MCU_CLK as microcontroller clock source?

[willem buitendyk]

From reading the ST25R3916 datasheet, it seems like using the MCU_CLK output as my microcontroller clock source will ultimately generate less undesirable noise and save on additional parts. My mcu and rf circuit will all be inside the antenna much like the str3911b-disco. However, I notice in the str3911b-disco design that the mcu is using an external 32 khz crystal instead and that mcu_clk is disconnected? Why didn't they use the MCU_CLK pin for this board?

[willem buitendyk]

I'm trying to understand the choice of clocks in the st25r2911b-disco design. HSE is not present but only the LSE. In the STM32G071G8 I'm using I only have one clock input so can't decide if I should use an LSE (32khz) or the MCU_CLK signal for more reliable UART communications as opposed to using the HSI. I've read I can tune the HSI using the LSE.

[willem buitendyk]

Ok I'm learning; or at least think I am. So if I understand USART correctly, a higher frequency clock combined with 16x oversampling allows for the ability to reduce baud rate error on receive. As my intention is to transmit reliably I would probably be better served with a crystal frequency that divides the desired baud rate evenly; such as a 7.3728 MHz crystal.

[Ulysses HERNIOSUS · Accepted solution]

Hi Willem,

the MCU_CLK is only accurate when running from divided main crystal oscillator (27.12). When running from 32kHz internal RC oscillator the accuracy will be limited.

USART in general can tolerate clock in-accuracies in the percent range.

ST25R3911B-DISCO is using USB where the deviation must be in the ppm range. We have rarely seen applications which use the MCU_CLK to drive the MCU. The main use case we have seen so far is to provide a clock to drive the SPI Clock of the MCU when using the transparent mode where one wants to modulate/sample synchronous to the field clock (e.g. fc/128, etc.).

Regards, Ulysses