nRF52840 USB HID with simultaneous SPIM transfer

RE: nRF52840 USB HID with simultaneous SPIM transfer

ovrebekk — Thu, 21 Mar 2019 11:20:06 GMT

Good to hear you figured it out, I will close the case then 🙂

RE: nRF52840 USB HID with simultaneous SPIM transfer

ilia — Tue, 19 Mar 2019 11:10:01 GMT

Hello,

The problem is solved now. It was not related to USB, SPI, or the board itself. I just unintentionally made one mistake in placing variables in memory. Basically I created a variable inside a function and stored its address in a global variable, and used this address in further calls.

RE: nRF52840 USB HID with simultaneous SPIM transfer

ilia — Mon, 18 Mar 2019 12:02:06 GMT

How can I make USB and SPIM working together? These features work perfectly if I use them separately, but if I turn on both USB and SPIM my application crashes. Maybe there is some sort of interruption conflict?

RE: nRF52840 USB HID with simultaneous SPIM transfer

ilia — Fri, 15 Mar 2019 17:28:08 GMT

It works, but only without call to __WFE();, and without USB functionality.

If I have app_usbd_event_queue_process() call in my main cycle, application breaks in the same way it was before any modifications.

RE: nRF52840 USB HID with simultaneous SPIM transfer

hmolesworth — Fri, 15 Mar 2019 17:06:51 GMT

For this I would just count wakeups rather than complicating it with another interrupt for the timer. Something like this:

// Trigger every 1,000 wakeups if 1mSec system tick, 100 if 10mSec system tick
static uint32 LoopCount = 0;

		if ((SPI_Status) && ((++LoopCount % 1000) == 0))
		{
			SPI_Status = 0;
			NRF_LOG_INFO("Transfer completed.");
			// Edit: remove this linenrf_delay_ms(10); 
			RGB_Refresh();  
		}

RE: nRF52840 USB HID with simultaneous SPIM transfer

ilia — Fri, 15 Mar 2019 16:58:46 GMT

Hello,

I agree, calling RGB_Refresh in interrupt is not a good idea. I placed it there in attempt to find working solution.

I have modified my main:

int main(void)
{  
    nrf_gpio_cfg_output( LOGO );
    
	Logo_Initialize(LOGO);

	HID_Initialize();

    while (true)
    {
		while (app_usbd_event_queue_process())
        {
            /* Nothing to do */
        }
      
		if (SPI_Status)
		{
			SPI_Status = 0;
			NRF_LOG_INFO("Transfer completed.");
			nrf_delay_ms(10); 
			RGB_Refresh();  
		}
        /* Sleep CPU only if there was no interrupt since last loop processing */
        __WFE();
    }
}

The behavior is the same as in my previous version, and the problem is still present.

I use nrf_delay_ms(10); just to slow down animation, perhaps it is not a good way to do it. I'll try to change it to timer api.

RE: nRF52840 USB HID with simultaneous SPIM transfer

hmolesworth — Fri, 15 Mar 2019 16:20:36 GMT

I would suggest that you have an issue with putting stuff in the interrupt which does not belong in an interrupt, particularly when that stuff generates the same interrupt before the current interrupt completes. Perhaps change this:

void spim_event_handler(nrfx_spim_evt_t const * p_event, void * p_context)
{
    NRF_LOG_INFO("Transfer completed.");
	
	nrf_delay_ms(10);
	RGB_Refresh();  
}

to something like this:

// Flag to trigger next display refresh outside of any interrupt
volatile uint16_t SPI_Signal = 0;

void spim_event_handler(nrfx_spim_evt_t const * p_event, void * p_context)
{
   SPI_Signal++;
}

   // in main loop after wakeup due to SPI interrupt above
   if (SPI_Signal)
   {
    SPI_Signal = 0;
    NRF_LOG_INFO("Transfer completed.");
	nrf_delay_ms(10); // <== what's this for? better to use a timer
	RGB_Refresh();  
   }

SPI_Signal should only be 0 or 1, but it doesn't hurt to allow more for debugging