Wet sensors (or humidity sensor troubles)

Intro

Well there is no surprise: i went through a lot of temperature and humidity sensors. Some I already reported on, and some more i tried afterwards. And none seem to survive the exterior readout very well. The defects are mostly related to humidity, but not temperature.

What happens when they fail: mostly the humidity saturates to 100%. Here is my most recently replaced one from outside, a BME280:

After a while, the sensor reads close to 100% humidity and only goes lower when the sun is shining on my terrace and the sensor read a higher temperature. By my accounts this is one of the longer lasting sensors and probably the 6th one i replace outside over the last 5 years. My sensor sits in a Stevenson screen on a covered terrace, so there is no direct rain on it. I tried baking it in the oven at 50C for an hour but it had no effect.

Problem with sensors seems to exist everywhere:

https://forum.arduino.cc/t/dht22-vs-bme280-vs-si7021-which-can-read-near-100-humidity-the-best/519665/9
https://forum.arduino.cc/t/humidity-sensors/613874
https://github.com/sparkfun/SparkFun_BME280_Arduino_Library/issues/15
https://stackoverflow.com/questions/45643298/humidity-sensor-over-i2c-saturated

 

But why is this happening?

Fog, mist and other situations when there is close to 100% humidity and it condenses on the sensor. Thus the sensor can be all wet. Lets look at what datasheets are showing.

Bosch BME280

This is by far my most favourite sensor, as it provides temperature, humidity and pressure readout. First, the datasheet says the sensor can operate from 0 to 100% humidity, but then it says that long term stability is only valid for 0-90% humidity range. And most importantly, in a footnote:

and then, which can imply that condensation cannot be allowed during operation.

Sensirion SHT21

Has a similar Operating range for humidity: 0-100% and a long term drift of max 0.5% per year. So it should not fail so soon, right? Well, another footnote may actually tell why:

 I hope it is clear that long hours of condensation may occur during colder climate.

Silabs SI7021

Is one of the first one i got my hands on quite a few years ago. They have also proven unreliable outside. But the datasheet is much clearer here, which is why it was clear from the start they will not be good outside:

HTU21

Now this sensor promises a lot more, same 20-80% normal operation range. But the extra: recovery in 10 seconds after 150 hours of condensation?This sounds like the ideal sensor for outside. Did not have one outside, but I had one in the bathroom that failed recently. Here is how, all of a sudden the temperature readout dropped by 8C  and the humidity jumped to 100%.

 

There is also a heater on the sensor, but that is not mean for fixing the large humidity but “The heater is intended to be used for functionality diagnosis: relative humidity drops upon rising temperature.”

Conclusion

Clearly none of these sensors are meant to operate in condensing environments. So they are not OK for outside sensing or rooms like bathroom. I accepted the fate.
Could a sensor with heating be better? No idea.

These sensors will fail soon when used outside in condensing humidity. I just made sure to have a pin header so i can easily replace it. The problem is when to consider replacing, when should I stop trusting the humidity data from the sensor?

Know a digital, low power sensor, that is meant to operate outside with condensation and recover from that? Please let me know.