Several points. I've not figured out this entire issue, but several things jump out at me:

1) The _primary_ current of a 5KVA 600V to 120V transformer is only about 8A. The load current will not be flowing through the primary coil, and I would expect small conductors on the primary side. The _secondary_ current of one of these transformers should be about 42A, and I'd expect larger conductors.

2) While you did calculations for an autotransformer, the transformers that you described using are standard dual coil transformers which you will be connecting in a buck configuration, requiring slightly higher KVA ratings. The _full_ load current must flow through the 120V coil, which means that the transformer KVA must be 120V * full load current.

3) I don't see how you are connecting _three_ 600V to 120V transformers in a buck configuration. You can use an 'open delta' connection with _two_ 600V to 120V transformers (with suitable ratings), and I can see symmetric connections that would work with three transformers that have _dual_ 40V secondaries (the secondaries would be connected in something of a 'zig-zag' fashion). You could also use these three transformers with the primaries connected _wye_ to buck the voltage down, but this uses the primaries at only 347V, which reduces the effective KVA of the transformers (you are using 600V to 120V transformers as 347V to 69V transformers, so the KVA available is reduced to 57% of the nominal rating).

4) I don't know the _code_ for any of the possible connections that I am dreaming up; just scheming with the physics [Linked Image]

-Jon