The kweld is a good design, no doubt about that. Unfortunately its global distribution model is not nearly as good. And by the time I can get one, its just not worth it. And its an interesting thing to build somethign new.
So, I'm going to create an 'lweld'. ( l comes after k. ). The concept of a constant energy delivery is very solid. Simple timing is not enough to get reliable operation. The question of how you deliver constant energy is key to everything.
The Kweld is measuring voltage drop, which provides a measure of currnet. Mathematically, you can then derive energy, as Power x time.. Obviously the Microprocessor and A/D needs to be doing this pretty quickly, and making a LOT of measurements, in order to get resonable accuracy.
The actual amount of energy being delivered is not really important.. We just want to be able to deliver the 'same' amount.. so, you can do some experiemental welds, and adjust up and down till dial in on the right weld.
I dont' want to use batterys for my power supply, I'm intending on use super caps for the job.. Super caps have some interesting charactertics about them, but something we can exploit is the voltage drop between the start of the weld and the desired end of the weld. This is a job for a comparator.. with a variable set point. It requires that the SuperCaps are recharged back to the same voltage between charges. There are jobs that sometimes where an analog solution is a good fit.. I may have a microprocessor in the mix. but it will be more or less in a supervisory role, and not direclty controlling the weld switch off.
HOwever its not the