STM32CubeExpansion_NFC5 Software migration problem

[BHuan.839]

After migrating to Nordic-nrf52832 platform with STM32CubeExpansion_NFC5_V2.0.0, I can initialize successfully (all initialization processes can succeed), and fail to Polling Tag Detect.

Refer to the example completely for program running process

int main(void)
{
    bsp_board_init(BSP_INIT_LEDS);
    bsp_board_init(BSP_INIT_BUTTONS);
    APP_ERROR_CHECK(NRF_LOG_INIT(NULL));
    NRF_LOG_DEFAULT_BACKENDS_INIT();
    timers_init();
  
	  spiInit();
    NRF_LOG_INFO("SPI example started.");
 
	  if( true != demoIni() )
		{
			NRF_LOG_ERROR("************ rfal NG **************");
		}
	
    while (1)
    {
				demoCycle();
        
        NRF_LOG_FLUSH();
//        bsp_board_led_invert(BSP_BOARD_LED_0);
        nrf_delay_ms(500);
    }
}

​Debug Log

Democycle() didn't crash, just couldn't get the card searching result.

[Ulysses HERNIOSUS]

Hi,

which voltage are you referring in the log output? How was it retrieved?

Best if you could provide a logic analyzer trace of rfalInitialize().

Regards, Ulysses

[BHuan.839]

The voltage is init RF power 5V

rfalInitialize -> st25r3911Initialize -> st25r3911MeasureVoltage

uint8_t spiTxRx(const uint8_t *txData, uint8_t *rxData, uint16_t length)
{
	uint8_t i=0;
	if((NULL == txData) && (NULL == rxData))
	{
	    return i;
	}
	
	if(length > 128)
	{
			return i;
	}
	struct rt_spi_message msg1, msg2;
	if(txData != NULL)
	{
		if(rxData != NULL) // RX �?�空
		{
			msg1.send_buf   = txData; // �?��?缓存
			msg1.recv_buf   = rx_buff;// 接收缓存
			msg1.length     = length;
			msg1.cs_take    = 1;
			msg1.cs_release = 0;
			msg1.next       = &msg2;
 
			msg2.send_buf   = RT_NULL;
			msg2.recv_buf   = RT_NULL; 
			msg2.length     = 0;
			msg2.cs_take    = 0;
			msg2.cs_release = 1;
			msg2.next       = RT_NULL;
			rt_spi_transfer_message(spi_dev_st25r, &msg1);
			memcpy(rxData, rx_buff, length);
		}
		else // RX NULL
		{
			msg1.send_buf   = txData; // �?��?缓存
			msg1.recv_buf   = RT_NULL;
			msg1.length     = length; // �?��?长度
			msg1.cs_take    = 1;
			msg1.cs_release = 0;
			msg1.next       = &msg2;
 
			msg2.send_buf   = RT_NULL;
			msg2.recv_buf   = RT_NULL;
			msg2.length     = 0;
			msg2.cs_take    = 0;
			msg2.cs_release = 1;
			msg2.next       = RT_NULL;
			rt_spi_transfer_message(spi_dev_st25r, &msg1);
		}
	}
	else // TX NULL
	{
		if(rxData != NULL) // RX �?�空
		{
			msg1.send_buf   = RT_NULL;
			msg1.recv_buf   = RT_NULL;
			msg1.length     = 0;
			msg1.cs_take    = 1;
			msg1.cs_release = 0;
			msg1.next       = &msg2;
 
			msg2.send_buf   = RT_NULL;
			msg2.recv_buf   = rx_buff; // 接收缓存
			msg2.length     = length;
			msg2.cs_take    = 0;
			msg2.cs_release = 1;
			msg2.next       = RT_NULL;
			rt_spi_transfer_message(spi_dev_st25r, &msg1);
			memcpy(rxData, rx_buff, length);
		}
	}
	rt_thread_mdelay(5);
  return i;
}

[Ulysses HERNIOSUS]

Hi,

• Your picture shows an ST Nucleo Board with your antenna. Can it read the tag?
• On the picture the antenna seems to have some shielding on the bottom top. Better to turn it around
• Did you adapt the matching for your antenna (inductance)? This is a necessary step for getting the antenna to work. Please refer to AN4974 Antenna matching for ST25R3911B/ST25R391x devices.
• The sampling frequency of your Saleae is at 2MHz, this is not enough to sample a 2MHz SPI!
• Please increase the sample frequency and provide the actual trace file of Saleae, not only screenshots. If you regard this as confidential you can also share through private message.
• I still see above an rt_thread_mdelay(5) - all NFC action will considerably slow down!
• The st25r3911MeasureVoltage() just gives the supply voltage. Please execute st25r3911MeasureAmplitude() and st25r3911MeasurePhase() to get the voltage and phase on RFI
• Please provide result value of st25r3911CalibrateAntenna()

Regards, Ulysses

[BHuan.839]

Under the same hardware platform, I can use STM32CubeExpansion_NFC5_V2.0.0 l476 example without any problem. The antenna matching is OK, and the SPI rate I have tried is useless. Delay time also tried all kinds of useless. Even if you know that the data of logic analyzer is wrong, what else can I do to improve it?

[Ulysses HERNIOSUS · Accepted solution]

Hi,

typically Saleae can be configured to sample much faster. If your device is really limited to 2MHz, then reduce SPI to 1MHz and send me the traces.

Regards, Ulysses ​