So a little background to explain the need. Occasionally we need to test starter motors (for engines) and provide enough power when the load (engine) causes the starter to pull in excess of sometimes 1000 A. Granted, this is for quite large starters. The test would need to last a minute or so, but foregoing the need to alter designs, we'd like to just go ahead with a continuous demand rating.
So I've got a fair understanding of linear power supplies and some common designs of higher currents, such as multiple transformers/taps, bridges, parallel transistors (2N3055, etc.), but in my own experience I have only "engineered", mathematically, relatively simple sub-circuits, RLC, and other simple circuits (555 timers, opamp 747, etc.) using Kirchoff voltage/current, Thevenin, systems of ODE/PDEs, phasors, etc. But really I rely on the community for a suggestion on what to look for. (And ideally a schematic!)
Some of the design requirements are
- 10-40 VDC (some are 24 VDC starters and a few are 36 VDC)
- Maximum of about 1000 A at 24 VDC ~ 25000 watts/VAC
- Be a current supply more than a voltage supply, similar to the idea of having the load in series with a much larger resistor.
- Lastly (I think), is to be able to use single-phase, two-phase, or 3-phase input.
For larger loads, 110 VAC at the main is not going to work. For some idea of portability (we are a start-up business and have plans to grow and move) it would be nice to be able to run off "single" phase or three phase. This way each leg could supply a good amount of the load. (Can these be joined after the rectifier bridge or smoothing cap?)
By the way, companies such as DynaPower, though awesome, are just way too far out of our budget for right now. Besides, it's much more fun to build it yourself!
For what it's worth, we had thought of purchasing a used portable welder but were worried about the voltage adjusting itself to give the required current, and therefore power, to the load. I wouldn't know how to limit voltages at these currents...