F28035 eCAN behaviour different when initialized by JTAG

Hello Pierre,

I'd recommend taking a look at the below wiki page:
http://processors.wiki.ti.com/index.php/C2000_Flash_Common_Issues/FAQs#Running_Code_Standalone

Particularly, I would make sure that any global variables used are being initialized inside main instead of with their declaration.


Thank you,
Brett

Hello Brett,

I've taken a look at the wiki page and I don't think it's of any help. Everything else in my program works fine running from Flash. The eCAN itself runs fine most of the time, except for the delayed edges I was mentioning.

I don't understand: could you please explain why I need to initialize global variables inside main? I think that's the job of the _c_int00 function, using the values from the .cinit section, and I've checked that it does it job with or without a JTAG emulator.

Best regards,

Pierre

The eCAN reference guide (SPRUGL7) states that bit 10 in the CANBTC register does nothing and is read-only.

Well it's not. You can definitely set it to 1, and when you do the eCAN starts resynchronizing on rising (dominant to recessive) edges, which is not allowed in the CAN 2.0B standard.

As you can see below, with TQ = 100ns and SJW = 2 and CANBTC[10] = 1 all recessive bits last 1.2us instead of 1.0us.

Best regards,

Pierre

Hi Pierre,

Sorry, I kept meaning to get back to this but I wanted to when I had enough time to put thought into it :).

Thanks for posting what you've found.  I'll follow up with the product team.


Thank you,
Brett

Hi Pierre,

Sorry for the delay in response. 

We have verified that what you are saw is accurate and have filed a documentation bug against this issue.  A future update of the reference guide will state that bits 10-15 of the CANBTC register are RESERVED and should only be written with all-zeros.


Thank you,
Brett