TMS320F28P650DK: Trip zone force action: strange behavior

Hi Ettore,

Since you are using a software force to trip your PWM signals, I would expect there to be some delay between the timing of when each PWM gets tripped. However 4us seems very large, I would expect something in the tens of nanoseconds at worst.

One approach you could take to have no delay between the PWMs getting tripped is to write to a GPIO output which is read by any of INPUTXBAR[1-3], and use the TZ[1-3] events to trip your PWMs. Let me know if this is a suitable approach. If not I can help investigate the cause of the 4us delay.

Thank you,

Luke

Hi Luke,

thank you for your reply.

Please let me dig my code, and in case I need your support I will back.

In the meantime, may I leave the ticket open ?

Hi Ettore,

Yes I will mark the ticket as "Waiting for Customer". It will reopen once you reply, however please respond with an update within 28 days or this thread may become locked.

Thank you,

Luke

Hi Luke, 

I'm still digging for the strange behavior. 

I have read the "Enhanced Pulse Width Modulator" chapter many times and it states that writing "1", through user code, on the OST field of the TZFRC register, the action immediately affects Pwm outputs. 

However, based on what I see on my code, there is a delay among the action and the effect; the delay time is a random value, but always, it is as integer number of half interrupt period or 3.3us time quanta.

Further, if I put a breakpoint on the code and I write the OST bit directly on Chip's register, through the Register Window I found on CCS, the flag OST on the TZFLG register doesn't rise even if I run the code step by step for a long. Please note that before this setting, I have set EALLOW on ST1.

Based on this observation, it seems that the "setting of TZCTL" register, acts with as shadow register or something like that, but I haven't found it on the user manual.

Does my story rise to you an idea about the issue ?

PS: I have repeated some tests with an evaluation board with a TMS version of the chip, but the results didn't change.

Thank you,

Ettore

Hi Ettore,

Could you try setting the FREE_SOFT bits in the TBCTL register to 11? It's possible the emulation mode for the PWM module is affecting the behavior of the TZFRC when the debugger is connected. If that doesn't have any affect I will test this issue on my side.

Thank you,

Luke

HI Luke

sharing file offline

thank you 

BR
Carlo

Hi Colombo,

Our SME is currently OOO, expect responses to be delayed till next week.

Best,

Ryan Ma

Hi Carlo,

I went through the file you sent via email. From your observations it seems the trip has a faster response time to an external trip compared to a software forced trip.

Do you still wish to use the software forced trip for each of the EPWM modules, or are you planning on writing to a GPIO as an external trip source instead?

Thank you,

Luke

Hi Luke,

what I want to do is:

1) GPIO as an external trip source for entire ePWM modules allocated.

2)  software forced trip for entire ePWM modules allocated.

The case 1, is a typical switch off due to an electrical protection, while case 2 is related to a switch off by the code.

For my application, case 2 must be as fast as case 1, and it must affect entire ePWM modules allocated.

Thank you,

Ettore

Hi Ettore,

Understood. Do you have a spare GPIO you could use to implement case 2 so that the EPWM modules are all tripped simultaneously, or do you need to use the software forced trip? There will be an unavoidable delay between when the different EPWM modules are tripped if you trip them one by one in software. 

Thank you,

Luke