CAN-Bus Problems

[dunkindonut]

Posted on July 09, 2014 at 13:53

Hi,

Has anyone

run

a

CAN

application

that works

also

in

error

cases

?

In my

application

the connection hangs

, for example,

after

pulling the

CAN bus

pin plug

and re-

connect.

One reason for this

is the

HAL_CAN_Transmit

()

function,

the

function

does not call

__

HAL_UNLOCK

(

HCAN)

in the

event of a fault

(HAL_ERROR return).

It seems

that

the same problem

also exists in

HAL_CAN_Sleep

() and

HAL_CAN_WakeUp

().

Another problem

is that

the MSR

>

ERRI

flag is

never

reset

,

after

bus errors

the system hangs

in the error

interrupt

Prodedure

(

endless)

.

How do I get

the system

working after CAN-Bus Errors?

Thanks

#bxcan #bug #stm32cube #can #can-sce-silent-mode

[jpeacock2399]

Posted on July 11, 2014 at 21:54

The way I handle CAN errors is to first detect it with excessive TX or RX errors in the SCE interrupt.  If it's obvious the node has to disconnect then I switch to SILENT mode in the BTR register.  I also disable the SCE interrupts, otherwise they will flood with constant errors.

I set a soft timer in the RTOS (FreeRTOS) to periodically retry the bus by bringing it out of SILENT mode and enabling error interrupts.  If the errors reappear then it's back to SILENT again, otherwise I start processing messages.

Leaving the bus in SILENT mode ensures it will detect when the bus fault is cleared (assuming it's remote).  The periodic enable is in case there are only two nodes and the other one needs a node to complete the network.

This is all custom drivers, I don't use the HAL since ST tends to skip any in depth fault handling.

  Jack Peacock

[Montassar BEN ROMDHANE_O]

Posted on August 06, 2014 at 13:36

Hi Helmut,

Thanks for the feedback. We understand that this is an issue, and that a ''__HAL_UNLOCK()'' is missed in case of error, and as a temporary workaround you can use ''HAL_CAN_DeInit()''. Concerning the 2nd point, yes the MSR>ERRI flag is never reset but other ESR flags are cleared and the generation of interrupt depends on ESR flags and not MSR ones. Please find below ERRIE bit description on RM:

Bit 15 ERRIE: Error interrupt enable
0: No interrupt will be generated when an error condition is pending in the CAN_ESR.
1: An interrupt will be generation when an error condition is pending in the CAN_ESR.

Nevertheless this improvement can be done.Thank you for your interest in our STM32Cube solution.Please continue providing valuable feedback. With regards.

[dunkindonut]

Posted on August 07, 2014 at 09:54

Hi Heisenberg,

thanks for your comment.

There is also a timing problem in ''HAL_CAN_Transmit_IT(CAN_HandleTypeDef* hcan)''.

The macro ''__HAL_CAN_ENABLE_IT(hcan, CAN_IT_TME);'' is called one line to early, so sometimes the ''HAL_CAN_TxCpltCallback()''  ist called when data is not send at this time.

Wrong Code:

..

..

      /* Enable Error Interrupt */

      __HAL_CAN_ENABLE_IT(hcan, CAN_IT_ERR);

      

      /* Enable Transmit mailbox empty Interrupt */

     __HAL_CAN_ENABLE_IT(hcan, CAN_IT_TME);    

      /* Request transmission */

      hcan->Instance->sTxMailBox[transmitmailbox].TIR |= CAN_TI0R_TXRQ;

    }

  }

  else

  {

    return HAL_BUSY;

  }

 

  return HAL_OK;

}

Working Code:

..

..

      /* Enable Error Interrupt */

      __HAL_CAN_ENABLE_IT(hcan, CAN_IT_ERR);

      

      /* Enable Transmit mailbox empty Interrupt */

      /* Request transmission */

      hcan->Instance->sTxMailBox[transmitmailbox].TIR |= CAN_TI0R_TXRQ;

      __HAL_CAN_ENABLE_IT(hcan, CAN_IT_TME);    

    }

  }

  else

  {

    return HAL_BUSY;

  }

 

  return HAL_OK;

}

Regards

Helmut

[ma.alam9]

Posted on September 14, 2016 at 14:21

Hello All,

I have been using CAN Tx Rx with polling and timeout for a long time but now I want to do it with interrupts. I am usnig an STM32f334c8t7 processor which is interfraced to Microchip's 2562 CAN transciever. I can enable interrrupt in the IER and some CAN messages are sent (6 to be precise)  before it all freezes. I can actually not make up the sequence for transmitting via interrupts. When should i enable the TME interrupt when and what should be in the TX Complete callback function. Is it necessary to implement it or not? can somebody please draw a basic outline of the sequence. PS: I am using MX Cube for code generation and i think somehow this HAL locking unlocking is messing it all up. I want to quickly fill all the 3 mail boxes one by one and then they get transmitted and i get the interrupt upon which i would transmit even more messages and so on.

Thanks