Hi all,

In Finland we have this thing called Winter. And it surprised us again, just like every year before.

It surprised some of our sensors too.

Temperature outside has gone down to about -20 celsius, so it removes humidity from air.

Images included from grafana, storing logs of 20 sensors in total for about a year now. Five of our H&T-sensors that we have been using are now showing zero percent humidity and I'm really having thought about trusting these values.

All images have temperature displayed in green and humidity with yellow.

I also did a spot test today (3.1.2024 1500) with Shelly H&T Plus, got ~20.5'C and 8%. Just to get another spot reading without knowing how calibrated that one is.

one has interesting humidity

These are installed in different rooms that are used daily as offices. All these use USB-power supply. We have no reported blackouts on the grid.

I assume that I would have noticed 8% change when testing these.

Do these sensors drift, or what should I make out of all this?

Did I miss some notes about calibration or should I think that sensors require replacement?

Your thoughts appreciated,

And without touching one of sensors it seems to have stabiliized a bit.

Now fifteen sensors have shown humidity of 0% or 100% without reason.

In touch with service desk. Asked to factory reset devices, but it didn't yet resolve problem on sensor that I tried it.

Thanks Scheuerer for the comment,

Today, it was -19'C again on my very frosty walk to work. There is a river nearby, so sometimes nice looking vapor clouds rise from warmer water to the air. Cold weather and snow make nature look very nice, but it's not meant for unprotected sensors or people.

-

However, in this case, we have no HT's directly outside. About 13 are in warehouse or office environments and rest six are in more requiring one. (usually staying under 40'C and over 25%RH at least by their sensor values)

I had one sensor gone bad just like ones here (RH jumps from 0 to 100 and back randomly or stucks to 100), but that was ruled out of this thread, because it was highly probable to have been caused by water damage from condensation. New one has a different location and better protection against it.

If the sensor element were a serviceable component, it would be a different story. However, having to throw away 15-20 sensors every few years doesn’t sound ideal.

I've now had some contact with customer support, but they seem very limited in their time, so I have had no luck with getting anything that would resolve this. I’ll still give them chances to recommend something to resolve this problem, as they may be best equipped to know their sensors.

-

If nothing then, maybe I need to revisit some of my old projects with AM2321 and BME280. I remember some years back reading a lot of comparisons when I was into sensors.

If I need to get involved, i'll probably try ESP8266-01 with a sensor like one you recommended.

But now I'll just work to get tickets done. Maybe one of the broken sensors could be disassembled and put under an expensive microscope. (to get some payback for the all time invested already. ;))

BR,

am