On that note, interesting to ponder methods to deliver heat. Is contact reliable enough to deliver a known peak temp on a fixed cycle? Works fine for sheet metal (spot welding), but that's with a modest to large ratio of resistivity (modest as in for aluminum, which works well enough; steels are quite straightforward). We don't have anything substantially better than copper to clamp the joint with; and also the heat loss through the cable and clamps is substantial, but equally poorly defined.
Is the ratio of electrical to thermal resistances reasonably constant? Hmm, maybe. Minus the default losses of course (conduction up the cable is independent of clamping).
We could use a constant-current source (a few hundred to some thousands of amperes, depending on wire/ferrule size), which accounts for the spread in electrical resistivity, but not thermal; it would likely deliver too much power into a high resistance. Constant voltage would draw less power with poor contact, but perhaps the reduced heat dissipation in the same case, balances properly? It seems like somewhere between these extremes should work, and that's quite convenient as it means a fixed impedance source should suffice; neat, shouldn't need any exotic controls. And time and applied voltage/current/power can be more-or-less inverse, should be very simple.
I guess the best case would be, say of a crimp lug: clamp it such that the barrel is under axial compression, between two massive clamps, one holding the wire, the other the lug face. This minimizes both electrical and thermal resistance outside of the barrel section, while leaving it as a point of least cross section, thus getting the most temp rise under load. Question: should the barrel be under radial compression as well (say by ceramic swage jaws, or a tubular collar)? It will release its internal tension as soon as it reaches annealing temps (sub red-hot), and it'll squish down like putty between the clamps. Seems like it should. And then, with tight clamping, internal oxidation should be at least greatly reduced, if not forming an entirely gas-tight seal in the process.
What of filler? The crimp lug could perhaps be prepared with an excess of braze*, which then wicks into the cable material, without having to rely on platings or anything (likely, tin plated cable simply isn't enough anyways, it's a quite thin layer).
*Brazed barrel type lugs are made from a single piece of flat stock, curled around to form the barrel; in the same motion, a small sheet of braze filler is clamped in the joint. These are sent into a reducing atmosphere furnace, where the braze melts into the gap, forming a solid, strong barrel (and also annealing the lug). It would be interesting if simply "too much" were used, enabling a process like this.
I suppose that would work, without too much trouble. Simple mechanical clamps could do, with periodic adjustment to account for wear / deformation. Bonus points for pneumatics or hydraulics. Regular cleaning of the contact faces would probably be needed, that's fine.
But yeah, still has the problem of no way to restrain the wire against fatigue. You'd need a pretty stout piece of heat shrink to account for that. Heatshrink or other wrapping is mandatory anyway, given the length of insulation that must be stripped to fit the clamp(s) -- but you'll need quite some thickness to account for both the missing insulation jacket AND the annealed metal joint.
The big question is, after such a process, can the metal in contact with the clamps, be held cool enough that it doesn't anneal appreciably? Thus solving the fatigue problem. If so, that'd be very interesting indeed.
Anyway, going to such efforts, when a simple cold (crimped) joint suffices, clearly isn't going to catch on. Just interesting to think about.
Tim