CC2640 SCLK_LF Clock Fault Handling

Hi Robb,

I am looking into this. Some of our core experts are out on vacation, so I will need a couple of days to figure this out.

Best regards,
Fredrik

Hi Fredrik,

Thank you for digging into this.  I am going to try some experiments myself.

Cheers,

Robb

Hi Robb,

I have not been able to get firm answers to your questions yet due to people being out on vacation, but here are my assumptions at least:

1) I think you are right. Nothing will happen automatically, and the code should switch to RCOSC_LF if the XOSC_LF fails and causes a reset.
2) Same as above
3) Is the question "what happens if we switch to XOSC_LF and it does not start up"?

Best regards,
Fredrik

And one more thing: At start-up, they must wait until after they have switched to XOSC_LF before turning on the clock loss detector.

Hi Fredrik,

I have been playing with the SimpleBLEPeripheral example code, and I have added the following to the main() loop before applying the BIOS_start().

=========================================

currentStat0 = (HWREG( AUX_DDI0_OSC_BASE + DDI_0_OSC_O_STAT0 ) & DDI_0_OSC_STAT0_SCLK_LF_LOSS);

currentResetCtl = (HWREG( AON_SYSCTL_BASE + AON_SYSCTL_O_RESETCTL ) &    

                             AON_SYSCTL_RESETCTL_RESET_SRC_M);

switch(currentResetCtl) {

   case AON_SYSCTL_RESETCTL_RESET_SRC_CLK_LOSS:    

       if (currentStat0  == DDI_0_OSC_STAT0_SCLK_LF_LOSS) {  // If SCLK_LF_LOSS occurs, switch to RCOSC_LF

           GPIOPinWrite(GPIO_PIN_6, 1);        

           OSCClockSourceSet( OSC_SRC_CLK_LF, OSC_RCOSC_LF );

           while( OSCClockSourceGet( OSC_SRC_CLK_LF ) != OSC_RCOSC_LF ) {

               // Wait until switched

           }              

       }    

       else {    // Then SCLK_HF_LOSS occurred, start OSC_XOSC_LF (External 32.768kHz Crystal)

           GPIOPinWrite(GPIO_PIN_7, 1);        

           OSCClockSourceSet( OSC_SRC_CLK_LF, OSC_XOSC_LF );

           while( OSCClockSourceGet( OSC_SRC_CLK_LF ) != OSC_XOSC_LF ) {        

               // Wait until switched

           }              

       }    

       break;

   case AON_SYSCTL_RESETCTL_RESET_SRC_VDDS_LOSS: // Did VDDS_LOSS cause reset?

       GPIOPinWrite(GPIO_PIN_25, 1);    

       OSCClockSourceSet( OSC_SRC_CLK_LF, OSC_XOSC_LF );

       while( OSCClockSourceGet( OSC_SRC_CLK_LF ) != OSC_XOSC_LF ) {

           // Wait until switched

       }                

       break;

   case AON_SYSCTL_RESETCTL_RESET_SRC_SYSRESET:            

       GPIOPinWrite(GPIO_PIN_27, 1);  

       OSCClockSourceSet( OSC_SRC_CLK_LF, OSC_XOSC_LF );

       while( OSCClockSourceGet( OSC_SRC_CLK_LF ) != OSC_XOSC_LF ) {

           // Wait until switched

       }                  

       break;            

   default:     // Default, start OSC_XOSC_LF (External 32.768kHz Crystal)

       OSCClockSourceSet( OSC_SRC_CLK_LF, OSC_XOSC_LF );

       while( OSCClockSourceGet( OSC_SRC_CLK_LF ) != OSC_XOSC_LF ) {          

           // Wait until switched

       }            

       break;

}

// Turn ON Clock Loss Detection.

HWREG( AUX_DDI0_OSC_BASE + DDI_O_SET + DDI_0_OSC_O_CTL0 ) = DDI_0_OSC_CTL0_CLK_LOSS_EN;

=========================================

This is to better understand the RESETs.  

On Powerup or EM Reset, I run through default and set the OSC_SRC_CLK_LF to use the external 32.768kHz Crystal.

When I short X32K_Q2 off the Crystal to GND, it generates an SCLK_LOSS Reset.  At this point, I check for SCLK_LF_LOSS in STAT0.  If this is set, then I change SCLK_LF to OSC_RCOSC_LF.

At this point, I run SCLK_LF from OSC_RCOSC_LF (Advertise and Connect Possible).

Now, when I short X32K_Q2 off the Crystal to GND, it still generates an SCLK_LOSS Reset.

Am I doing something wrong with the switch?  I would not have expected the SCLK_LOSS to occur if I am running off of the internal RCOSC_LF.

I do have a follow-up question about the BOD operation, but I will start this as a separate post.

Thanks,

Robb

Hi, Robb:
My customer has 32K external osc failure issue, the failure protect code as your showed seems could help solve the problem.
So have you already dig out the bug and make it work as you expected? Thanks!

Vivian