Slow GPIO Output Speed Issue: Inquiring About Causes and Solutions

[HHJ]

Hello, we are currently testing the output of square waves through GPIO output pins. However, I found that a single operation to set the GPIO output pin high or low using the gpio_pin_set function takes about 3 microseconds, so the transmitted square wave does not meet our speed requirements. I would like to know what is causing the slow GPIO output speed—is it the internal clock being too slow? Or too many parallel tasks? (In debug mode, the chip does not jump to other tasks and executes GPIO operations continuously.) Are there better GPIO configuration functions? Please provide us with assistance.

[HHJ]

We are currently using the SPI of the chip hardware without any modifications, directly reading the DTS file via macros, and setting the CS_delay to 0. Our clock is set to 8M, which works well (32 clocks take 4us), but the CS line pulls high 33us after the clock line ends. This makes a single SPI operation take 46us, resulting in the SPI frequency failing to meet our time requirements—since the time consumed by the CS line determines the final SPI frequency. This is quite unreasonable; why does such a strange phenomenon occur?

[Håkon Alseth]

Hi,

 

You need a bit more flexibility and control over the hardware module.

I would recommend that you look into using the nrfx_spim driver directly instead of using the zephyr driver.

https://github.com/zephyrproject-rtos/hal_nordic/tree/master/nrfx/samples/src/nrfx_spim

 

Kind regards,

Håkon