On Wednesday, 9 September 2020, 18:58:39 CEST, Andy Shevchenko wrote: > On Wed, Sep 9, 2020 at 6:45 PM Christian Eggers <ceggers@xxxxxxx> wrote: > > We found some sensors which are much slower (20% at room temperature) > > than nominal. According to the data sheet, up to 27% is possible. Now I > > add 33% to the nominal time out, hopefully this is enough. > > I'm wondering if this is a function of temperature. > And if that has an actual thermistor inside... I haven't designed the hardware, but I guess that clock generation is done using an RC oscillator which depends on temperature. Temperature measurement is possible, but I think the recommended way is to externally control the integration time using the SYN pin and to get the actual number of RC cycles from a register. But generating the required waveform for the SYN pin using a uC general purpose timer seems challenging. I wouldn't be surprised if the measurement results also depend on the exact value of the supply voltage.
