RADIO_IRQHandler not called after power cycle

[Johannes]

I am having a strange issue with my custom radio protocol. I have ported the code  to the 52840 and SDK15  from the nrf51822 where it is running with no issues.

The issue is that everything runs correctly when using the J-Link debugger. as soon as I disconnect the J-Link and perform a battery pull, the system will enter a boot loop. So far I have tracked it down (with LED blink signals, since it works again once I connect the debugger) to a point where the RADIO_IRQHandler is not being called after the battery pull.

The sequencs is like this (with SEGGER SES):

Debug->Run    ---> works!

pull battery and re-insert ----> bootloop

Target->Reconnect J-Link  ----> works fine after next boot!

Here is my Handler:

void RADIO_IRQHandler(void) {
    LED_ON(LED_1);  // LED never goes on that is how I know the IRQHandler is not called
    uint8_t event = 0;
     if (NRF_RADIO->EVENTS_READY) {           // put each event separately in the scheduler, even if they occured simulatenously due to shorts
        NRF_RADIO->EVENTS_READY = 0;
        event                   = RADIO_EV_READY;
        radio_handler(&event, sizeof(event));
    }
#ifdef NRF52840_XXAA
    if (NRF_RADIO->EVENTS_PHYEND) {
        NRF_RADIO->EVENTS_PHYEND = 0;
        event                 = RADIO_EV_PHYEND;
        radio_handler(&event, sizeof(event));
    }
#endif
    if (NRF_RADIO->EVENTS_END) {
        NRF_RADIO->EVENTS_END = 0;
        event                 = RADIO_EV_END;
        radio_handler(&event, sizeof(event));
    }
    if (NRF_RADIO->EVENTS_DISABLED) {

        NRF_RADIO->EVENTS_DISABLED = 0;
        event                      = RADIO_EV_DISABLED;
        radio_handler(&event, sizeof(event));
    }
}

 and this is how I enable the interrupts:

radio_enable_irq() {
    NRF_RADIO->INTENSET = RADIO_INTENSET_READY_Enabled << RADIO_INTENSET_READY_Pos |
                          RADIO_INTENSET_END_Enabled << RADIO_INTENSET_END_Pos |
                          RADIO_INTENSET_DISABLED_Enabled << RADIO_INTENSET_DISABLED_Pos |
#ifdef NRF52840_XXAA
                          RADIO_INTENSET_PHYEND_Enabled << RADIO_INTENSET_PHYEND_Pos |
#endif
                          0;


#ifdef NRF52840_XXAA
    NRF_RADIO->EVENTS_PHYEND      = 0;
#endif

    NRF_RADIO->EVENTS_END      = 0;
    NRF_RADIO->EVENTS_DISABLED = 0;
    NRF_RADIO->EVENTS_READY    = 0;
    NVIC_SetPriority(RADIO_IRQn, 1); 
    NVIC_ClearPendingIRQ(RADIO_IRQn);
    NVIC_EnableIRQ(RADIO_IRQn);
}

Is there any known mechanism how the interrupts would become disabled after a battery pull? And what does the J-Link change that it works again?

[wpaul]

The major difference in operation between having the debugger connected and not is that when the debugger is attached, it prevents the CPU from entering low power mode, i.e. the WFI instruction won't actually spin down the CPU or turn off any clocks.

Unfortunately I can't really say why that would cause the symptoms you describe without seeing the code, but hopefully this will provide a clue. Do you have code to put the chip in low power? It might be it still needs some adjustments to work correctly on the 52840.

-Bill