Variance between sensor A & B

I have a new Zen monitor. How much variance between the two sensors is expected and/or acceptable?

This is my average variance on downloaded data, using 427 measurements at 10-minute intervals. I calculated the numbers below based on the a and b values.

| Variance | A vs B |
|27%| 0.3_um_count_a vs b |
|24%| 0.5_um_count_a vs b |
|21%| 1.0_um_count_a vs b |
|25%| 2.5_um_count_a vs b |

On the PurpleAir map the confidence is 100%

This is an indoor monitor. I’m using it to try to locate the source of second hand smoke coming from my downstairs neighbor.

Hi @howardz

According to the laser counter manufacturer, Plantower, the expected channel output deviation is as below:

  • ±10μg/m³ @ 0~100 μg/m³
  • ±10% @ 100~500 μg/m³

It sounds like your device could be outside of that range. If it is, please send an email to with the Device ID of of your sensor.

I was using the particle count. Does Plantower have a tolerance for that?

When I use the “cf_1” datapoint – which is the Plantower approximation of μg/m³ – the variance is less than 1.


With respect to the Plantower folks, or perhaps with disrespect, their statement that Plantower can only get to within 10 ug/m3 is wildly mistaken. In fact, based on 14,400 indoor and outdoor monitors over 5 years, the mean precision of those monitors (based on abs(A-B)/(A+B) was on the order of 10% for ALL concentrations. I used a 20% cutoff for 85 million hourly outdoor values and 78 million values beat that cutoff point. The median outdoor PM2.5 level was 8 ug/m3, and only a tiny percentage exceeded 100 ug/m3.


Hi @howardz

Plantower does not have any suggested variation for particle counts. Have the counts continued to consistently vary from each other?