CC2745R10-Q1: Detection of high-frequency clock loss

Hello, sorry for the delay, we'll get back to you on July 1st.

How about this matter?

Sorry, but I appreciate your reply.

Hello Toshiki, 

Apologies for the delay. I am working with the power driver expert to understand. 

Can you reiterate the question a bit? Are you expecting the function PowerCC27XX_oscillatorISR() to occur when there is high frequency clock loss, or high frequency clock loss to occur when PowerCC27XX_oscillatorISR() is executed? 

From a brief review of the function, I do not believe the CPU will be reset. 

Let me know. 

Thanks,
Isaac

Hi,

High-frequency clock loss occurs -> interrupt handler PowerCC27XX_oscillatorISR() is called.

This is the expected flow.

At this time, are you restarting the oscillator within PowerCC27XX_oscillatorISR()?

Hello,

I'm going to take a look into this, and get back to you tomorrow.

Best,

Nima Behmanesh

Hello,

Toshiki Nakamura said:

At this time, are you restarting the oscillator within PowerCC27XX_oscillatorISR()?

This is correct, the oscillator is restarted within PowerCC27XX_oscillatorISR. This occurs when there is TRACKREFLOSS or HFXTFAULT.

Best,

Nima Behmanesh

Hi

If a high-frequency clock loss occurs, the interrupt handler PowerCC27XX_oscillatorISR() restarts the oscillator, but is a CPU reset unnecessary?

I am concerned that restarting the oscillator may have some effect on software operation.

Hello,

You may want to reset the CPU as well. I spoke to a colleague, and they mentioned that although the CPU reset doesn't occur, it should since the digital and analog domains could be out of sync.

So I believe that a reset is necessary.

Best,

Nima Behmanesh

Hi,

In order to perform a CPU reset, I need to notify the user application that a "high-frequency clock loss" has been detected.

"High-frequency clock loss" is not one of the eventTypes elements that can be specified with Power_registerNotify().

Is there any other way to notify the user application that a "high-frequency clock loss" has been detected?

Hi,

Let me take a look at see how this can be achieved. Please allow me until Wednesday to get a response.

Best,

Nima Behmanesh

Hello Nakamura-san, 

There is no notification for high-frequency clock loss. 

The power event PowerLPF3_HFXT_AVAILABLE occurs once at startup, and once per standby wakeup. Using this knowledge, you can check if this event occurs multiple times in a row, and if the event correlates with either startup or a standby wakeup event. If PowerLPF3_HFXT_AVAILABLE occurs multiple times without a standby wakeup event, then an HFXT fault has occurred. 

The specific fault will not be known, but when an HFXT fault occurs, the HFXT will be reset, or the software will enter a fault, so a reset is recommended. 

Let me know if this helps. 

Thanks,
Isaac

Hi,

Please tell me the number of times PowerLPF3_HFXT_AVAILABLE event must occur to judge that an HFXT failure has occurred.
If PowerLPF3_HFXT_AVAILABLE occurs twice without a wakeup, it means that there is an HFXT failure?

Best Regards,
Yukine

Hello Okita-san, 

I will verify with the team and get back to you. I believe if the PowerLPF3_HFXT_AVAILABLE occurs twice without a standby wakeup notification, then this would indicate HFXT failure. 

I will provide a response by tomorrow (21/08/2025). 

Thanks, 

Isaac

Hello Okita-san, 

This method does not provide a definitive way to identify HFXT loss. The event will occur at startup, when standby is entered, and when the device wakes up from standby. That being said, the HFXT will restart every time the device enters and exits standby. 

I think it would be sufficient to reset the CPU after the device hits the PowerCC27XX_OscillatorISR() function. I am working with the team to add a notification for HFXT loss, but there is no timeline for this notification at the moment. 

Thanks,
Isaac

Hi,

> I am working with the team to add a notification for HFXT loss, but there is no timeline for this notification at the moment. 
Please let me know the interim measures or workarounds until the above-mentioned issue is resolved.

Best Regards,
Yukine

Hello Okita-san, 

I will need to discuss this with the team. Currently there is not work around until a notification is added from my understanding. 

Thanks,

Isaac