At the moment the main problem is figuring how I will get 10e5-10e6 currents for 1ms or greater.
Options:
0) Really huge lithium battery bank. This would appear doable for a few thousand based on the ACIR 13.2 mohm and max charge voltage of cells like the Samsung 25R. Used Tesla modules appear to be initially less expensive but may have inferior resistive losses at pulse current due to the fused interconnects. Not sure about practicality of either option. Over 1000 high discharge 18650s would be required in parallel. This option would almost certainly preclude a plasma armature due to the inherently low voltage without an impractically large bank size.
0.5) Supercaps? I don't know anything about supercaps.
1) Homopolar generator. I would have to design and build a really big spinning wheel and brushes that carry the foregoing amps which is not trivial. Or buy a used one from a physics lab.
2) Inductor. How do I charge a really big inductor? Probably from one of the previous two options, so it might be more practical to just use one of them. Any suggestions are welcome.
3) Pulse alternator. This option looks impractical from an engineering standpoint since it is even more complicated than option 1. On the other hand, compensated pulse alternators or compulsators are apparently a very attractive means of compact pulse power supply and also entirely unknown outside of research.
4) Cap bank. Already ruled impractically expensive, unless I can find a physics lab throwing out their old particle accelerator.
\There are quite a few "cheap" SCRs on Ebay that are sold as NOS because the seller can't find the data sheet.
The whole point of pneumatic injection was to avoid the problem of putting in a switch somewhere else. But it looks like pneumatic injection may be undesirable anyway.
I've been looking for SCRs on ebay and it's not immediately clear which ones will do ludicrous current. They all look kind of small. At least I would have to parallel a handful of SCRs, which is fine.