It is tempting to make a spot welder from the Maxwell ultracapacitors e.g. the 3000F ones rated for 1900 Amps each. There is something nice about the idea of a spot welder that does not draw large pulses of current from the mains, and a cordless spot welder would be a novelty.
I figure that something in the region of 10kA to 30kA output current would make it useful enough to justify the effort. I understand that the welding voltage is normally a couple of volts or so, but if the weld is in an awkward location then cables or tongs might drop more voltage than the weld itself.
If you work out the energy required for a spot weld, it is far less than the energy contained in one of those capacitors, let alone several in parallel. Therefore it is going to be necessary to switch off the current somehow.
Once one is doing that, it becomes tempting to regulate the current with a bunch of switching current regulators or buck converters, perhaps with multiple phases interleaved.
It seems like most cheap FET packages are limited to about 100 Amps, so a bunch of 100 Amp current regulators with the outputs in parallel might be good.
What is the cheapest FET per current? The voltages would be small, even a series-parallel capacitor bank might be charged to 10 Volts at most. Since there might be a hundred of these mosfets to reach a total of e.g.
10kA, a small package would help. It seems to me that for any FETs that have a high enough current rating, the RDSon will be far better than required.
It only needs to withstand the current for a few hundred milliseconds, maybe every 5 seconds.
Also, any ideas for making cheap high current inductors? Perhaps the cables to the welding electrodes would have enough inductance for a single switching current regulator or buck converter, though if instead of one large buck converter, many small (100A) converters are interleaved, then separare insulated strands would be needed for each converter, and the mutual inductance might mess things up.