configuring DCO

Hi,

Thanks for your reply. But I am still facing a problem. XT2 is not present on my board but a 32.768 KHz watch crystal  connects to Xin & Xout.

I am configuring DCO for SMCLK clock sourced from the watch crystal.

Here is my code:

  FLL_CTL0 |= DCOF + DCOPLUS; //DCO oscillator fault condition present, DCO O/P is not divided
  FLL_CTL1 |= XT2OFF;  //Turn off XT2,
  SCFI0 |= FN_2;// f_DCO range = 1.3 to 12.1 MHz
  SCFI0 &= ~FLLD_1;// divide f_DCOLCLK by 1
  SCFQCTL |= 0x1F; //f_DCOCLK = 1.(31 + 1).f_crystal = f_DCOCLK = 32x32.768 kHz = 1.048576 MHz
 
  // Loop until 32kHz crystal stabilizes
  do
  {
      IFG1 &= ~OFIFG;                     // Clear oscillator fault flag
      for ( i=50000; i==0; i--);           // Delay
  }
  while (IFG1 & OFIFG);                   // Test osc fault flag
 
  // disable watchdog
  DisableWatchdog();    // Stop WDT - Password(write - 05Ah) + disable WDT

 

Problem : When I am setting FLL_CTL1 |= XT2OFF;  //Turn off XT2,system is getting reset always. After commenting it, receiving and sending are still not working.

 

Saurabh Arora said:

FLL_CTL0 |= DCOF

DCOF (and the other OF bits) is set if an oscillator fault is detected (and NOT reset automatically). You can only reset it, if you think the fault condition has been removed, and it will reappear after some time (some 100us) if the oscillator is still not oscillating. If you manually set this bit, you tell the clock system that there is no working DCO. MCLK will then switch to XT2 (which isn't there). and then to VLOCLK (LFXT1CLK is not up yet). the WDT is also running on VLOCLK, so the watchdog will reset your system after 32768 VLO clock cycles, long before you disable the watchdog. Also, after setting DCOF, SMCLK will switch from DCO to XT2 as the only available clock source, and then you disable XT2, leaving SMCLK without any clock source.

You need to disable XT2 (if your device has an XT2 input), or OFIFG will never clear, since XT2OF will be always set.

Saurabh Arora said:

for ( i=50000; i==0; i--);           // Delay

And once again the classical delay loop. Since I isn't used inside the loop or ever after, any good optimizing compiler will completely eliminate this as dead code.
If you do something inside the loop that isn't redundant (e.g. writing I into a hardware register that doesn't do anything), the loop will work as expected. As you do now, it will or won't, depending on the used compiler and its optimisation settings.

Also, you do not set the XCAPxPF bits for the watch crystal capacitors. Do you have external capacitors? If not, the crystal will never come up.