The first weld controller that I have built had no synchronisation and therefore the weld time was not very accurate (+/-20ms) and the minimum weld time was set to 50ms because of reproducibility. For me this was fine. But with the synchronisation, the weld time is more accurate and a shorter weld time is possible, but I have not tested this. The weld times can be changed in the software easily.
Maybe with a shorter welding time there will be no need for switching to the lower 400A current mode.
I think that very short weld times are tricky, in some circumstances it may work, it should further be examined.
OK, that makes sense, thanks.
Some capacitor discharge spot welders can use dual pulses where the first is a short pulse in the order of 10 ms and the second (after, say, a 10 ms pause) is a bit longer. It was in this context that I had asked the initial question.
As you say, this will require some investigation and testing. In theory the current from the capacitor discharge will be a decreasing exponential while the one from the transformer should be sinusoidal. These different waveforms will probably produce different welds.