> > > Moin Attila, Here is what my homebrew weather station actually uses: A HTU21D from TE Connectivity for temperature and humidity. These can be bought already fitted to a breakout board (I2C), interfacing with the RPi or whatever else. A BMP388 from Bosch for atmospheric pressure, again on a breakout board which has both I2C and SPI. All of this has been running fine since the end of 2018. As I write the BMP reads 1021.9 mbar, with the airport down the road reporting 1022. I remember checking this frequently in the first few weeks of operation, with never more than 1 mbar deviation. I haven't looked at it in years and am surprised it is so close still. The HTU21D says 5.58C and 87.7% rh, the airport 6C / 84%. As I said the Bosch probably needs to be replaced with a BMP390 nowadays. The greatest unreliability comes from the RPi which crashes once in a while because I am such a lousy programmer. There is even a time-nuts feature as the RPi also talks to a GPS module as well as a RTC - I think it is that very combination which makes it unhappy. Sorry about the confusion with the Bosch numbers in my initial reply. BR, Thomas.

I have been using a variety of sensors and instruments over the last 50 years. I do research on processes that occur in limestone caves and their atmospheres. I also worked in a government research laboratory on developing measurement systems. I stopped using dew point instruments because they were not suitable for ambient conditions. They have errors of fractions of degrees. Thin film sensors do no better that 1% RH and are not useable in the long term above 85-90% RH. Sampling is a major problem in measuring environmental (air ) measurement. I built a version of the WMO Reference Psychrometer, and I can measure air temperature in a cave to 0.01K and ) 0.1% RH (including 100%RH) for an average reading sample of 1m3 of air. These measurements preclude the presence of an observer. Having taken the trouble to get consistent results in these stable conditions I realise just how meaningless are the numbers that are bandied around of measurements in laboratories and work shops with active heat fluxes of kilowatts. I use a four terminal Pt RTD with a 5 digit HP ohm meter, in a stirred thermostatic water bath to calibrate sensors, and an ice point cell to obtain a 10mK zero reference. See: “Traceable Temperatures” J.V.Nicholas and D.R.White (1982) DSIR Bulletin 234, Wellington. I built my own barometric sensors using altimeter capsules and an optical detection system. I refer them to a laboratory grade Fortin Barometer. When considering monitoring ambient conditions you must take account of radiation on the instrument and always remember that an error of 1 centigrade degree of air temperature causes anerror of ~10% RH. Cheers, Neville Michie