External rail-to-rail 32K clock - internal xtal driver damaged?

[Anders Nelson]

Hi guys,

I have a custom board containing a Raytac nRf52840 module running the Arduino core. The Arduino core configures the nRf52 to expect a 32K crystal on the xtal input pins by default. However, I applied a rail-to-rail 3.3v 32K oscillator to the nRf52 with this configuration active. BLE does not function and USB CDC behavior is somewhat erratic.

I replaced the rail-to-rail oscillator with a crystal and BLE continues to not function. I see the voltage rating for the xtal pins in xtal mode (< 1000mV swing) - can I expect the inputs to be damaged now? If I configure the core to use the internal LF oscillator, can I expect BLE to work?

Thanks!

[run_ar]

Hi,

It is possible that the inputs can be damaged. But you could still try to use the LFRC and see if BLE works. remember to also change the ppm setting to 500ppm for the clock.

[Anders Nelson]

Hey, thanks for your response! I configured the LFRC to run and BLE works now. I'm now having trouble with the bootloader not responding to the PC - is it possible there is a baudrate error due to use of the LFRC?

Where can I find the area to change for a 500ppm tolerance?

Thanks!

[run_ar]

You  can change to 500ppm in sdk_config.h. change #define NRF_SDH_CLOCK_LF_ACCURACY 1

Also note that there are multiple entries for *LF_SRC that need to be updated: NRF_SDH_CLOCK_LF_SRC, NRFX_CLOCK_CONFIG_LF_SRC and CLOCK_CONFIG_LF_SRC

Not sure why there should be a baudrate error due to the LF SRC. This would be derived from the HF clock. So you might want to check that it is using the crystal and not the HFRC.