nrf5340 SPI 32MHz, HFCLK128M 128MHz and QSPI usage

RE: nrf5340 SPI 32MHz, HFCLK128M 128MHz and QSPI usage

Håkon Alseth — Wed, 27 Nov 2024 15:19:45 GMT

Hi,

Sorry, since you mention SPI at 16MHz, you are ofcourse referring to the high-speed SPI4. My apologies for my mix up.

SPIM4 can run on 32 MHz, but only if the HFCLK128M is used.

[quote user="tcube"]What happens if the nrf5340 SPI starts a transaction à 32Mhz and changes the HFCLK128M to 64MHz to start a QSPI transaction?[/quote]

The clock will continue to run at configured frequency, but it is not a supported combination and there is not guaranteed to work. If running SPIM4 at 32 MHz, the CPU must be configured at 128 MHz.

If you are asynchronously (with respect to the SPIM4 handling) switching CPU frequency, then I would recommend that you lower SPIM4 to 16 MHz.

Kind regards,

Håkon

RE: nrf5340 SPI 32MHz, HFCLK128M 128MHz and QSPI usage

tcube — Wed, 27 Nov 2024 14:31:03 GMT

A last question for my understanding:
Configuring the SPI to 32MHz requires the HFCLK128M to be at 128MHz.
What happens if the nrf5340 SPI starts a transaction à 32Mhz and changes the HFCLK128M to 64MHz to start a QSPI transaction? Does the SPI clock remain at 32 Mhz?

RE: nrf5340 SPI 32MHz, HFCLK128M 128MHz and QSPI usage

Håkon Alseth — Wed, 27 Nov 2024 13:51:02 GMT

Hi,

[quote user=""]Can we change the HFCLK128M speed to 64MHz before any access to the QSPI and restore it to 128MHz when it's finished?[/quote]

Yes, it is designed to be performed at any time (except for scenarios given in the errata you pointed to), as stated here:

https://docs.nordicsemi.com/bundle/ps_nrf5340/page/chapters/clock/doc/clock.html#ariaid-title3

[quote user=""]Is this can have a negative impact (corrupted data) on SPI, I2C, or BLE if we change the core clock during a transaction?[/quote]

This changes the scaling to the CPU, not the peripheral clocks.

The PCLK nets listed here will remain to whichever configuration they are set to use:

https://docs.nordicsemi.com/bundle/ps_nrf5340/page/chapters/clock/doc/clock.html#ariaid-title2

Kind regards,

Håkon