Strange frequency shift on STM32 about mozzi HOT 4 CLOSED

Very interesting. Possibly our audio output timer is slightly off.

Some questions:

1. Am I right in guessing that the sample also plays 1.5% faster?
2. Have you tried different AUDIO_RATEs?
3. What STM32 is this, exactly? Any chance F_CPU is not set quite correctly? Are you seeing the same thing on other boards of the same or different type (if available)?
4. Perhaps you might want to take a look at

   Mozzi/MozziGuts.cpp

   Line 523 in 53889db

   audio_update_timer.setPeriod(1000000UL / AUDIO_RATE);

   . Possibly we run into a case of bad rounding, there.

from mozzi.

1. You are right there.
2. Now I have tried 16384 and the frequency is near exact... See point 4!
3. It is STM32F103C at 72 MHz, a classic "Blue Pill" board. I don't have another type to try right now, but I don't think the problem is there...
4. Well spotted! the problem is indeed a bad rounding:
   1000000/16384 = 61.03515625 => rounded to 61, error of less than 0.05%
   1000000/32768 = 30.517578125 => rounded to 30, error of 1.7%!
   I think it will be better to set that timer prescaler and overflow manually instead of using setPeriod function. I will look into that and propose a PR in the next few days.

from mozzi.

Thanks! Looking forward to that!

from mozzi.

Hi,

I think the Teensy are also subject to this. Will try to append this correction to #137 .

from mozzi.