The results!

When your Pi boots, it should start recording air quality data. It won’t flash or bing or do anything science-y sounding. Your only chance to notice it will be once an hour, when the fan starts spinning to suck air into the monitor. (If you really want to check, you can log into the Pi over VNC and see if it’s working by searching through running processes.)

As it measures the air quality, the Pi is recording data to a spreadsheet in the AQM folder called “alldata.csv”. It is also trying to send data to my webserver, because I haven’t got around to fixing that yet—not to worry, though; no data is being sent because your Pi is not able to log into my server (and I’m not able to log into your Pi).

The Pi saves a lot of data to alldata.csv. It saves 40 measurements an hour (20 for each of PM2.5 and PM10). There’s no good reason for this, and I should make it save only an average, but it has proven useful¹, and there’s no discerenable harm (after 15000 measurements, alldata.csv is still less than a megabyte in size).

The number of measurements does make drawing inferences a little difficult. The trick is to use the moving average function on your spreadsheet software of choice. Chart 1 shows the AQM data for a week in May, 2018, in my  backyard. I’ve drawn two moving averages, one for each of PM2.5 and PM10.

¹ The Pi takes 20 measurements, once an hour. Weirdly, the first measurements are always lower than the others. I’m glad I kept all the data (over Mohammad’s objection) because we were able to find this flaw. It doesn’t make a lot of difference to the results because we are making relative comparisons and the error is consistent. But it’s there.

Figure 2: Measurements over one minute. Note that the first measurements are lower than the final measurements. This is a consistent error.