Beautiful, isn't it?
The complexity of the circuit has increased considerably over my first Experiment: Power supply, NodeMCU, to the left an I2C-switch and 6 (six!) sensors! Why? Because We Can
And it makes a nice chart, see below.
Now, actually, I was disappointed with the performance of the second DHT22, showing much different humidity readings than the first one. So I got a BME280 in addition to compare and that was reading much lower than the DHTs. What the ... So one more BME from a different source and this one sits somewhere in the middle, seems reasonable. But still, I had hoped for better results for the Bosch devices. After some more research I got another device, or rather: a bundle of 4 SHT31-D and a I2C-switch, you never know what that stuff might be good for. After putting feed on the tiny boards I tested them all on a simple circuit and yea, very consistent readings across the bunch!
But comparing measurements became more and more difficult. Time for that TCA9546A to show me what it is good for! So I present 2 BMEs and 2 SHTs on the I2C bus via the switch and 2 DHTs connected directly to IO pins on the 8266.
Readings are again pulled by the OpenHab server and made into that chart. Where you can see that the SHT31 are rather close with one of the BMEs, two others a bit off and one DHT22 seems to be broken or whatever.
Looks like I will go with the SHT31 for the actual implementation later this year.
Observer language filtered content at https://firstname.lastname@example.org