Testing Various Low-Cost Hygrometers
See also the companion page on thermometers.
Described here are my own experiences assessing accuracy and calibrating the following devices. Results only represent my own specific devices which I have bought from various retail sources.
|Bosch Sensortec BME280|
|Measurement Specialties HTU21D|
|Silicon Labs Si7021|
|Aosong DHT11, DHT22, AM2302, AM2320, AM2321|
I would like to stress here that my interest in this is rather more about the methodolgy of testing and how accurately I can set up the calibration experiment using low or zero cost equipment than it is about truly calibrating these sensors. It would be presumptuous of me to suggest I know better than the metrology experts at some of the world's premier digital sensor manufacturers. Where I see a deviation between my measurements and multiple sensors I happily assign the error to my experimental procedure and look for improvements. My setup went through many iterations on that basis. Once I can start to see differences between sensors, that might be 'error' in one or other of the devices. Please do bear in mind at all times that a discrepancy from prediction could be fault in the sensor or in my test.
My conclusions are
- In a one-off test, the Bosch BME280 won easily as the most accurate. It performed in quite a different class from the others.
- Almost all do actually measure relative humidity to an accuracy sufficient for most realistic tasks.
- Individual devices will not necessarily match the advertised 'typical' performance.
- The actual accuracy achieved depends on your instrumental setup as much as the devices themselves.
- Reliability and long term stability may be more important to you than absolute precision.
Update (March 2017): I have a major revision of these conclusions in preparation. There is still a long way to go and data collection is ongoing, but I want to be as open as possible. Newer tests have suggested that in my previous tests I got lucky with an excellent Bosch specimen but possibly had a damaged Sensirion device. In that first one-off test, the BME280 was the clear winner, but I do not want anyone to interpret that as definitive. In my ongoing tests, the Bosch and Sensirion devices are indistiguishable at the limit of what I can measure. Work is still in progress.
This project has grown way out of hand. As my friend Bob said, I either need to get a good editor to sort out all these web pages or else get a life and stop writing them. The links below go to comprehensive write-ups.
- Direct tests of specific devices
- Compare AM2302/DHT22, AM2320, AM2321, SHT71, HTU21D, Si7021 & BME280.
- Compare DHT22(AM2302), DHT11 and SHT71
- Six DHT22 (AM2302) sensors. Includes the most complete description of my test methodology.
- Other observations on the sensors and the test method
- DHT22 (AM2302) reliability and failure of one device
- How fast can DHT22, SHT71, BME280 devices be read?
- Comparing the response speed for the HTU21D and Si7021.
- Effects of solution supersaturation.
- Absolute temperature accuracy for the BME280.
- Noteworthy idiosyncrasies: a few miscellaneous things worth knowing when using these sensors.
- First generation experiment with DHT22 only. Retained because some mistakes I discovered in my original setup may be of interest to anyone repeating the experiment.
2017-03-25 10:06 AM