Dear Sir/Madam,
We are conducting research on power converters, controlled by DSPs for the last 7 years. This is what we notice and we need your help. We produce pulses for our opto-drivers using an eZdsp f28335. When we do that using the ePWM there are spikes every 1,5s (15000 samples with 0.0001s sample time) in the opto-driver output (figure 1). In figure 2, you can see the magnification of the low frequency pulses of the opto-driver. When we disconnect the cable from the ePWM output and reconnect to a GPIO and produce identical pulses there, there is absolutely no problem (figure 3). Remember, everything else, hardware-wise, remains the same and only the software loaded on the dsp changes. No spikes whatsoever when connected to the GPIO (figure 4). We have tried the same setup with different f28335 DSPs, noticing exactly the same spikes with the ePWM, so there is absolutely no defect on the specific DSP. As you know we cannot produce high frequency PWM with the GPIO, so we have to find out what is wrong with the PWM. We also attach the simple mdl file that we use to produce the above results, eventhough we do believe that software is irrelevant, except if there is some hidden property in the board configuration settings.
We have noticed exactly the same spikes with different chopping frequencies, different clock dividers etc. Always spikes happen at 1.5s when ePWM is used. No issues with GPIO. Do you have any idea why is this happening? Any help would be highly appreciated, as we are at complete standstill for the last 1 year.
Thank you,
Panagis Vovos
Lecturer
El. Engineering dept.
University of Patras, Greece