This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

dMAX example fails on Weuffen C6727 EVM

Hi all,

We are having a problem with the dMAX example not generating the DMA completion interrupt (INT8) reliably. Whether it works or not seems to depend on startup conditions. More frequently it doesn't work. The data gets transferred however, but the interrupt never happens. The same code works fine on a PADK board according to the TI support team. We are using the PADK GEL file that was provided to us with a minor modification for the 12 MHz clock (PADK board uses 25 MHz clock). Any ideas will be highly appreciated. 

 

/Mikhail

 

 

  • Mikhail said:
    Whether it works or not seems to depend on startup conditions. More frequently it doesn't work. The data gets transferred however, but the interrupt never happens.

    Is the problem intermittent, or does it never work. These three sentences sort of say different things about that.

    If it works sometimes under specific startup conditions, what are those differences?

  • Randy,

    Thanks for your response. We now believe that it does work every time if the startup (or rather power-up) conditions are favourable. However, we haven't been able to figure out what these conditions are so far and they happen to be non-favourable more often than not. However, if it worked once after power-up, you can reload program via JTAG any number of times and it will work every time. It seems that something doesn't get initialized in the code and/or GEL file.... 

     

    /Mikhail

  • Did you get a GEL file from Weuffen for their board? Comparing the two and comparing as much as you can between the two boards may help you find the key difference.

    Starting from no special knowledge, I would be concerned about power supply sequencing and clock settings (input and internal clocks). Some pin configurations could cause problems, but it is hard to think of something that would make an interrupt fail to occur.

  • Randy,

    We couldn't find anything wrong with the GEL file, but we found that everything works fine under CCS v.3.3 but not under CCS v.4.1.2.  Any comment on this?

     

    Thanks,
    Mikhail 

  • There is no reason I can think of for the IDE to have any impact on the application running. As long as you are not doing anything with CCS interaction while the program is running, that is.

    I do not know if CCSv4 uses GEL files any differently than CCS 3.3 does. That might be a question to post onto the Code Composer Forum to get the CCS experts to offer some ideas.

    But you could try manually running all the same GEL functions that get executed automatically, like StartUp, OnTargetConnect, OnReset, OnPreFileLoaded, OnFileLoaded, and OnRestart, for example, to see if their functionality is needed and missing in CCSv4.

  • Randy, Mikhail,

    CCSv4 does not treat GEL files differently than CCS3.3 (although some minor functions are deprecated like GEL_MapAdd), but a significant point that would affect the application is the DSP/BIOS version used. CCSv4 only supports BIOS 5.4x releases, therefore if the application was developed in CCSv3.3 chances are it was developed using DSP/BIOS 5.3x. To rule out this issue I would install a release 5.4x to CCSv3.3 and retest the project with this.

    Apart from this I cannot think of any other details that could affect the application execution, since the emulator interactions are very similar between the two IDE releases.

    Best regards,

    Rafael

     

  • Rafael, Randy,

    Because we were able to successfully run the test in CCS3.3 we decided to take the compiled  output from the CCS3.3 and load it onto the target from the CCS4 IDE environment.  This was also successful, loading the CCS3.3 output with the CCS4 IDE showed the interrupt at the end of the DMA transfer.  This suggests that the GEL file and IDE are working the same for initializing the board, but maybe the compiled output is behaving differently.  We confirmed that the compile tools are the same, 6.1.12 in both IDEs.  Do you have any other ideas as too what could cause a difference in the compiled output between IDEs?

    Vladymyr

  • Hi All,

    I got Dmax interrupt working by using  c672x  ROM bootloader with System Patch V2_00_00 in CCS v 4.02 as recommended in  SPRAAS8—Using ROM Contents on TMS320C672x.

    Vladymyr


  • Vladymyr,

    Excellent, and thank you for responding back to the forum.

    For future users, would you please share which section of the document was relevant to your project?

    Regards,
    RandyP

  • Normal 0 false false false EN-CA X-NONE X-NONE MicrosoftInternetExplorer4

    Randy,

    Sections 1…4 are relevant.

    Vladymyr

  • Do you agree this thread is answered?

    How do you enter your text, such that it ends with the "Normal 0 false false..." line? I would like to submit this information to our support team. Perhaps the question is "which internet browser do you use?"

    In one posting, the document name SPRAAS8 has a link to the latest version plus a long description when you hover over it. How did you do that? I know how to create links, but getting that much information in there is more effort than most users go through. If I can learn from this, I would like to.

    Regards,
    RandyP

  • Hi Randy,
     
    Do you agree this thread is answered?
    -Yes, I think so.
     
    How do you enter your text, such that it ends with the "Normal 0 false false..." line?
    -I copied and pasted from MS Word 2007. I tried to delete this line in edit window but I could not see it there. Sorry for that.
     
    Perhaps the question is "which internet browser do you use?"
    -I am using Mozilla Firefox.
     
    In one posting, the document name SPRAAS8 has a link to the latest version plus a long description when you hover over it. How did you do that?
    -First I copied the SPRAAS8 link from CCS v 4.2 help and pasted to MS Word 2007 and then copied and pasted to E2E reply.
     
    Vladymyr