TCAN4550-Q1: Issue with Tcan4x5x Driver

Hi Ravindra,

This bit is part of the CAN FD Controller IP called "M_CAN" that is used in the TCAN4550 and developed by Bosch.  The M_CAN User's Manual (Link) provides the following explanation for this bit.

If you are trying to access an invalid register address or a reserved address, then this bit can be set.  Please verify you are only trying to read or write to a valid register address.

Regards,

Jonathan

Hi Jonathan,

Thanks for the reply.

We are getting this error on a valid register access. The error happens when we read a valid register as part of interrupt processing. In the interrupt handler task (woken up by isr), we first clear IR register. After clearing, we verify that IR is zero. Then we read 10F4 (FIFO status in case of Tx completion) and 1044 (PSR in case of Rx). In either case, immediately after accessing these registers, when we read back IR, it shows 0x20000000. 

Just to cross check whether it is only the problem with 10F4 or 1044 registers, we read some other registers (for debugging). We get Reserved Address Access error in some registers like 10F8, 1048, 1024, etc. But not when we access registers like 1004, 10C0, 10D0! All these are read operations. 

The bigger problem is that even if we write back the same value to IR to clear, it does not get cleared. 

Hi Ravindra,

I am unable to replicate the ARA bit getting set while reading the registers you have indicated.  I followed your process of first reading and clearing the interrupt register 0x1050 and then I tried reading these registers and between each register read I read the interrupt register 0x1050 to see if the ARA bit was getting set.  But ARA=0 for all reads of 0x1050.

Is it possible to log the SPI signals with a logic analyzer to see how the signals are being generated according to our expectations?  Perhaps there is some form of logical error or timing error that is causing the device to se an incorrect address.

Regards,

Jonathan

Hi Jonathan,

Enclosed is a capture from the logic analyzer.

Hi Ravindra,

If I'm decoding this correctly the MOSI data bytes are: 0x41, 0x00, 0xC4, 0x00, 0xFF, 0xFF, 0xFF, 0xFF

This would be a SPI Read (0x41) of address 0x00C4, with a Length field of 0x00.

I see two problems with this.

1.) There is no defined address of 0x00C4

2.) The length field should not be 0x00, instead it should be 0x01 because there is only one additional "word" of data (or an additional 32-bits) to be returned before the chip select signal goes high again to end the SPI read transaction.

There appears to be some error in the driver causing an incorrect bit sequence to be output from the MCU on the MOSI line, which is likely causing the ARA error bit to be set.

Regards,

Jonathan

Hi Jonathan, 

Thanks for your continued and immediate support. 

Sorry for the confusion. The above read was a result of a bad test code we put in while debugging the issue. The enclosed are the images of the IR register (0x1050) read before and after reading register 0x1044. We did three reads immediately after initializing the can device. We did not perform any other operations before testing these reads.  As can be seen immediately after accessing 0x1044 register, the IR register value has change from 0x00000000 to 0x20000000.

Thanks

Hi Jonathan,

Thanks a lot for your support. That seems to have solved this issue. Appreciate all your help.

Ravindra.

Hi Ravindra,

Thanks for letting me know this issue is resolved.  I'm glad I could help.

Regards,

Jonathan