I have two D1 Mini’s, running Tamota, with DS18B20 temperature sensors. I have both of them right next to a Sonoff (far enough away to be influencesd with any radiated heat from the Sonoff), running Tasmota, with a BME280 sensor and there is a substantial difference in readings between them. Currently they are:
I know that I can set a float in HA to bring them inline, but the DS18B20 do not seem to be consistent in that I have adjusted the float several times but after some time them are misaligned again.
I believe that the DS18B20 can have different resolutions, using 9 up to 12 bits, but I have no idea of what Tamota runs at as default, or whether it can be changed, but even at 9 bits I would expect them to have a more consistent temperature.
Not sure which one you have but data sheets seem to point to a ±0.5°C Accuracy. This reflects the data you have. Worse case these two sensors could be 1 degree apart and within spec.
Even given a ±0.5°C accuracy, there is no way my living room is currently at the temperature they are reporting. The BME280 (top figure on table below) is more like the real temperature:
They are not exposed to any direct sunlight or any other heat source that the BME280 is not also exposed to. Below is a picture, the BME280 is in the sensor enclosure.
I just got three DS18B20 and ESP8266 with Tasmota and have a higher temperature in my living room (around 31°C instead of 23°C) and in my fridge with <10°C I got 18.9 / 19.9 and 21.2°C from the DS18B20
according to the images they look like original
I compared it with 433MHz-temperature-modules (also not really calibrated) - in my flat and the fridge - and the DS18B20 deliver around 7°C more than it should be
I could make offsets, but I wrote here because maybe there is a logic mistake what I made (because from bedfellow they also measured more than it should be)
But I will write a mail to the reseller, maybe he has an idea