Is TXP Necessary for SPI Slave?

[EthanMankins]

I am having issues with my slave responding to my master. I am trying to lessen complexity by only using RXP on the slave, without TXP. The slave (Nucleo-H743ZI) responds one byte late.

On the CS interrupt I send 0 for the first byte and enable RXP.

*((__IO uint8_t *)&hspi1.Instance->TXDR) = 0;
__HAL_SPI_ENABLE_IT(&hspi1, SPI_IT_RXP);
__HAL_SPI_ENABLE(&hspi1);

I then receive a byte and set the next transmit byte. I use FPint so the master does not send data when it is low (used for later implementation).  

void spiISRSTM(SPI_HandleTypeDef *hspi) {

	//Receive if ready
	if (hspi->Instance->SR & SPI_FLAG_RXP) {
		HAL_GPIO_WritePin(nFPINT_GPIO_Port, nFPINT_Pin, GPIO_PIN_RESET);	
 	rxBuffer = (*(__IO uint8_t *)&hspi->Instance->RXDR);
 	(*(__IO uint8_t *)&hspi->Instance->TXDR) = 1;
 	HAL_GPIO_WritePin(nFPINT_GPIO_Port, nFPINT_Pin, GPIO_PIN_SET);

	}
	__HAL_SPI_CLEAR_UDRFLAG(hspi);
	__HAL_SPI_CLEAR_OVRFLAG(hspi);
	__HAL_SPI_CLEAR_EOTFLAG(hspi);
	__HAL_SPI_CLEAR_FREFLAG(hspi);

}

The MISO line for this setup is:

0011 1111

I would expect:

0111 1111

Any Ideas what is causing this?

[Saket_Om]

Hello @EthanMankins 

Could you try activating DXP interrupt and checking on DXP flag instead of RXP? 

 

[EthanMankins]

Hi @Saket_Om 

I have tried this and it does not make difference to the timing.

[EthanMankins]

The issue was an underrun after the first byte. Disabling and reenabling spi after read and before transmit solved the issue.

sr = hspi->Instance->SR;
if ((sr & (SPI_FLAG_RXP | SPI_FLAG_TXC)) == (SPI_FLAG_RXP | SPI_FLAG_TXC)) {
		HAL_GPIO_WritePin(nFPINT_GPIO_Port, nFPINT_Pin, GPIO_PIN_RESET);	//Lower nFPINT
		rxBuffer = (*(__IO uint8_t *)&hspi->Instance->RXDR);
		__HAL_SPI_DISABLE(&hspi1);
		__HAL_SPI_ENABLE(&hspi1);
		pPacketSlave->ProcessSPI();
		(*(__IO uint8_t *)&hspi->Instance->TXDR) = txBuffer;
	}