That’s exactly the idea behind kWeld. It calculates the instantaneous power that it delivers into the weld spot (not to confuse with the power that it draws) and accumulates it during the pulse, in order to determine when to stop. Keeping the amount of energy/heat dumped into the weld spot constant makes the results more consistent, and mostly independent of mechanical pressure, dirt/corrosion, and the battery voltage. I’ve also recently added an arc detection feature that interrupts a pulse with an error in that case. Arcing can easily puncture a cell can otherwise, when it happens. It’s resistance spot welding, not arc welding in our case 
I’ve actually once used kWeld as a magnetizer, by mounting an air coil instead to its output. It doesn’t make coins jump however, you apparently need more voltage for that.
Good point And, yes and no. As long as the mass that you deposit the energy into is kept constant, then you’ll get a constant temperature rise. As we are heating up a small spot underneat the electrodes, this is a good approximation when we can keep the pulse duration short. Once that gets longer, then heat starts to spread out form the spot to nearby metal. I could implement an algorithm that estimates this heat spreading, but that would need to be configured for parameters like metal thickness and type. As long as everybody reports me that the repeatability that kWeld produces is good then I don’t think that it’s worth the effort.
At the beginning of our conversation, I was convinced that he overtorqued the current trim pot as these are quite delicate, and didn’t see this as a warranty case. What had worried me much more is that the used Maxwell caps are not intrinsically safe when being overcharged and can apparently go wild in such a case. Therefore I immediately started developing a protection circuit that will be added to a next gen kCap module. It doesn’t matter whether the overvoltage is resulted from a user error, a bad trimpot, or some kind of fault, kCap has to be safe to use. I have the first prototype of this on my table right now. This got me distracted from my communication with this guy, and although I wrote him that I would treat the kSupply fault as a warranty case despite not being entirely convinced, I forgot about the bad kCap that he also has on his table. I just checked my email history, he didn’t ask any further and also didn’t send the kSupply for repair yet. I’ll contact him again to resolve this.