Hi all!
Its been almost 2 years since I was last playing around with a bunch of emonTH’s, TX’s and the Pi as a base station… I was working on a project to use the emonTH idea as a sort of ‘learning’ tool in order to develop my own more large scale low cost temperature solution for datacenter moinitor applications…
However here I am again with the horn for getting my head into such things again 
Anyway… I have set two of my emonTH’s back up, these were the older versions with the DHT22 onboard. Now alarmingly I have noticed pretty major innacuracies with the temperature measurement of the DHT22 from both units, I have tried a few different DHT22’s I had lying around as well and the same thing happens…
Outside, when the temp was in the low 0’s, the DHT22 was reporting around 2-3c above actual temperature… So I wired in an external DS18B20 and that gives more accurate readings.
On my emonTH inside, I have just sat 3 different K type thermocouples next to it, each going into different meters and it turns out the DHT22 is reporting around 1.5c above what the K types are reading (which were all within 0.2c of each other).
So… I got my el cheapo IR thermometer out and compared the surface of the emonTH case to the surface of the DHT22 itself and sure enough… Its at least 1c - 1.5c above ambient…
Some research online tells me that self heating is common with this sensor… But is it really this bad? I fully understand you can offset this in code as a thermal ‘correction’, but I wasn’t expecting it to be CONSISTENTLY hot running across 3 different DHT22’s.
Anyone else noticed this?
Tom