My understanding is, that a system that doesn’t oscillate under no_load conditions will not oscillate when put under load.
It could still happen when you tune PID with a certain load and in the field this load is suddenly lower. (in case of a fan it can be a wind gust)
It would be cool to trigger a re-tune algorithm when oscillations happen. But that requires an autotune procedure and a way to detect oscillations. Or detect oscillations and fall back to save parameters…
Until then, we have to use moderate PID values as a tradeoff.
Yeah I think you are right, if you increase the mass of the system that would make it less prone to oscillating, not more. But suppose you design a system where like my fan there is a load on the motor. You assume it’s always going to be there. Then one day something changes. Or the voltage of the power supply changes, that could also de facto increase the P gain, I think.
To be fair this doesn’t seem to happen very often. Still I have run into issues a number of times where a system includes a PID and I can’t tune it and so it interferes with the flexible use of the system. It goes against my belief in the importance of modularity and reusable flexible general purpose systems at little bit. But whatever. I just think autotune should be worked out a little better, we are almost there. It’s like automatic gain control for a microphone. It’s just a helpful feature that would probably become very widely used after being sorted out.
