WeatherShield: SI7021 humidity readout offset

[rusheee]

Hi All,

I am building a wireless sensor network for logging temperature and humidity around the house. Comparing the humidity data from the SI7021 sensor on the one Weathershield i have with other sensors (SHT21) and also a calibrated instrument, i have found the reading from the SI7021 to be 6% higher than it should be. I'm using the SI7021 library from the Github repository.

Does anybody have reliable references for their sensors to tell if they show an acceptable offset that is within spec? I haven't cross-checked the code of the library yet, but i assume the calculations are according to the data sheet recommendations?

[jra]

The Si7021 data sheet indicates that the worst case error should be less than 4.5% @100% humidity and less than 3% error below 80%, see https://www.silabs.com/Support%20Documents/TechnicalDocs/Si7021-A20.pdf for details.  I would try using the saturated salt technique (http://www.omega.com/temperature/z/pdf/z103.pdf, http://nvlpubs.nist.gov/nistpubs/jres/53/jresv53n1p19_a1b.pdf) to calibrate your sensors as that is probably the easiest way to verify them without a lot of specialized equipment.

[rusheee]

Hi jra,

It is a very good advice to check the sensor readings using a saturated salt solution and also calibrate them this way. The documents you linked are a great reference.
Since i had compared the readings of my SI7021 sensor to those of a calibrated instrument, i am very confident that they are indeed out of spec (+6% vs. +3% as allowed below 80% r. H.). That error has been pretty consistent for over a week now, so i also does not seem that it is an error that would decrease by conditioning the sensor as described by the data sheet.

Has anybody else found the humidity readings to be out of spec?