• State TI Thinks Resolved
  • Locked Locked
  • Replies 12 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 1089 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Original question:

TMS570LS3137 Watchdog Timer Expire Reset

TMS570LS3137: TMS570LS3137 Watchdog Timer Expire Reset

Hugh Shane
Intellectual 565 points
Part Number: TMS570LS3137
Other Parts Discussed in Thread: HALCOGEN

In a related thread, the application engineer seems to be saying that the Digital Windowed Watchdog Reaction Control (RTIWWDRXNCTRL) register specifies the reaction of the DWD when the DWD counter is decremented to zero as well as the reaction of the DWWD if the watchdog is serviced outside the time window. Am I reading this correctly?

Hugh

  • 0
    TI__Guru**** 196566 points
    Hi Hugh,

    You are right. As I mentioned in another thread, please don't service the dog (write key to WDKEY register) before the watchdog window is opened up. Otherwise, it will reset the device.
  • 0 in reply to QJ Wang
    Intellectual 565 points
    QJ,

    I understand. I've configured my WDT for a window size of 100% using this HALCoGen API:

    dwwdInit(Generate_Reset, 218, Size_100_Percent);

    However, I'm still observing occasional system resets due to the WDT system. (If I disable the WDT system, the resets don't occur.) When I read the RTIWDSTATUS register after these resets I see the value 0x00000032 or 0x0000003A, which seems to indicate a window violation. Why might this be happening if I specified a 100% window size?

    Hugh
  • 0 in reply to Hugh Shane
    TI__Guru**** 196566 points
    Hi Hugh,

    The correct sequence is:
    1. configure preload, window size, reaction
    2. enable DWD (RTIDWDCTRL)
    3. service the DWD (RTIWDKEY)

    If you service the dog before step #2, it will trigger the reset.
  • 0 in reply to QJ Wang
    Intellectual 565 points

    Yes, I'm following the sequence you specify.

  • 0 in reply to Hugh Shane
    TI__Guru**** 196566 points
    Hello Hugh,

    You said "If I disable the WDT system, the resets don't occur." Did you disable the WDT by writing a wrong password(5312 ACEDh) to RTIDWDCTRL to disable the DWD?

    Can you share your code with us? Thanks
  • 0 in reply to QJ Wang
    Intellectual 565 points

    When I say "disable the WDT system" I mean don't initialize the WDT following a reset and don't write to the WDKEY register at any time during program execution. I fear my previous remark in this regard has distracted us from the central issue...

    My point of confusion concerns the values that I'm seeing in the RTIWDSTATUS register following a reset generated by the WDT. In spite of the fact that I've specified a 100% window size, the register is indicating that a time-window violation has occurred (bit 5) and some combination of end time violation (bit 4) or start time violation (bit 3). Are these indications what you would expect in the case where the DWWD feature isn't enabled?

    It would be difficult to share my code in its current form. I'd have to create a stripped-down version of my software

  • 0 in reply to Hugh Shane
    TI__Guru**** 196566 points

    Hello Hugh,

    I understand that servicing the dog before the DWD counter is enabled should not reset the device. 

    I did many test using the following code, but did not generate reset. 

    i=1;
    /* Initialize and start DWD */
    dwwdInit(Generate_Reset, 4095, Size_100_Percent);
    dwdCounterEnable();
    while(1){
         if(i==1){
               dwdReset();
         }
    }

     

  • 0 in reply to Hugh Shane
    TI__Mastermind 49120 points
    Hugh,

    A 100% window size means that the watchdog service window is always open. So it is essentially a timeout watchdog, once enabled. What is the frequency of the RTI1CLK domain? If it is 90MHz, then you are allowing for only 2.433 us (219 / 90MHz) before the watchdog generates a reset.

    Is this the timeout period you desire?

    Regards,
    Sunil
  • 0 in reply to Sunil Oak
    TI__Guru**** 196566 points
    219*2^13/90Mhz = 19.7ms
  • 0 in reply to QJ Wang
    TI__Mastermind 49120 points
    Of course, thanks QJ. Sorry for the confusion on my part.
  • 0 in reply to Sunil Oak
    Intellectual 565 points

    @Sunil: The 100% window behavior you describe is also how I understand the system to work.

    @QJ: I concur with your calculation. I'm looking for a timeout period of approximately 20 msec.

  • 0 in reply to Hugh Shane
    TI__Guru**** 196566 points
    Hello Hugh,

    I did many (>10000) test using following code, and did not get the reset issue:

    i=1;
    dwwdInit(Generate_Reset, 4095, Size_100_Percent);
    dwdCounterEnable();

    while(1){
    if(i != 0){
    dwdReset(); //feed the dog after the counter is enabled
    i++;
    }
    }

    1. configure the reaction, preload, window size
    2. enable the counter
    3. service the dog