dbuckley said:
On pure single phase the current waveform is still a pulse, but it's identical on the hot and the individual phase's neutral. [...] If you make it 120/240 I think you are removing the benefit of not sharing neutrals.
I can actually answer this point for _single_ phase systems. I don't really know what happens when you have a three phase bank of single phase systems, though I do have a strong guess (above).
As you mention, any complex periodic waveform can be broken down using harmonic analysis. The nifty thing is that we can then analyze a system for its response to each individual harmonic, and add up the results to get the response of that system to the complext waveform.
Well, it is _not_ true that all harmonics present on the load waveform will add up on the neutral.
Instead only those 'multiple of 3' load harmonics will add up on the neutral. Everything else will balance out. This means that if a particular power electronic load happened to produce lots and lots of 5th or 7th harmonic, this would not cause any neutral loading problem. It is only the multiple of 3 harmonics, in particular the 3rd harmonic itself, that cause problems.
In a single phase center tapped system, the harmonics which would cause problems are multiples of _2_. 3rd harmonic does not cause a problem for the neutral in a single phase system. The characteristics of most power electronic loads are such that they tend to produce very little 2nd harmonics. So in a single phase system, you could have multiwire branch circuits (2 circuits sharing a neutral) without having harmonic problems on this neutral or on the secondary of the transformer.
This would also be true for the six phase 'star' system that I described, although it would have extensive practical problems.
-Jon