>10 to 50 A igbts

  You can only use those if they can sustain switching at over 20 kHz. For
extra large units 5 kHz or even 400 Hz will do (but you don't want to deal
with 100 kW range units).

>220 V DC

  To make 220 V AC out of this, you need to raise it to 315 V dc at least
(i.e. disregarding losses, more like 350 to allow for some headroom when
switching and losses in diodes and chokes). You can't apply 220 DC to the
switching bridge and get 220 V AC from it because you can't reach the
required peak voltage like this. And you need more like 6 'H' switches to
drive such a thing (minimum 4).

>3 phase

  Concentrate on 1 phase, then add more as you go. There is no need to
complicate yourself from now. The electronic generation of 3 phases is so
much like 1 phase that the only thing really required on the controller
side is extra ROM to store the larger code table...

  imho, you will need to do some real good figuring on the output filter
that will make the high frequency switching waveform into 50 Hz sine. An
iron transformer (3 separate primaries, one 3~ secondary star/triangle)
will save the day if you can afford the weight. This also obviates the
need for a voltage raiser from 220 -> 315 V DC.

  Note that if you do NOT use a transformer, then you will have a
triangle-type output and will not be able to ground the center point (and
no star center in any load, which most motor and heating controllers do
not like at all, being compelled to use 380 V 3~ only).  The center of the
transformed 'triangle' will float at 158 V DC above the ground (i.e. - of
220 V DC).

  Looks like a serious project to me... probably not worth the effort for
one off. BTW I think that I heard of rotary converters (ancient). They are
made of a special motor running at 220 V (or whatever) and extra windings
on it that give the 3-phase for the load. Maybe you can obtain one of
these 2nd hand from where the tug comes...

Peter (plp@nospam.actcom.co.il)