Zephyr SPI 1 issue after migration from F746zg to F756zg

[STM32Fuser]

Hi all
I migrated from a Nucleo F746zg to a F756zg. Code and overlay is the same except that I west build against F756zg. Now SPI 1 is not working properly anymore. On the scope I can see that SPI communication starts (CS goes low, MOSI byte is sent, MISO byte is sent back) but the transfer does not finish. I ran the same but using SPI 3 this worked flawlessly. Don't know maybe an APB or FIFO issue?
Anyone an idea?

Sorry, I forgot to say that I tested this with a Zephyr RTOS firmware (which worked with the F746zg) and as well with a simple straight forward Board-to-Board HAL_SPI_TransmitReceive Ping Pong F756zg (SPI 1 master) <-> F746zg (slave). Besides I want to mention that the same Ping Pong with F756zg (SPI 3 master) <-> F746zg (slave) worked flawlessly.

[TDK]

SPI1 works on both chips, issue is elsewhere.

> but the transfer does not finish

What does this mean?

[STM32Fuser]

HAL_SPI_TransmitReceive on the master side is entered but never left. I see the outgoing byte on the MOSI line and the same byte with a short delay on the MISO line sent back from the slave. When I check the RXNE register on the master it is set to zero.

This are my settings for F756zg SPI1 master on the:

 

hspi1.Instance = SPI1;

hspi1.Init.Mode = SPI_MODE_MASTER;

hspi1.Init.Direction = SPI_DIRECTION_2LINES;

hspi1.Init.DataSize = SPI_DATASIZE_8BIT;

hspi1.Init.CLKPolarity = SPI_POLARITY_LOW;

hspi1.Init.CLKPhase = SPI_PHASE_1EDGE;

hspi1.Init.NSS = SPI_NSS_SOFT;

hspi1.Init.BaudRatePrescaler = SPI_BAUDRATEPRESCALER_64;

hspi1.Init.FirstBit = SPI_FIRSTBIT_MSB;

hspi1.Init.TIMode = SPI_TIMODE_DISABLE;

hspi1.Init.CRCCalculation = SPI_CRCCALCULATION_DISABLE;

hspi1.Init.CRCPolynomial = 7;

hspi1.Init.CRCLength = SPI_CRC_LENGTH_DATASIZE;

hspi1.Init.NSSPMode = SPI_NSS_PULSE_DISABLE;

 

/**SPI1 GPIO Configuration

PA5 ------> SPI1_SCK

PA6 ------> SPI1_MISO

PB5 ------> SPI1_MOSI

*/

GPIO_InitStruct.Pin = GPIO_PIN_5|GPIO_PIN_6;

GPIO_InitStruct.Mode = GPIO_MODE_AF_PP;

GPIO_InitStruct.Pull = GPIO_NOPULL;

GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_VERY_HIGH;

GPIO_InitStruct.Alternate = GPIO_AF5_SPI1;

HAL_GPIO_Init(GPIOA, &GPIO_InitStruct);



GPIO_InitStruct.Pin = GPIO_PIN_5;

GPIO_InitStruct.Mode = GPIO_MODE_AF_PP;

GPIO_InitStruct.Pull = GPIO_NOPULL;

GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_VERY_HIGH;

GPIO_InitStruct.Alternate = GPIO_AF5_SPI1;

HAL_GPIO_Init(GPIOB, &GPIO_InitStruct);



/* USER CODE BEGIN SPI1_MspInit 1 */

/**SPI1 GPIO Configuration

PA4 ------> SPI1_CS

*/

/*Configure GPIO pin : PA4 */

GPIO_InitStruct.Pin = GPIO_PIN_4;

GPIO_InitStruct.Mode = GPIO_MODE_OUTPUT_PP;

GPIO_InitStruct.Pull = GPIO_PULLUP;

GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_LOW;

// GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_VERY_HIGH;

HAL_GPIO_Init(GPIOA, &GPIO_InitStruct);

//

// /* Set CS high initially (inactive) */

HAL_GPIO_WritePin(GPIOA, GPIO_PIN_4, GPIO_PIN_SET);

On the slave side I use hardware NSS and the same pins but with a slightly different configuration (due to hardware NSS). 

The same setup F746zg (Master) SPI1 <-> F746zg (Slave) SPI1 worked.

[TDK]

Note that reading DR within the debugger will cause this behavior.

> HAL_SPI_TransmitReceive on the master side is entered but never left.

Run the program, hit pause when it's "stuck". Where is execution at? What is it waiting for?

[STM32Fuser]

It stucks here in stm32f7xx_hal_spi.c:

/* Wait until RXNE flag is reset */

if ((__HAL_SPI_GET_FLAG(hspi, SPI_FLAG_RXNE)) && (hspi->RxXferCount > 0U))

{

Update:
Sorry, actually I mean in this loop:

while ((hspi->TxXferCount > 0U) || (hspi->RxXferCount > 0U))

{

 
As far as I can see is the RXNE flag a read only flag. But is there some other flag I must set/reset myself after all is received? And if yes, why did it work with the F746zg without any such measures?

[TDK]

Are you reading the DR register in the debugger? Doing that, or having SCK signal be subpar can cause this issue.

[STM32Fuser]

No I don't have any Live Expressions, any Watch nor anything intentionally in my code which could periodically access the DR. When I access it, then only manualy during a pause. Besides the scope shows the exact same behaviour when running the code without debugging.

But the SCK is a good hint. The shapes look clean but have quite an overshoot. Maybe reducing the slew rate would help. What do you think? 

[mƎALLEm]

Hello @STM32Fuser and welcome to the ST community,

1 - In next time please use </> button to share the code. See this post.

2- I can confidently say that there is no difference between Nucleo F746zg and Nucleo F756zg. Most probably it's a hardware issue and something was modified in the non working board example a resistor a solder bridge or something else. Check the schematic in the SPI pins path.

[STM32Fuser]

Hi
1. Yes sorry, I will keep in mind. I'm new here but it's not an excuse.
2. Yes, R121 must be moved to R122. I did that and checked both lines with a multimeter if everything is correct. I also checked directly at the MCU. I think F746zg share the same schematics with F756zg or did I miss something?

[mƎALLEm]

---

@STM32Fuser wrote:

I think F746zg share the same schematics with F756zg or did I miss something?

---

Indeed both boards have the same schematic and even the same PCB. That's why I have a doubt about the hardware and something was modified in it ...