Best way to deal with accuracy/inaccuracy of humidity sensors with disparate readings?

I have 2 Aqara Temperature and Humidity Sensors and 3 ESPHome Temperature and Humidity Sensors. I have them all sitting next to each other to compare the values.

I made the ESPHome sensors using this Tutorial,
with these Boards,
and these DHT22s

The temperature readings match pretty closely I think:

But there seems to be a very large discrepancy in the humidity readings:

The difference is over 10%.

What would be the best way to handle this difference and know what to trust?

I bought the sensors about 2 years ago, and I notice the product page for the DHT22 only mentions temperature. I thought all DHT22 sensors would do both, but is it possible that it’s only returning a “fake” humidity reading?

Thank You,


It’s possible?

Those seem to be in two distinct fairly tight groups? Am I reading that graph correctly?

Does one of those groups match what you consider reality? If so I’d probably try to template sensor the ones that don’t match with some kind of correction if you can figure out the math (looks like a simple +/- amount)

1 Like

Yes, I apologize for not making it clear.

The 2 Aqara sensors are the bottom group of the humidity graph, and they agree with each other. The 3 ESP sensors are the top group, and they seem to agree for the most part.

I don’t know if I trust my own judgement of what’s right, and I worry maybe both are wrong.

I also wasn’t sure if a constant offset would be accurate in the long run. but it is worth trying.

Calibrate it. There an offset and probably fancier lambda stuff you can do in ESPHome, and all kinds of corrections you can do in HA.

1 Like

A couple of resources that will be of use.

Humidity Fixed Points of Binary Saturated Aqueous Solutions - PMC - This one gives a value for Sodium Hydroxide: 8>9% @ 20°C plus a formula to calculate the RH at different temperatures.

1 Like