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.

MSP430F5635 Watchdog Timer Accuracy

Other Parts Discussed in Thread: MSP430F5635

The Issue 

  • The PRODUCT product clock becomes slow sometimes to about 30s off after 1 or 2 hours, depends on unit.  
  • The real time clock runs from a timer on the MSP430.  Timer source is the 32khz crystal and ACLK ocscillator.  

Ideas

1.Our product uses a new chip, the MSP430F5635, maybe the chip has an issue?, or

2.The setup of the timer could be wrong, or 

3.The crystal itself could be bad, or

4.Some other interrupt or other operation in the system is corrupting the clock values

 

 

For #1, I called our contact at Avnet (Rinzai.bell@avnety.com): he never heard of the issue, but he will check with TI

 

For #2 above, We use an older MSP430 in another released product. I thoroughly checked code: I migrated and checked the SW down to every bit in every register… I cannot find any error in this setup.  

 

For #3, we checked the crystals: they gave error of typical 100 sec per month or less, standard for all the products

 

For #4, We now setup a basic program with the same OLED hardware, but only the timer running, to eliminate the rest of the dive computer… we will check at 4pm today … if it is of by even a few seconds, then we know the timer itself does not work right, because we have isolated it in the program and nothing else is running.  

 

Update: this test was done, a simple program was setup to run the WDT in counter mode, using a 1000ms interrupt, setup using the code below, provided by TI supplied header files.  After 1 hour and 40 min the time was off by 17 . At this point the only issue we can think of is the crystal or the MSP430F5635, and we believe it is not the crystal, it has to be the MSP430 ... please look into this as it is a production release for us, critical, and very bad for our product.  We will not use this chip if it has an issue.  

 

MOV.W   #WDT_ADLY_1000,&WDTCTL      ; WDT 1000ms, ACLK, interval timer

  • Rinzai Bell said:
    Our product uses a new chip, the MSP430F5635

    What is the old chip? How do you initialize LF oscillator of both?

    Rinzai Bell said:
    and we believe it is not the crystal

    Don't believe but test. - Solder crystal in board you trust and test.

    Are you aware of crystal load capacitance? - It is suspicious that you don't even mention it..

  • Check the ACLK source. The 5xx chips default to the REFO and not LFXT1 on POR. Have them check the clock register settings as well as the OFIFG flags for the xtal.

    Thx,

    Rinzai

  • You can be sure: the WDT is counting correctly. It is a simple 2^n counter. It either counts to 2^n clock pulses or it doesn’t count at all.
    The problem must be the clock you’re feeding into it.

    How did you check the crystals? If you feed the crystal into SMLK, try to output SMLCK on a clock pin and do a precision frequency metering. 2800ppm are unlikely, but with wrong load caps it is not impossible.

    Also, how do you use the WDT and count the time? Not just the WDT init code, but your ISR or whatever? Maybe the MSP resets on every WDT interrupt (once a second).This could well add an error of 17s per 1:30h

**Attention** This is a public forum