Anybody ever use 250mcm or 350mcm or 500mcm cable?

One big "but" about using parallel cables for handling large currents...

If you have one connection/cable fail, then the rest of the current flows in the remaining conductors.

And another caveat that I have experienced multiple times in high end test equipment...

Many times, the test equipment sent (for example) +5 volts on 4-6 parallel wires (and another set for ground/return current). plus other signal cables using one large 24 conductor (for example) mating pairs of connectors.

What I saw happen was not a somewhat higher resistance current path (say a connector pin+socket connection) that failed.... But the "perfect" pin+socket+crimps+wiring run that had the lowest resistance... The lowest current path would take the most current (I=V/R) and would heat up the most (P=I*2R). Notice that heating goes with the square of the current... And I would pop open a piece of test gear and I would see the connectors "unzipping" from overheating pin+socket connection. First one at the end, then it got hot and caused the next one to fail, which got the next pin+socket hot then in failed... Until the last cables (good connections) had to fail because they were too small to carry the whole current by themselves.

What should have been done? First, the loads should not have been paralleled, but divided into sections (each less than the rated capacity of the cable+connector). Sometimes that cannot be done--Next, would have suggested longer wire run/smaller diameter cable--You want the resistance of the cable to be the "major" ballasting factor (i.e., if the connector went a little bit low resistance than the cable, it would not matter since the resistance of the cable is >> than connector resistance, or lack of resistance).

And/or more parallel connections... I used 1/1 = rated current for one cable. For two parallel cables, I used 1/1 + 1/2 = 1.5x single conductor rated current. For three, 1/1+1/2+1/3 and etc. for more parallel connections. I never did have any failures in my designs (that I know of from following that rule--And as you see, 1/N gets very small for lots of parallel paths, and why I suggest more than 3 parallel battery strings is not recommended by me--if at all avoidable).

Using heavier/higher rated contacts+cables is obviously another answer--But not always possible (costs, space, dealing with >0.5" diameter copper cable, etc.).

Remember that many times we are paralleling cable to lower voltage drop--And a single cable is "capable" of carrying the current without failure--But would have "excessive" voltage drop--Another issue.

-Bill