If you are sure, mark it as solved
Does anyone have an explanation about that 5V "scale" on a 3.3V Arduino ?
It's not a 3.3volt-logic processor.
It seems to be a 328 clone running at 32Mhz.
Leo..
Ah, OK. The thermistance is powered by a 3v3 pin but the Arduino is power by a 5V PS. That makes sense.
The thermistor should be powered by the MCU supply,
so powering it with 3.3volt is wrong in this case.
Leo..
Can you please elaborate on this? Plugging it on the 3.3V Arduino pin seemed to work fine for temperature sensing.
I guess it's to take advantage of the full 12 bits ADC resolution and lower the signal/noise ratio but maybe you could tell us if there's more...
If you do that, the ADC results will be referenced to a 5V that may vary a few tens of millivolts, while the 3.3V remains the same since it's regulated.
That will produce an offset error, it will be sensitive to supply voltage variations.
A ratiometric sensor, like a thermistor, should be powered from the same supply that Aref of the A/D is connected to. The Aref of a 328 is by default connected to 5volt. (I guess your clone is the same).
If you do, then voltage on the thermistor increases/decreases with the same and opposite amount of the sensitivity (mV per A/D step) of the A/D. Resulting in the A/D value being stable.
If A/D and thermistor are powered from different supplies (a stable regulated3.3volt and a potentially unstable USB supply), then that ratiometric behaviour is lost. Resulting in a varying temp readout with supply variations.
Leo..