USART line low

[jgauthier]

Hello,

 I'm working with an STM32G05, on a very simple custom board. I expose both USARTs to external headers.

USART1 seems to be operating fine - in that I can send/receive data.

However, USART2 is not.

When I inspected the TX line I discovered it was low, even at boot.

I check for anything that might be pulling it to ground, and there isn't anything on the board itself.

Is there something with the processor that could be causing this?

[Andrew Neil]

What board are you using?

If it's a custom design, has it ever worked before?

If you change the TX line to be just a plain GPIO, can you set it high & low?

[jgauthier]

Thanks for that idea. The custom board has never been used - it's brand new.

I made PA9, and PA10 both GPIO Outputs and they are both toggling correctly.

I set them back to USART1, and USART1_TX (PA9) stays low.

[jgauthier]

I believe I have found a bug in cubeMX code generation.

Code generated for USART1 looks like this:

 GPIO_InitStruct.Pin = GPIO_PIN_9|GPIO_PIN_10;
 GPIO_InitStruct.Mode = GPIO_MODE_AF_PP;
 GPIO_InitStruct.Pull = GPIO_NOPULL;
 GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_LOW;
 HAL_GPIO_Init(GPIOA, &GPIO_InitStruct);

The code generated for USART2 looks like this:

 GPIO_InitStruct.Pin = GPIO_PIN_2|GPIO_PIN_3;
 GPIO_InitStruct.Mode = GPIO_MODE_AF_PP;
 GPIO_InitStruct.Pull = GPIO_NOPULL;
 GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_LOW;
 GPIO_InitStruct.Alternate = GPIO_AF1_USART2;
 HAL_GPIO_Init(GPIOA, &GPIO_InitStruct);

I added

GPIO_InitStruct.Alternate = GPIO_AF1_USART1;

to the initialization structure for USART1 and now it works as expected.

[Tesla DeLorean]

It does look wrong, but you originally stated that the issue was with USART2, not USART1

What did that failing code look like?

[jgauthier]

You're right. I made a mistake because for some reason I thought it was the secondary UART. After posting a bug I received this response:

https://community.st.com/s/question/0D53W000019Ctw4SAC/uart-not-configured-properly-for-stm32g061

It was confirmed by ST that it is a code generation issue,

[Andrew Neil]

If the issue is now resolved, please mark the solution: