nRF51822 TEMP internal temperature sensor characteristic

[Bouke]

I'm trying to use the internal temperature sensor of the nRF51 to know the die temperature. Its accuracy is specced as +/- 4 degC, even +/- 8 degC at the extremes. Does anyone know more properties of this uncertainty? Is it a fixed thing, that I can improve by calibrating an offset? Or is it noisy, something that I might filter out using multiple measurements?

Edit: See my own answer, the offset I measure is unusually large. Any thoughts where this might come from?

[Susheel Nuguru]

Bouke, we still have to consider that the external temperature sensor measurement is still not the same as the internal sensor measuring die temperature. The specced accuracy calculations was made specific to the die temperature at a very large variations of temperature it is working in. How much of the time is the system kept on? maybe we can get a better result by trying to keep the chip in sleep mode most of the time (like a minute) and just wakeup to read the temperature very fast and go back to sleep. That COULD ensure that the die temperature is much closer to outside temperature and then we can use the external temp sensor reading to compare.

[Bouke]

Hi Aryan, a strategy similar to this was implemented, although some communication takes place right before the measurement. I would expect (and accept) perhaps a slightly higher die temperature, but what surprises me most is that it is fairly constant over the whole temperature range. If the environmental temperature is -15 degrees, the chip would lose its heat very easily, and in a 35 degrees environment it would be much warmer. However, there is no such effect; the difference is very similar over the whole temperature range.

[Albert Proust]

In fact is not a fixed value, unfortunately! We use our reflow oven to have an exact temperature (24°C) and we set via bluetooth the offset value between the oven's temperature and the internal temp sensor. It's a bit tricky but that's work.

[Bouke]

Wow, that sounds like a really nice setup! Unfortunately this is not viable in my case...

[Susheel Nuguru]

"what surprises me most is that it is fairly constant over the whole temperature range" what do you mean by this? you mean that the temperature of die remained fairly constant at the whole tested temperature range? looking at your chart it seemed that the die temperature was varying accordingly. Also the thermal conductivity of chip is another charecteristics which neccessarily is not linear with outside temperature.