- Ask a related questionWhat is a related question?A related question is a question created from another question. When the related question is created, it will be automatically linked to the original question.
Hi,
I'm working on a project very similar to the example "Control of a 3-Phase Interleaved DC/DC Converter" from the ePWM reference guide, shown below. 
I have 3 ePWM blocks, one working as the master module, the other two as slaves 120 and 240 deg phase-shifted. I would like them to output exactly the same signal, just phase-shifted.
My input to the ePWM blocks, which is the duty cycle (just one value for all three of them) obtained from the current values read in by three ADC blocks and their successive elaboration through a closed-loop control system, is varying because of the nature of my hardware. Even if this varies at the same frequency as my ePWM blocks, the two frequencies don't seem to be synchronised. As a consequence the three ePWM blocks read in the given duty cycle value at different moments (each when its own timer starts counting-->timer=zero. Alternative options are available, but none puts different ePWM blocks in relation with each other). This means, that if the duty cycle varies between the timer start of ePWM1 and that of ePWM2, the latter will output a different PWM signal.
Is there a way to synchronise all three PWM blocks to read in the duty cycle value at the same time, independently from their timers running at different phase angles?
Alternatively, I have tried triggering all 3 ADC blocks through the same ePWM block ( the master ), so that the current inputs and thus the processed output value for the duty cycle wouldn't change throught the whole ePWM1 timer period (master). That somehow didn't work, though. ((didn't make use of any interrupt option))
Thanks in advance for your help,
Andrea
Hi Cody,
Thanks for your reply. If I could make sure that the PWM function block only updates the value at a specific moment inside the counter period, I would do that. Unfortunately options for that are very limited (CTR=ZERO/PERIOD/etc.). Any change in the duty cycle value between TBCTR=0 of ePWM1 and TBCTR=0 of ePWM2 would be updated by the latter only, which would output a different signal, too.
Interestingly, I've tried feeding one and more PWM function blocks with a repeating sequence that fits exactly in a PWM period ( 5 values, each changing every fifth of a period (=100 µs because PWM frequency = 10 kHz)), as shown in picture below. 
The result was, that ePWM1 (the lower one, the master with 0 deg shift) wouldn't always read in the value 5 or 40. And I've made sure with that sequence that acquiring the value 5 once would mean having to acquire it in the next period, too, IF the PWM block was to update the value at the exact same point in the TBCTR period. Independently from whether the sequence and the PWM period were synchronised, which I would assume should be the case.
( I've set "Reload for compare B Register" = "Counter equals to zero" )
Surprisingly, ePWM1 doesn't always read the same value. It seems to be following a pattern of 5 and 40, that repeats every X periods, like 100 periods. As if the PWM period (100 µs) was slightly shorter than the sequence period and therefore shifted slowly ahead.
These results make me something is wrong with these PWM function blocks.
Regards,
Andrea
