What are the consequences of using low-frequency GPIO lines with a high-frequency signal?

[havingnordicfun]

I have an external clock running at 3Mhz which I'd like to align with in software. Consider this contrived example:

// P0.03 is recommended 10kHz or less, but using for 3Mhz
#define CLOCK_PIN NRF_GPIO_PIN_MAP(0,3)

// P0.19 has no specified maximum frequency, and presumably is appropriate
#define CLOCK_PIN_FAST NRF_GPIO_PIN_MAP(0,19)

void init(){
    nrf_gpio_cfg_input(CLOCK_PIN);
}

void align(){
    while(!nrf_gpio_pin_read(CLOCK_PIN){};
    while(nrf_gpio_pin_read(CLOCK_PIN){};
}

void main(){
    init();
    int counter = 0;
    while(counter < 100000){
        align();
        counter++;
    }
}

What would the consequences be of attempting to read values on a GPIO pin not recommended for such a high frequency? Please disregard that a PPI channel and hardware timer would be more appropriate for this contrived example, I would really just like to know any and all consequences for not following the recommended specification for low-frequency GPIO pins.

Thanks!

[Andreas]

Hi,

 

Expected consequences are degraded RX sensitivity and possibly modulation index.

 

Best regards,

Andreas

[havingnordicfun]

Thank you for the reply. So would it be fair to say that the counter in the example above may not align properly in every case and miss cycles in the external clock due to decreased sensitivity?

[Andreas]

Hi,

 

Not exactly, the timer accuracy should not be affected, at least if using the HF crystal as clock source. RX sensitivity and modulation index are radio 'performance parametres', you might see degraded radio performance while the pin is being toggled.

 

Best regards,

Andreas

[havingnordicfun]

Hi,

Are you referring to a hardware timer going unaffected? To be sure, I'm asking about the ability to properly sense every change of the 3Mhz GPIO line in software alone, without GPIOTE or PPI or a timer tied to the HF crystal.

Thank you!

[Andreas]

Hi,

 

No I probably just mixed up counter and timer when issuing my response to you, sorry about that. Using the HF crystal as source for the HFCLK will give a stable CPU clock, every sample should happen as accurately as possible. I do not believe have investigated whether there can be differences between the pins, the note in the PS only refers to degratation or radio performance parametres.

 

Best regards,

Andreas