TMS320F28379D: Looking for support with code for workaround option

Hi Pranav,

The workaround is described in the application report. Do you have a specific question on it that I can clarify ?

Thanks

Hello Prarthan,

I am confused on how to implement it. Should I just use an if condition to compare my present CMPA with the present TBPHS and update it to (TBPHS+1) if it is smaller? I tried that but it is affecting the phase shifting between my PWMs. I think I am missing out on something else along with this logic

Hi Pranav,

Yes, you need to compare the TBPHS value with the CMPA. Because if TBPHS value is more than CMPA you will miss the CMPA event on up count and if TBPHS value is less than CMPA it will miss the CMPA event on down count.
I dont understand how that affects phase shift ? Can you explain.

See posts below where this is discussed : 

https://e2e.ti.com/support/microcontrollers/c2000-microcontrollers-group/c2000/f/c2000-microcontrollers-forum/1161072/tms320f28033-when-tbphs-load-value-is-less-than-cmpa-value-the-below-figure-will-the-aq-action-be-lost

https://e2e.ti.com/support/microcontrollers/c2000-microcontrollers-group/c2000/f/c2000-microcontrollers-forum/1294852/tms320f280039c-pwm-skip-when-tbphs-value-is-less-than-cmpa/4930730?tisearch=e2e-sitesearch&keymatch=TBPHS%252525252525252520greater%252525252525252520than%252525252525252520cmpa%252525252525252520epwm#4930730

Thanks.

Hello Prarthan,

If I include the following code, I can see the issue is resolved but it affects the 180 degree phase shift and equal duty maintained between EPWM1A and EPWM2A. I have attached the code snippet and the affected waveforms.

if(new_phase > new_cmpa)
{
    cmp = new_phase + 1;    
}
else
{
    cmp = new_cmpa;
}

EPwm2Regs.CMPA.bit.CMPA = cmp;  

EPwm2Regs.TBPHS.bit.TBPHS = new_phase;

EPwm1Regs.GLDCTL2.bit.OSHTLD = 1;        

Hi Pranav,

Since the SYNC can happen on any TB CTR value before the CMPA, the difference between counter value when sync happened and CMPA is lost in the duty cycle, but we cant adjust that unless we know sync position precisely.
This workaround prevents the condition where PHS value load causes the compare event to miss 

Thanks.

Hi Prarthan,

What can we do to resolve my issue with the incorrect phase shifting due to this workaround? Have I implemented the workaround correctly?

Hi Pranav,

As I mentioned above if we know exactly where the SYNC TBCTR position we can add that to the CMPA and achieve the exact duty cycle.

But in case we don't know where SYNC would happen on TBCTR we cant compensate for that in the intermediate CMPA setting.