal thresholds for PM1. Additionally, some sensors attempt to calculate an AQI internally using black box machine learning algorithms, which can make things even more confusing and is nonsensical when you want to work with the data.