STM32 HSE oscillator stability problem
02 May 2017
I have implemented support for Precise Time Protocol in STM32 and it still performed bad. After some investigation, I have descovered stability problems of HSE.
I have reported these problems to the ST, they have confirmed this behavior but unfortunately, it does not appear in any errata of SMT32 line. The problem was discovered on STM32F107 and later confirmed on STM32F207 and STM32F407.
I used several evaluation boards with ST’s MCU. I did just internal connection of HSE directly to MCO output of the microcontroller so no PLL was used. The program in the MCU performs just infinite loop after configuring the MCO. Using MCO also enables measurement of other clock sources and I was able to connect frequency counter directly to the microcontroller.
I was measuring frequency every second and storing it using GPIB and SCPI to the PC. After some minutes there ware enough data. The base frequency of the oscillator was 24999407,7 Hz, but it was not stable! I assumed, thet there would be visible just temperature drift but it was not true. Measured frequency was hopping within several Hz. Depending on X-tal, these spikes ware from 2 Hz to 8 Hz.
I did’t perform any tests of HSI, but for precise time, RC oscillator was no-go anyway for me.
Only solution for this was to use complete external oscillator at 25 MHz that was stable enough. With external oscillator, I was able to reach expected performance of the Precise Time Protocol.
Also others confirmed this behaviour https://blog.dan.drown.org/gps-module-measurements/