TMS320F28335: TMS320F28335

Akbari,

I would say that bit bashing a PWM waveform isn't the best idea, nonetheless it can be done. Using "DELAY_US()" isn't a good idea either, it will stall the CPU wasting cycles.

If you don't want to use the ePWM modules that leaves us with two reasonable option, 1 eCAP and 2 CPU timers.

The eCAP has an "APWM" mode of operation that allows basic PWM output, likely much better than you would achieve through toggling GPIOs. 

If you have available CPU timers it would provide consistent way to measure delays between the GPIO toggles. It will also leave your CPU free to run application code.

What is your justification for not using the ePWM module? I cannot think of a valid reason for not using the ePWM modules other than " I already allocated all of the ePWM modules and need another PWM output."

Regards,
Cody 

Thanks for your Reply Cody,

The reason I am not using ePWM modules is that I am working on multilevel inverters. in Multilevel inverters, we need more than two carries which for level-shifted SPWM method, these carries need to be shifted in amplitude. Then, these carries should be compared to the Sin reference to produce proper signals for driving power switches. In multilevel inverters, one switch maybe needs to be turned on for more than one voltage level. Consider the simple T-type inverter shown in Fig. below. Switch S1 needs to be turned on to produce levels +2,+1. I tried to use ePWM modules to produce the gate signal for this switch, but the thing is it is not allowed to do logical operations (and, or, xor, ....) on GPIOs. So, I produce the gate signal and then assign it to the GPIO.

I tried DELAY_US() but it just produces a delay, and I cannot produce dead-time intervals with it.

All the best

Amir

Dear Cody,

Thanks for your reply,

So, you mean we cannot use TMS320F28335 for multilevel inverters applications by using ePWM modules? I tried different ways but at the end of the day, I have to use my proposed procedure to produce gate signals.

Thanks for your precious time

Amir

Amir,

no not at all. I have seen customers implement multilevel inverters on this device. While I cannot comment on the exact methodologies they use, it would either be done through code controlling the PWMs or explicit external logic(And, Or , etc.). The code could be interrupt or state machine based, but would use the PWM modules. I don't know of any one who has successfully used GPIOs to emulate as many PWMs as you are trying to do. 

Regards,
Cody 

Dear Cody,

Thanks for your reply and time you spent to help me,

Well, I guess I should try it with different methods! It is getting very annoying for me!

All the best