Let assume that the motor is 12 poles pairs (not 12 poles) then the mechanical to electrical ratio is 12 so we have
120 mechanical Hz becomes 1440 electrical Hz
180 mechanical Hz becomes 2160 electrical Hz
In the latter case we have a step time frequency of 2160 x 6 = 12960 Hz between two commutation event.
Then we can hypnotize that we have the sequence (Commutation, Demagnetization, Zero crossing, Commutation) each PWM cycle, we need to set 38.8kHz of PWM frequency.
First question is: can STM8 do the job?
Using the data reported in the validation document I see that it is required 14us between interrupt latency and execution time over a period of 25.72us. So it seems that the job can be performed using 54.4% of workload.
Second question is: these values are practically possible considering the stabilization time that depend on the power stage? And only one PWM period can be sufficient to achieve a good zero crossing detection?
I think that the electrical frequency of 680Hz (as explained in the thread you mention) can be more realistic.
Another suggestion, if we want to preserve the added value of the sensorless control, can be to analyze the possibility of a sinusoidal control using for instance a STM32 value line.
Sorry for the late answer.
Retrieving data ...