PMS5003 PM2.5 Sensor Accuracy?

So I have been experimenting with the Plantower PMS5003 sensor to measure air quality.

But I have no way to know what the default values should be.

As usual, everyone else who reviews one of these things only ever does so with a single unit, and therefore does not test repeatability.

I am testing three units, all from different sellers (but all the same Plantower PMS5003) and getting inconsistent results.

The first two units somewhat agree with each other, but the third gives what I would guess to be more appropriate results.

Here is a table with a random "average" base reading from each of the three sensors, taken when there was no additional added particulates in the air.


This data collection was the result of letting them run overnight in the house.

For reference, these are the values the sensor gives us:

All three respond well to burning incense or solder, at which time all the numbers spike very high, however I am hesitant to believe that my house, which is dusty as hell, would be getting so many 0's in those columns for the base readings. It just seems more appropriate that the third set of values would match my house. (But I am basing this solely on my intuition rather than any quantitative data.)

It would be nice if the manufacturer gave any kind of calibration instructions or average values- but they don't.

Included are the serial dumps from the sensors. (Could not upload .txt files for some strange reason.)

pms5003_incenseTest_unit1.h (23.1 KB)
pms5003 test3sensors.h (3.8 MB)

I should also mention that one of the suspected bad sensors also zeroed out completely at multiple times during the run- returning 0 values for every value (which I assume should never be possible). One of those times is visible in the first dump file.


Am I correct to assume that the ones giving 0 values for the PM readings are faulty? (I realize they do concur with each other, however 3 units is still not a very good sample size for such a product.)

And can anyone share from their own experiments what the base values should be reading in the average home interior?

Thank you!