Low-Cost PM2.5 Monitor at Make-Think-Code

PM2.5, fine particulate matter, is the air pollutant most associated with damage to health, yet many aspects of its distribution are not yet understood. So, when the Make-Think-Code Lab announced a "Sensing the Environment" hackathon we were excited, as Make-Think-Code is:
...a technology-focussed lab at the Pacific Northwest College of Art that brings together members of Portland's vibrant creative, tech, civic, and educational communities to explore the powerful role that creativity and technology play in the search for imaginative and impactful solutions to complex and urgent problems.
So, the hackathon provided a perfect opportunity to collaborate on the development of a low-cost PM2.5 outdoor air quality monitor.

At the hackathon, we set ourselves the goal of designing - and prototyping - a monitor that would be easy for citizen science groups, individuals, and researchers to assemble and deploy. Since several PM2.5 monitors are already available on the market (e.g. Dylos, Purple Air, and Shinyei), we focused on designing a sensor that would not need to depend on nearby A/C power or internet connectivity, allowing us to monitor under cloudy conditions, in shaded areas, as well as in parks and open spaces located away from a power source or internet connection. Designing and prototyping the monitor meant working on two aspects: the monitor itself, as well as a design and prototype of a shelter for the monitor.


For the monitor assembly, we chose a Plantower 5003 PM sensor; mainly because it showed good performance in tests conducted by the South Coast Air Quality District in California. We used a Teensy 3.2 Arduino controller, a DS1037 real time clock from Sparkfun for getting the observation timestamp, and a Teensy audio shield for holding the SD card (used for logging the data). In the very first prototype, we used 3 AA batteries to power the monitor assembly. However, the monitor ran for only 6 hours on the 3 AAs. In our current prototype, we use an 8000mAh power bank which powers the monitor for ~30 hours. Under this design, a week to 10-day long measurement campaign will require several power bank changes. We expect soon to test a 26000mAh solar-powered power bank in the hope that it can run for a 1-week campaign without a battery change.


The shelter design has some interesting constraints: it needs to be reasonably weatherproof as it may have to sit out in the sun or rain for weeks. At the same time, it must also allow for battery changes and data downloads from the datalogger. Cost is also concern, as is simplicity of manufacture and assembly, since our goal is to have large numbers of sensor kits available for assembly by interested community members. Again, our designs have shifted over time, from a Stevenson Screen design based on the Canary through two acrylic versions. Our current design is drafted in Fusion 360, which lets us parameterize to allow for easy changes in material thickness or battery compartment size. The door (in gray) lifts out to give access for changing the battery or the SD data card.

The shelter is designed to be laser-cut out of acrylic or other weatherproof material. The laser-cutting pattern is first generated as a CAD drawing from the Fusion 360 model, as shown here. nothing A second pass in Illustrator or the equivalent may be necessary to get the parts arranged efficiently for the particular piece of material being cut. Once the parts are cut out, the circuit board and sensor get clipped to the center divider and the whole thing will slot together in moments with no glue or fasteners required.

We are looking forward to completing and deploying three to five prototype PM2.5 monitors and beginning to design the back-end database and visualizations by January 2018.