Aaron, the SD data is uncorrected and unmodified in all cases. API data is also not corrected in any way, so you will get what the sensor sent with no conversions or corrections applied. On exception is the CF=3 field.
Aaron and Adrian–
If I am understanding what Aaron is saying, this is bad news for people wanting to use the PurpleAir data. The “new” 5003 makes a completely arbitrary change to the >0.3 um/dl, dropping it down by a huge margin. This affects the ALT-CF3 data in one major way: it reduces the contribution of the 0.3-0.5 um/dl (which you get by subtracting the >0.5 u/dl from the >0.3 um/dl) from typically about 16% of the total PM2.5 value to < 1%!. I have been running two “new” 5003s side by side with three “old” 5003s for a couple of months now. But it did not take long to see the effect.
Adrian, do you think this change will be permanent? If so, we would need a marker to identify the “new” 5003s. It may be time to look for an alternate sensor.
This change with the 5003 laser counters is not permanent. We received a batch of laser counters with this issue, but the ones we are receiving and sending out now are the standard ones we were using previously. We check all laser counters for this issue before they are sent out.
1 Like
Thanks, Ethan for that news. I feel better.
Good news that this is a temporary problem. It would be really great if there were a way of ordering sensors directly from PurpleAir in countries other than the US (e.g. in Europe). I know it’s possible but the customs and shipping costs make ordering sensors prohibitive for non-US based customers.
We are working on a solution to this with a 3PL and soon you will be able to order from various shipping locations.
1 Like
Adrian,
You mentioned “We identified a way to id these new devices in the field both from the PA firmware (with a future update) or directly in the data they provide now.”
Do I read this correctly? Is this a temporary problem that will be solve through firmware or must the affected sensors be replaced?
Hi Ethan,
Our group is following this issue of changes in the PMS5003 because of the strong relationship between the Channel 1 [ >0.3 um] value and the submicron aerosol scattering coefficient. How have your latest batches of PMS5003 performed, particularly the first size channel?
Thanks,
Jim Ouimette
Hi Dennis,
These sensors can’t be changed via firmware. Contact us via email (contact@purpleair.com) with your device IDs if you’d like us to see if any of your sensors are affected.
Hi Jim,
Since discovering the issue, we have been able to successfully vet out these sensors. Our recent batches of PMS5003 have performed very well, particularly at lower size measurements (i.e. 0.3 µm, 1 µm, etc.).
A recent study by the University of Utah evaluated this new version of the Plantower PMS 5003 in both the laboratory and field. These new sensors were deployed in PAIIs between approximately June 2021 through January 2022. As of May 2023, approximately 10% of outdoor PAII sensors are this newer type, and they are biased low by approximately 3 ug/m3 (CF=1). This low bias is unlikely to affect measurements during particulate matter pollution episodes, like wildfire events. However, this bias could be potentially important for individuals using these new sensors for health, indoor air quality, or environmental justice studies. These new sensors can be identified by the ratio of PM counts (new sensors have 0.5 counts / 0.3 counts = > 0.4), similar to PurpleAir’s original post.
The study also proposed the following correction relationship for the new Plantower PM2.5 measurements below 16 ug/m3 (CF=1):
PMS-corrected = 1.348 × PMS-raw – 2.797
Applying this correction reduced the bias between the new and the original version of the Plantower PMS 5003 sensors. Measurements above 16 ug/m3 appear to remain unaffected.
The study can be found here
https://www.sciencedirect.com/science/article/pii/S0021850223001210
Or feel free to contact the authors if you’d like more detail.