It is a transformer, which to the output of the H-Bridge looks like a
capacitive load. there is a very large turns ratio in the transformer, and
the few picofarads on the secondary of the transformer end up looking like
tens of microfarads at the input. before we slowed down the h-bridge we
were pulling ~40A inrush over a couple hundred nanoseconds.

I managed to 'solve' the problem today, although it left me with other
questions. On the prototype, we used through-hole parts, and surface mount
parts on the PCB. it turns out that there are some differences between the
surface mount and through hole versions of the same diode even though all
the data sheets specs are identical.

As i was pulling my hair out today, (looking over the layout and dreading
having to start to cut traces and run wires) i decided to try putting the
through hole diodes inplace of the surface mount diodes and it works like
a charm now..

I am not quite sure what the differences between the two packages could
be. some sort of parasitics?? In all the reading I did over the past few
days, everything pointed to reducing the stray inductances and
capacitances around the fets, which is why I am guessing that maybe the
surface mount diodes have larger stray capacitances?