STM32L476 Nucleo UART/HAL looses characters on TxD

JuM · ‎2017-10-12

Posted on October 12, 2017 at 10:04

Hi,

trying to transmit data via USART1 of STM32L476 Nucleo evaluation board. Use PA9 for TxD.

Character count is 254*2+14 = 522 or 255*2+14 = 524.

522 always is ok, 524 gives errors at receiver site.

Data switches between 'all 0x00' and 'all 0xFF' every cycle.

Use HAL_UART_Transmit (...) for sending data.

Receiver always flags Error on 0xFF packets.

Took measurement on TxD with oscilloscope. That shows that with 'all 0xFF' always 8 characters are not sent!!! (s. attachement).

What may be the cause of this?

P.S.

By the way: Took a look into HAL_UART_Transmit () source code.

Found that writing TDR does not check TXE flag first (but that is not the cause of the 8 byte error).

waclawek.jan · ‎2017-10-12

Posted on October 12, 2017 at 13:54

What baudrate? Isn't the timeout too short?

JW

JuM · ‎2017-10-12

Posted on October 12, 2017 at 14:03

Oh yes, you're absolutely right.

10 msec timeout is too short (although we are using 500 kbit/sec).

What about TXE bit checking before writing TDR in HAL_UART_Transmit () ?

Thank you for support!

waclawek.jan · ‎2017-10-12

Posted on October 12, 2017 at 14:25

Oh yes, you're absolutely right.

we are using 500 kbit/sec

Was I? Now I'm confused. I was under the impression that the timeout is per character. The ehm 'documentation' does not tell much.

What about TXE bit checking

I don't know. I don't and won't Cube.

JW

JuM · ‎2017-10-12

Posted on October 12, 2017 at 14:38

Looked again at 'HAL_UART_Transmit ()' source code...

TXE checking IS done (prior to TDR writing by call to 'WaitOnFlag...(UART_FLAG_TXE)...' ).

So at first sight timeout parameter seems to be a 'by char' timeout.

But: Before entering the loop ('TxFerCount--'...) a ticker value ('HAL_GetTick()' is fetched.

And further on the timeout is checked every char against this initial HAL_GetTick().

So timeout parameter becomes a frame timeout...

And: My frame with 524 chars has 10.06 msecs...

waclawek.jan · ‎2017-10-12

Posted on October 12, 2017 at 14:46

Thanks for the explanation.

Jan