Public Lab Research note

  • 1

Two weeks with a Speck

by mathew |

Carpentry. 80 times worse than rush hour.

What I want to do

Illah Nourbakhsh and Beatrice Dias from CMU CREATE Lab sent me a Speck particulate monitor. I plugged it in and let it run.

My attempt and results

The first thing I noticed is that it is minutely sensitive to its placement. It draws air in the side and out its bottom, and needs to be placed on a flat surface. Also, as the instructions mentioned, it didn't like direct sun. The measurements went up in sun.

There was a definite correlation between the rush of morning traffic and particulate levels. They would roughly double. But its also pretty cloudy in the morning, and I should really look at humidity too before making a judgement but... I don't have any software package that stands up to the CSV file produced by the speck. Excel, LibreOffice, both crapped out when I made a graph. So I'm going to visualize the data using this screenshot from a frozen LibreOffice.


I was hoping to present the dates out of their Unix time stamps and into a regular date format, but should've been faster at the screen captures.

Maybe I'll try again sometime with some other software. here's the CSV file and the ODF spreadsheet with a column of human readable dates.



Taking it Apart

the scoop underneath is where air enters.

USB power/data and the small fan exhaust on the side.

Air comes in the rectangle in the bottom.

Notice how the Syhitech DSM501A sensor's ports line up with a baffle to direct air through it, and the intake port lines up with the bottom of the sensor.

Still sawdust collected on the inside of the fan. sheesh. That was last week.

Everything is press fit in place, after the three screws holding the two parts of the case are out, everything slides in and out easily.

Why I'm interested

This thing is really sweet- very finished, easy to use- it made me immediately sensitive to changes in particulates. Also there's a good chance its as good as expensive things.. Just look at the study linked in the Dustduino wiki page.

Also, check out this cool visualization of air quality around Pittsburgh.

air dust open-air particle-sensing particle-monitoring particulate-monitoring dust-monitoring openair


Thanks to Bea at CMU CREATELab for the blog cross-post !

My followup post on moving Speck data into Fluxtream to visualize data.

I couldn't get Excel to parse the time stamps either. I averaged every 60 particle counts, so I assume I got a number for every minute. Here is a graph from Excel for approximately the same time range as your graph above.

do you want me to make an .xls file from the .odf file? i'm not sure the date translation will work, but the dates are int he odf I made.

They sure are. Excel opened the ods file without too much trouble, and the timestamps are right there. I was just curious to look at the data. Don had the Speck running all day Sunday at the Barnraising, and it really responded to cooking bacon. Maybe he will parse the time stamps and send me the file.

you can always try Fluxtream, although I don't know how you get the data in there on windows. actually I do-- make Don do it!

Let me know if you need help with data conversion and/or uploading.

You must be logged in to comment.