I have 7 of them. From 3 batches. Separated by purchase period and channel. Two with quality silicone leads, 5 el cheapo black leads. All 7 agree with each other on the temperature, but they all are 1 C low when compared to multiple other thermometers (electronic and liquid, IR camera, expensive house thermostat, etc) which also all agree with each other.
Also, I seen it in both Tasmota and ESPHome.
There must be a systemic error, but where and how? Is it a know thing one must add 1 degree offset for some reason? As stated, it would be rather presumptuous to think the systemic error is not mine, but I would not mind if other people could please also check their DS18B20 against other (known good) meters.
Isn’t it standard procedure to never trust and always calibrate with ice water and boiling? Where boiling isn’t 100 but an ambient pressure lookup.
It may be useful in some cases, but it should at least be within manufacturer specs most of the time. And consistently ~1 C low, that is not a missing calibration problem, that is some sort of structural problem.
Are all [random] Dallas DS18B20 [clones/counterfeits] always a degree low [than the original (specification)]?
Not necessarily but the chances are not low
some of the counterfeit sensors actually do not work in parasitic power mode, have a high noise level, temperature offset outside the advertised ±0.5 °C band, do not contain an EEPROM, have bugs and unspecified failure rates, or differ in another unknown manner from the specifications in the Maxim datasheet.
Did you care to get your hold on a original one to compare them to your fakes/clones/counterfeits? Or did you just buy random? As a hint: From over 100 DS18B20 “type” sensor bought from aliexpress or ebay in the year 2019 and 2020 less than one (in numbers: ) were originals
Yepp, on the other hand if you got a bunch of them (or a calibrated reference sensor) you can easily bring them in line thanks to the glorious filters calibrate_linear or calibrate_polynomial.
Calibration of 53 pieces of DS18B20 Dallas sensors simultaneously. 32 of them waterproof model, so doing 2-point calibration for these:
[…]
Conclusion
DS18B20 solderable PCB (TO-92 packaging) are worse quality, and since they are not waterproof, they can only be calibrated to a reference, and then the measured offset can be applied (possible in ESPHome and Tasmota, also in HA with templating). Try calibration close to the temperature where it will be used intentionally, as it’s likely that the linear offset is still unprecise.
DS18B20 waterproof version are better, and can be calibrated using 2 points. Since we all learned in school that water boils at 100C, and water and ice present in the same time is at 0C (at sea level) it’s possible to do actual measurements of what the sensor sees in these practical conditions. ESPHome has calibrate_linear and calibrate_polynomialfunctions which allow for much more precise compensation than a simple offset.
SHTC3 is simply great. 0.1 or 0.2 degrees difference is acceptable.It’s not waterproof but it still can be tweaked with a linear offset, given that there’s a good reference to be compared with.
One could select a reference sensor from the pack of multiple DS18B20 waterproof sensors. From my 32 pieces I found one which was measuring 99,9C at water boiling, and -0,1 at iced water so I’ll set that aside and label it as reference sensor for the future.
Good point. I did not even consider this aspect, you are probably (almost certainly) correct.
In my defence: I was carefully wondering if I missed some systemic error of my own, and if so, what would it be. You found it: the systemic error is I buy from AliExpress.
(actually, the two silicone ones came from a local vendor, and were quite a bit more expensive. But I’ll assume they are fake too)
) were originals 
