S32K148 CAN physical layer issues

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

S32K148 CAN physical layer issues

Jump to solution
2,457 Views
michael_kirches
Contributor III

Dear community,

I have a question regarding the CAN configuration of S32K148.

For our current project, we are using CAN interface in combination with a third-party CAN stack. For a first start with CAN, we tried to get the CAN 2.0 example running, which is described in S32K cookbook. This one worked fine, we could send and receive messages.

With the integration of the CAN stack, we are now facing issues, which indicate that there's something wrong with the physical layer. From a CAN tracing tool, we try to send a single CAN message to test, if our hardware can receive it. In the tracing view of the CAN tool we can see that the defined message is put on CAN bus, but we also see a lot of error frames, which report about stuff errors and form errors. According to the CAN stack vendor, this is a clear indication, that there's something wrong with the physical layer. Thus I'd like to ask, if you can see a problem in our current CAN configuration:

void CAN_port_init()
{
    printf("+++ CAN_port_init\n");
    PCC->PCCn[PCC_PORTE_INDEX] |= PCC_PCCn_CGC_MASK; /* Enable clock for PORTE */
    PORTE->PCR[4] |= PORT_PCR_MUX(5); /* Port E4: MUX = ALT5, CAN0_RX */
    PORTE->PCR[5] |= PORT_PCR_MUX(5); /* Port E5: MUX = ALT5, CAN0_TX */
    printf("--- CAN_port_init\n");
}

void CAN_configure_hw() {
    printf("+++ CAN_configure_hw\n");
    PCC->PCCn[PCC_FlexCAN0_INDEX] |= PCC_PCCn_CGC_MASK;
    CAN0->MCR |= CAN_MCR_MDIS_MASK;
    CAN0->CTRL1 &= ~CAN_CTRL1_CLKSRC_MASK;
    CAN0->MCR &= ~CAN_MCR_MDIS_MASK;
    CAN0->MCR |= CAN_MCR_FRZ_MASK | CAN_MCR_HALT_MASK;
    /* Good practice: wait for FRZACK=1 on freeze mode entry/exit */
    while (!((CAN0->MCR & CAN_MCR_FRZACK_MASK) >> CAN_MCR_FRZACK_SHIFT)) {
    }

 /*
 * 00 7A 00 03 (values from GENy for configuration: Clock = 8MHz, BaudRate = 500kBaud)

   Checking settings:
    - PRESDIV: 0
    - RJW: 1
    - PSEG1: 7
    - PSEG2: 2
    - PROPSEG: 3

      Sclock = f_canclk / (PRESDIV+1) = 8000000 / 1 = 8MHz
      Tq = (PRESDIV+1) / f_canclk = 1/8000000 = 125ns ( Time Quantum )
      BitTime = (1 + [PSEG1+PROPSEG+2] + [PSEG2+1]) * Tq = (1+12+3) * Tq = 16 * 125ns = 2µs
      BitRate = 1 / BitTime = 1/2µs = 500kBaud (OK)
 *
 */

 // CAN_TIMING_REGISTER is '#define CAN_TIMING_REGISTER 0x007A0003'
 CAN0->CTRL1 = CAN_TIMING_REGISTER; //0x007A0003;

 for (int i = 0; i < 128; i++) {
    /* CAN0: clear 32 msg bufs x 4 words/msg buf = 128 words*/
    CAN0->RAMn[i] = 0; /* Clear msg buf word */
 }
 for (int i = 0; i < 16; i++) {
    /* In FRZ mode, init CAN0 16 msg buf filters */
    CAN0->RXIMR[i] = 0xFFFFFFFF; /* Check all ID bits for incoming messages */
 }
 CAN0->RXMGMASK = 0x1FFFFFFF; /* Global acceptance mask: check all ID bits */

 CAN0->MCR = 0x0000001F; /* Negate FlexCAN 1 halt state for 32 MBs */
 while ((CAN0->MCR && CAN_MCR_FRZACK_MASK) >> CAN_MCR_FRZACK_SHIFT) {
 }
 /* Good practice: wait for FRZACK to clear (not in freeze mode) */
 while ((CAN0->MCR && CAN_MCR_NOTRDY_MASK) >> CAN_MCR_NOTRDY_SHIFT) {
 }
 /* Good practice: wait for NOTRDY to clear (module ready) */
 printf("--- CAN_configure_hw\n");
}

void CAN_init()
{
   printf("+++ CAN_init\n");
   CAN_configure_hw();

   // Calls to third party CAN stack...

   printf("--- CAN_init\n");
}

In order to initialize CAN, we call methods  CAN_port_init and CAN_init as described above. 

Do you see anything which could lead to error frames with the above configuration? Any help would be appreciated.

Kind Regards,

Michael

0 Kudos
Reply
1 Solution
2,233 Views
PetrS
NXP TechSupport
NXP TechSupport

Hi,

the setting looks normal. You can try to test different CAN bit timing setting, e.g the one from S32K cookbook.

There is also 16 time quanta used for bit time, but different sample point is set.

Anyway you should have same sample point on MCU side as well as on CAN tool.

BR, Petr

View solution in original post

4 Replies
2,233 Views
michael_kirches
Contributor III

Hi Petr,

thanks for your reply. We tried with different sample points and checked settings on both sides (MCU and CAN Tool). Still error frames are shown. As we are currently using the S32K148 evaluation board, I want to reassure ourselves that no special initialization of the CAN tranceiver is  required, all settings should be applied properly for CAN usage by chosing correct pin mux and CAN bit timing settings. Is our understanding correct?

BR, Michael

0 Kudos
Reply
2,233 Views
PetrS
NXP TechSupport
NXP TechSupport

Hi Michael,

from the factory the SBC comes in Forced Normal mode and no special init is needed for the SBC and its CAN transceiver. If the SBC leaves this init mode (by clearing FNMC bit of the SBC configuration control register) then the SBC init must be done. Or restoring factory preset values can be performed as mentioned in SBC datasheet.

Also discussed here https://community.nxp.com/message/1047022#comment-1105605

BR, Petr

2,233 Views
michael_kirches
Contributor III

Hi Petr,

thanks for the link, I will check it in case nothing else remains to check. Anyway, I think SBC has not been changed so far.

Currently we are quite confident, that our problems are not related to CAN module configuration.

BR, Michael

0 Kudos
Reply
2,234 Views
PetrS
NXP TechSupport
NXP TechSupport

Hi,

the setting looks normal. You can try to test different CAN bit timing setting, e.g the one from S32K cookbook.

There is also 16 time quanta used for bit time, but different sample point is set.

Anyway you should have same sample point on MCU side as well as on CAN tool.

BR, Petr