Hi Kai,
Thank you very much for your response.

Assume for instance you have a short voltage dip on mains. If now the PID loop shows some relevant differential term, then a much higher power will be applied on next spot welding event. But what if then mains voltage dip has dissapeared?

The resistance welding is usually 2-20 AC cycles. If there was a voltage dip in previous cycle yielding less heat generation and if it is restored then I will be more heat generation which can compensate for previous cycle. And if voltage is not restored I will be getting required heat anyways. [ I will need to do some more experimental trials on this issue using PID to see the effects In my current design I am using voltage feedback to detect voltage fluctuations and no PID I will continue using this voltage feedback in next PID thing as well to take care of short
fluctuations ].

What if there's suddenly a voltage spike? Then, welding spot would be extremely overrated and burn down

Since resistance welding lasts for 2-20 cycles a short spike is usually not a problem. The thyristors used for supplying welding transformers contains MOV which also helps in protecting against such spikes.

I think a better approach is to design a constant current source, means to control firing angle every half period in order to keep welding current through welding spot constantly, and this for the duration of actual spot welding event.

That's what I am designing a constant current source. which would deliver a constant set current irrespective of fluctuations sheet metal variations etc. I want my PID loop to calcualte new firing angle for every half cycle to ensure set current [ RMS ] through job. For all 2-20 cycles.

Thanks & Regards,
Prahlad Purohit