Odd Si7021 humidity readings over 6 nodes in same jar

[WhiteHare]

As a cross-check on one another, let's compare notes.  My reading of the BME280 datasheet (https://ae-bst.resource.bosch.com/media/_tech/media/datasheets/BST-BME280_DS001-11.pdf) is that between 0C and 60C, the sensor's operating range (from page 8 ) is just what one would want: 0%-100%.  Outside that temperature band, the upper end of the humidity operating range starts to decrease, as per Figure 1.  So, if you were to put it into a freezer where the air had very high RH, it's operating range might be exceeded.  What happens then, though, isn't clear: does it just inaccurately report the humidity while under those conditions but return to "good as new" (well almost) when it returns to within its operating range, or is it ruined, or does it require a reconditioning as per Section 7.9, or... what?

It does give the caveat that for the given operating range, the environment must be "non-condensing," which I interpret to mean the sensor shouldn't have a temperature lower than the dewpoint.  Well, again, what happens if that were to occur?  Does it eventually return to normal, or does it require a bake/rehydrate cycle or... what?

I'm not trying to nitpick, but I am actually planning to put one in an ordinary home freezer that we've been having trouble with to better troubleshoot what's going wrong.  So, at least for me, the answers are not purely academic.  Likewise, in Joe's use-case of an outdoor TH sensor, his sensor might  (?) encounter those conditions during the winter. 

[joelucid]

I'm with you. Imagine how I feel after manufacturing with a sensor that has such a gotcha  .
But by default (assume you have no clue about this gotcha and are shopping for a nice TH sensor) would you rather trust SiLabs or a noname chinese boat fab that makes the DHT11?

It seems that measuring rh is just inherently hard. You need a sensing element that's exposed to the air but also extremely sensitive to contamination - that's just a bad combination. I don't expect the other sensors to be much better.

I certainly draw one conclusion: with rh motes it's not worth to worry beyond a couple of years of battery life. I don't think technology is where you can expect sensible readings after more than that.

[joelucid]

Likewise, in Joe's use-case of an outdoor TH sensor, his sensor might  (?) encounter those conditions during the winter.

Oh absolutely. The sensor only needs to be a bit colder than your environment for condensation to occur at 100% rh. It's easy for that to happen since it will lag air temp changes. That's why I actually like the si7021 for outside. If I'd use the bme280 I'd have to include a separate heating device.

[joelucid]

I've been running with heater for a month or so outside. It's working pretty well, the data seems sound. I'm running at heater level 1, which heats the sensor by about 5 celsius.

My old outside humidity sensor was a dht22 which I had just replaced 4 months back. That one is pretty much pegged to 100% humidity, showing the high humidity exposure drift.

Just now for example the dht22 measures absolute humidity of 7.89 g/m3, while the heated si7021 comes to 6.43 g/m3, a huge difference.

I found I had to correct the measured temp of the si7021 with heater down by 1 to get correct values for abs humidity. The app note from silabs also instructs to do something similar with heater on.

Joe