Thanks for the link. I didn't mean to stray from the topic. I personally don't believe the FARS is the culprit here. My personal feeling is the 31 sensor is the problem, not temperature, just humidity.
We've identified 3 problems with the SHT31 humidity sensor so far in this thread:
- A wet bias that develops over time
- A wet bias that ships with new sensors
- A relatively small (3-5%) dry bias that's confined to readings 90% and above
Fan aspiration affects #1 and to a much lesser extent #3 too. In non-arid climates active ventilation bombs the sensor with moist air that leads to premature aging. It also elevates nighttime temperatures on clear nights which leads to lower humidity readings. Although, I would argue that any good passive shield will have the same effect during radiational cooling so long as it's 2 meters off the ground.
The clearest and most frustrating aspect of the dry bias is during fog events when most SHTxx sensors won't achieve 100%. This has nothing to do with your shielding choice. It's simply a limitation of the sensor. If your sensor reads 96.5% humidity in saturated air it is considered within specification.
By far the biggest problem is the wet bias (worst between 40-70%). This thread contains reports of max humidity errors in the range of 10-20%! That leads to wildly inflated heat indices in the summer. A survey of other VP2's online compared to nearby airports reveals a widespread problem that appears to stretch back to the SHT11 sensor.
EDIT 10:52AM: As for when the wet bias develops, that's uncertain. I have one good SHT31 sensor that's 7 months old and one with the wet bias that's 16 months old. Both have been in a mixture of FARS and passive shields. Others have reported receiving new sensors recently and all of them have run wet.