It is hard to say exactly what is the most important thing in Digital Control of Power Electronics. There are lots of things to know.
Some that are important are
- How sampling works on a frequency domain basis, mainly because control stability is most easily viewed in the frequency domain
- The inherent processing delay of digital systems and its impact on how you need to design the power converter
- Calculating the frequency response of a z domain transfer function.
- Implementing a z domain transfer function as a repeated, sequential, multiply and accumulate discrete calculation
- Understanding the effect of the quantization on the response of any digital filter or system.
- How to deal with the effect of this quantization.
And then there are other fundamental software type things
- The processing for the control loop must be completed before the control output is required. And if there is a delay it must be consistent. I have a neat example of exciting a mechanical resonance by changing the string length on a pendulum.
- The processing system can run at the same speed as the switching, a multiple of the switching or any other rate so long as the beat frequency between them is not annoyingly in the control band.
But the most fundamental thing to know is how to deal with the non-linearity that digital systems have. This non-linearity is noticeable in the small signal when the digital value is meant to be constant. The smallest step is a one bit change.
This is a staircase non-linearity and as digital control is usually about the error being small (or zero) the quantization is the cause of all sorts of small signal instability. “I am pretty sure I don’t have that” you say. What does that look like? – it looks like a slightly periodic noise.
So the most important thing is that digital control is non-linear. And as power electronics is more often than not non-linear we now have two nonlinear systems connected.