TMS320F28379D: Matlab Simulink PWM Signal Problem with variable Frequency / Clock Cycle

Hi Kevin,

Is it possible that you can isolate the issue by scaling down your model.

Work on a single PWM module and try to reproduce the behavior.

You can then import the code in CCS and debug. Here is how you can create a CCS project from MATLAB/Simulink.

So I scaling down my program, but i must have 2 PWM signals.

It must have something to do with the "Synchronization action" i'm setting it  on "Set counter to phase value specified via input port".

When I disable this option, so the PWM signal has no phaseshift,  it works fine...

But I need the phaseshift between the signals.

Here is the generated  ccs project.https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/171/PWM_5F00_3PHASE_5F00_DAB_5F00_TEST_5F00_ert_5F00_rtw.7z

Hi Kevin,

Unfortunately you need to debug your model. But if you can pass the Frequency values at which this behavior can be reproduced and if you can provide a list of PWM signals that are affected, we can pitch in. But we need a scaled down model where we can reproduce the issue.

Few points that can help.

1. Try to calculate the period and other count values when problem occurs.

2. Try to feed the same values through constant block in a different model.

3. See if these values are blocking the PWM operation.

4. Analyzing those data points can help understand the underlying issue.

HTH,

Hi,

the thing is that I can't name precise frequencys, the effect apears at all frequencys ( tested between 1 and 250 khz)

With constant values is it no problem.

When I debug the model and switching the frequency, some PWM signals aren't working anymore.

Sometime they're apearing again, when i'm again switching to another frequency and also stay when i'm switching back to theprevious frequency. disapears

But most of the time the signals don't come back.

So it's completely random. Don't seeing a pattern.

Three PWM signals are stable and aren't affected. Marked this in the screenshot.

It must have something to do with the Counter value settig at the synchronization.

Best regards

Kevin

precise

Hi Kevin,

I just found this picture from the device data sheet.

I see in your model the ePWM is configured for 200Mhz because of which we may be seeing this problem.

Can you scale down the ePWM freq to 100Mhz and see if it helps.

Hello Venkatesh,

I changed the EPWM clock divider to SYSCLKOUT/2, but I still have the same issue.

Hi Kevin,

I could simulate the problem. However when I debugged using the CCS the values for TBPRD, CMPA and CMPB for the slave PWM matched that of Master. Hence we cannot suspect that these values are getting corrupted. I also observed the TBCTR of the Slave PWM is running which indicates the PWM module is ticking but no output. As noted by you, the pattern is random. We are looking for clue from TI what settings if present/missing could cause this behavior. That can help in investigating. I will ask them to look in to this issue.

Hello Venkatesh,

thank you for your effort.

I'm hoping you guys find a solution for this issue. It would really help me out.

Best regards

Kevin