SO8N STM32G031C6 USART not working.

[fancyrat]

I'm trying to setup an USART for a very small MCU of the STM32G0 family, the MCU it's theoretically being programmed, but the USART is not responding, I configured the MCU just like I did in other projects (of bigger MCUs), I'm not able to find what I'm doing wrong in this case.

This is what I can see from the USART (theoretically in this image I'm sending "0xFF" all the time, but it does the same whatever I put in the buffer).

Codewise I'm enabling the USART DMA with LL_USART_EnableDMAReq_TX(USART1), then configuring with LL_DMA_ConfigAddresses.

Some checks with LL_DMA_IsEnabledChannel, and finally sending stuff with both LL_DMA_SetDataLength & LL_DMA_EnableChannel once the buffer has something on it.

Also tried to clear TC1 flag & disable channel every communication, checking it with LL_DMA_IsActiveFlag_TC1.

The USART is configured as Half-duplex (as it need to be) & baudrate, word length, etc...are correct:

[KnarfB]

And if you don't use DMA but simply write to TXD?

Is is a pin issue or a DMA prog issue?

hth

KnarfB

[fancyrat]

Hi,

I've always worked with DMA so I'm not confident in the code I've just wrote, but it seems that the same is happening.

I write this very quickly anyway, so I'll keep investigating in this direction. This is the code I've just written that doesn't use DMA.

void comms_init(void)
{
 // USART clock
 RCC->APBENR2 |= RCC_APBENR2_USART1EN;
 
 // USART baud rate
 USART1->BRR = (uint16_t)(SystemCoreClock / 115200);
 
 // USART data (8 data bits, no parity, 1 stop bit)
 USART1->CR1 = USART_CR1_TE | USART_CR1_RE | USART_CR1_UE;
 
 // USART peripheral
 USART1->CR1 |= USART_CR1_UE;
}
 
void USART_SendData(uint8_t data)
{
 // Wait for empty transmit data register 
 while (!(USART1->ISR & USART_ISR_TXFE))
 ;
 
 // Write data to the TDR
 USART1->TDR = data;
 
 // Wait for the transmission
 while (!(USART1->ISR & USART_ISR_TC))
 ;
}
 
void comms_task(void)
{
 
 while (1)
 {
 USART_SendData('H');
 USART_SendData('e');
 USART_SendData('l');
 USART_SendData('l');
 USART_SendData('o');
 USART_SendData('\r');
 USART_SendData('\n');
 }
}

[fancyrat]

EDIT: If I remove the watchdog the "spikes" that go to 0 disapear.

[KnarfB]

Your later code works* if you replace the flag with:

while (!(USART1->ISR & USART_ISR_TXE_TXFNF))

*) have added more init code for the pins, namely the auto-generated MX_USART1_UART_Init() for my STM32G031 Disco..Note that USART1_TX pin is same as BOOT0 and SWCLK, which might be confusing...

hth

KnarfB

[Viktor POHORELY]

Did you tried just simple GPIO toggling, to see if the issue comes from pin or IP ?

There are several pins in paralel on those small packages, thus in case you have other one configured as output it may not work correctly.

[fancyrat]

Thank you all,

I built the project from scratch and I manage to sent data both with the DMA & HAL, now the problem I'm having is that the data I'm receiving it's not exactly correct, I'm receiving incorrect characters, I'll try to check if with an oscilloscope, but at a simple glance, everything looks good.

Also I would try again with my old code, as @KnarfB​ confirms that it worked with a very small change.

Edit: Well I just tried with USART_SendData code and I'm still receiving weird stuff.

[fancyrat]

Just as a reference, this is what I've got now, it should be saying "HOLA" using the UART with an DMA, I'm receiving other stuff mostly random symbols, I can't seem to find the relation to it:

[fancyrat]

Now I've tried to write "*\n* just to have some other signal to compare like the one in this link: https://hackaday.com/2016/06/22/what-could-go-wrong-asynchronous-serial-edition/

It seems that in my case the message keeps repeating multiple times. But the bits are correct:

[Bob S]

Your polarity looks inverted. With TTL/3V level signals a UART should idle high. The start bit is initiated by a high to low transition. These images look like your line is idling low. Unless I am missing something, like you have a TTL-to-RS232 level shifter that isn't able to drive below 0V???? I don't know how that (inverted could happen in the G0, but then I don't know much about the G0 family.

[fancyrat]

Now that you said it...yeah, that's weird, There's nothing you're missing, I don't have a TTL-to-RS232 shifter.

I'm taking a look at it, everything is really weird. I managed to receive signals if I sent messages very quickly and it's very inconsistent, if I had a delay to sent messages every second then it's impossible to make it work and the signal you see it's like the one were talking about.

I don't understand anything.

Maybe it's just noise, what I see now looks like a word, but the characters are wrong. Maybe he's interpreting that the start or stop bit or whatever is in the wrong position.

[Viktor POHORELY]

Do you have baud rate set correctly?

[fancyrat]

Yes, baud rate is set correctly. I haven't managed to make it work with the DMA (as I want). But the following code is working right now:

void send_uart_data(const uint8_t *data, uint32_t size) {
 for (uint32_t i = 0; i < size; i++) {
 while (!LL_USART_IsActiveFlag_TXE(USART1)); // Wait until TX buffer is empty
 LL_USART_TransmitData8(USART1, data[i]);
 }
}
 
void comms_task(void)
{
 const uint8_t data[] = "Hello, world!";
 send_uart_data(data, sizeof(data)-1);
}

I don't know, I'm probably missing a configuration or initialization in the DMA. This is what I've done in the last try, I've tried more complex stuff, but wasn't working, so I'm trying simpler things now:

/* Define UART Tx DMA buffers */
uint8_t uart_dma_buffer[32] = "Hello, world!";
uint8_t dma_buffer_size = sizeof(uart_dma_buffer);
 
void comms_init(void){
 /* Enable DMA in the USART */
 LL_USART_EnableDMAReq_TX(USART1);
 
 /* Configure DMA for USART Tx */
 LL_DMA_InitTypeDef dma_init = {0};
 dma_init.Direction = LL_DMA_DIRECTION_MEMORY_TO_PERIPH;
 dma_init.PeriphOrM2MSrcAddress = (uint32_t)&USART1->RDR;
 dma_init.MemoryOrM2MDstAddress = (uint32_t)uart_dma_buffer;
 dma_init.Mode = LL_DMA_MODE_CIRCULAR;
 dma_init.PeriphOrM2MSrcIncMode = LL_DMA_PERIPH_NOINCREMENT;
 dma_init.MemoryOrM2MDstIncMode = LL_DMA_MEMORY_INCREMENT;
 dma_init.PeriphOrM2MSrcDataSize = LL_DMA_PDATAALIGN_BYTE;
 dma_init.MemoryOrM2MDstDataSize = LL_DMA_MDATAALIGN_BYTE;
 dma_init.NbData = dma_buffer_size;
 
 LL_DMA_Init(DMA1, LL_DMA_CHANNEL_1, &dma_init);
 
 /* Enable DMA channel */
 LL_DMA_EnableChannel(DMA1, LL_DMA_CHANNEL_1);
}
 
void comms_task(void)
{
 /* Wait for DMA transfer complete */
 while (!LL_DMA_IsActiveFlag_TC1(DMA1)) {}
 
 /* Clear DMA transfer complete flag */
 LL_DMA_ClearFlag_TC1(DMA1);
 
 /* Restart DMA transfer */
 LL_DMA_SetDataLength(DMA1, LL_DMA_CHANNEL_1, dma_buffer_size);
 LL_DMA_EnableChannel(DMA1, LL_DMA_CHANNEL_1);
}

[KnarfB]

The DMA in your above code is setup for receiving, not transmitting:

dma_init.PeriphOrM2MSrcAddress = (uint32_t)&USART1->RDR;

hth

KnarfB

[fancyrat]

Ok, this is really stupid.

I managed to get a consistent (wrong) message using DMA. If I use my hands as a "resistance" between both Tx wires, I receive the correct messages.

I guess that I'm stuck now working as a wire for the rest of my life.

[KnarfB]

STM32G031C6 has no SO8N in CubeMX.

[fancyrat]

In CubeMX I'm working with STM32G031J6Mx SO8N. I think that both chips are mostly the same.

[KnarfB]

Yeah, the disco board? For your above plots: double-check that you have probed the correct pins. On my disco, I have swapped RX and TX pins internally (CubeMX), so PB6 is free for SWD debugging. Make sure that TX is configured as push-pull PP, not OD by CubeMX.

Its also advisable to leave/config PF2 as NRST (option bytes), so you don't lock out yourself like several people here already did.

[Edit]: Have tried it with HAL_UART_Transmit_DMA and .. it worked as expected.

hth

KnarfB