Over the past week I've been comparing my primary SHT75 sensor with my new Davis 31. The Davis 31 has a manufacture date of June 2019. The sensors are co-located and the Davis 31 has been outdoors only for this week. The temperature of the two sensors is nearly identical, so my comparison is in RH.
The SHT75 has been in use outdoors for 6+ months. For most of the week, humidity has been high up near saturation as is typical this time of year in western Washington state. The Davis 31 has reached as high as 98%. The 75 will hit 100%. From about 87-88% and above, the 75 reads about 1-2% higher.
Yesterday we had an unusual down-sloping east wind off the Cascades and this dried out the air very quickly after prolonged days of high humidity and some fog. The humidity dropped from a high of 89% just after midnight to the upper teens with dewpoints in the single digits to around 0°F. Temperatures were in the upper 30s to mid 40s.
I don't quite have the elaborate set up as some of you, but I have data in Cumulus for both sensors every 10 minutes. The comparisons are based on the raw readings of both sensors. After analyzing the data, the Davis 31 read on average 3-4% higher than the SHT75 in the humidity range of the upper teens to the 30s. The range in difference was the Davis 31 reading from -1%RH to +8%RH, but +3% to +5% was typical. I did notice that over time in the dry air, the Davis 31 did seem to improve a bit reading only 2-3% higher on average. The lowest humidity both sensors reached was 21% for the SHT75 and 23% for the Davis 31. The highest humidity for the day for both sensors was 89% for the SHT75 and 88% for the Davis 31.
The SHT75 read consistently about 3-4% high in the same range compared to the Belfort 566 I use for the best true reading I can afford. What I report online has a humidity calibration in Cumulus of 1.061x-5 which matches well with the Belfort. So the Davis 31 is about +6% to +7% RH from true as best I can tell in this humidity range, sometimes a little better, sometimes a little worse.
Just thought I'd pass along what I saw and it is consistent with past comparisons I've done as well as the Prodata studies. Not sure if Jerry is still using the SHT75 for comparison as well or what he is seeing more generally. I just may have bad luck with these Davis sensors.
The picture was fairly typical. Sensor 2 data is the Davis 31. The Belfort at this time was reading about 19%.