STM32 CAN bus

[pilotak]

I tried CAN bus sucessfully on LPC1768, now i moved to NUCLEO_F142ZG and NUCLEO_F103RB but i can't get it to work. Same code works on LPC but doesn't on STM32. Is CAN not supported on STM? I looked at transmit pin through scope and there is nothing = pin HIGH.

#include "mbed.h"
CAN can1(PB_8, PB_9);
DigitalOut led1(LED1);
char counter = 0;

int main() {
    printf("start\n");
    
    can1.frequency(250000);
    can1.reset();
    can1.mode(CAN::Normal);

    while(1) {
        if(can1.write(CANMessage(1337, &counter, 1))) {
            printf("sent\n");
        } 
        else {
            printf("error\n");
        }
        counter++;
        led1 = !led1;
        
        wait_ms(100);
    }
}

It prints 3x "sent" (but not visible on scope) and then error forever

Description

• Type: Question
• Related issue: #abc
• Priority: Blocker | Major | Minor

---

Target
STM32

Toolchain:
GCC_ARM & online

[LMESTM]

Hi @pilotak
You get an error after 3 messages being sent because there are 3 transmit mailboxes in STM32 F103RB and none of them is free. So it seems like the transmission is failing. How have you connected the board to the CAN bus ? Do you have tranceivers in place like suggested here https://developer.mbed.org/handbook/CAN ?

[pilotak]

Hi @LMESTM ,
As i already explained , it works perfectly on LPC1768 so yes it tranceivers are working and they are connected as suggested. I also tried to swap RXD/TXD just in case but it's the same - no activity on those two pins, checked with scope.

I did test on LPC: when i disconnect transmitter from LPC i can still measure activity on TXD pin - it tries to send data.

F412ZG and F103RB there is no activity, even that program sends me sent (which should be confirmation that it arrived to the mailbox) there is no activity on TXD pin. When i press restart button on nucleo RXD and TXD gets LOW for as long as button is pressed, otherwise HIGH all the time

[LMESTM]

@pilotak
I think I could reproduce the issue.
There is a recent fix on can reset that has not been merged yet #4932
until this fix is merged, the reset is not well managed.
Can you try one of 2 possibilities:

• either cherry-pick fix from Fix STM32 CAN reset to not lose context #4932
• or remove the call to can1.reset() from your example as this is not needed.

Let me know if any of the 2 above proposals work ok.
If this is ok, we just have to wait #4932 to be merged.

[pilotak]

@LMESTM
by removing reset() it works. Thanks a lot

I tried to checkout your PR #4932 which didn't work for me, but i'm sure it works i just didn't want to spend time finding why (because removing one line from code is faster )

I guess we will have to wait until merge to properly work.

[LMESTM]

Thanks for your quick feedback - glad it works.
When #4932 is merged, you can re-introduce the use of reset in your code (if/when reset is needed ... ;-) )

[pilotak]

just for reference, it has been merged

[amrithmmh]

will this work on stm32f103ct6 ? I meant the default CAN api in mbed

[pilotak]

should, you can try