Accuracy of the internal temperature sensor

[espruino-discuss3]

Posted at 2016-09-26 by @yerpj

Since 1v48, we can call E.getTemperature() in order to retrieve the chip internal temperature. I just read from the datasheet

the internal temperature sensor is mainly suitable for applications that detect temperature
changes instead of absolute temperatures. If an accurate temperature reading is needed,
then an external temperature sensor part should be used. ```
As well as in the reference manual: 
```The temperature sensor output voltage changes linearly with temperature. The offset of this
linear function depends on each chip due to process variation (up to 45 °C from one chip to
another).```

I actually never tried to quantify the real accuracy of this sensor. Does anyone have any idea on the actual accuracy ? Does anyone already used an Espruino that has a temperature offset around 45°C ?

[espruino-discuss3]

Posted at 2016-09-26 by @gfwilliams

I haven't actually noticed much variation between them. I think it's ST covering themselves.

They seem to read a few degrees higher than room temperature, but that's usually because the chip really is hotter if you've got it connected to USB and it's not sleeping. It'd be interesting to connect a DS18B20 up and run some tests comparing the 2 temperatures when the chip is asleep most of the time.

---

Posted at 2016-09-26 by @yerpj

It was also my assumption.
I already compared this sensor on my 3 Picos, I got a dispersion of +- 2 degrees, which is quite good for most of applications.

I propose to keep this conversation opened in case of someone would notice problems with the offset of the internal temperature sensor.

---

Posted at 2016-09-26 by @gfwilliams

Yes, absolutely.

I guess it's one of those try it and see things - if it works, it's going to keep working. ST aren't saying that it'll change the reading over time - so if your Pico reads 2 degrees high, subtract 2 degrees and it'll be spot on from then on.