hi @guolivar, thanks a lot for the feedback, much appreciated. i understand that this is a practical, empirical solution, that could be correct most of the time, but still it is quite a simplification. i had a look at the 2 scientific articles that the DustDuino wiki refers to ('data quality') but they don't provide any detail about the sensor's P1 and P2 channels and how the measurements should be calibrated. the DustDuino sketches use an average radius of 0.44 micron for the P1-P2 range (diamter 1-2.5 micron; a bit odd) and radius 2.6 micron for the P2 range (diameter >2.5 micron). i think this is the correlation that you refer to above; is this based on those scientific articles?
fyi, i've summed up a couple of remarks on my blog, http://lantaukwcounter.blogspot.hk/2015/10/pdd42-sensor-can-it-measure-pm10-and.html
Is this a question? Click here to post it to the Questions page.
Hi there @tomtobback !
Just to clarify ... I'm in no way associated with the DustDuino team ... I'm just a "concerned citizen" ;-)
I am an air quality scientist at New Zealand's National Institute of Water and Atmospheric Research (http://www.niwa.co.nz) and I'be been using low-cost air quality sensors for a few years. (google "PACMAN niwa" or "ODIN niwa" for some info on my team's work)
Moving from counts to mass is always very tricky and you need to make a number of assumptions, not only about the density distribution (specific mass of the particles) but also about the composition of those particles. In typical urban settings, where there are several sources present at any given time, it is particularly difficult to accurately and reliably convert between the two metrics as the average size and composition changes every instant and that has nothing to do with the quality of the measurements ... I remember that back in Sweden we put a very expensive laser spectrometer (think of the Shinyei but so sensitive and accurate in its sizing that has 32 channels reporting different size fractions each) in a subway station and when we went to analyse the data it didn't make sense ... in the end we found out that the sizing algorithm based on light scattering wasn't "calibrated" for mostly iron particles (as the wheel wear from the trains) and we just couldn't reconstruct the size distribution reliably.
My advise is always that there is no perfect measurement and that you shouldn't be afraid of making assumptions, just document them and try to evaluate their impact. But, above all, be very clear about the purpose of the measurement and choose the instrument accordingly.