Yes, very interesting, although the important thing it seems is to
saturate the transformer core itself. Once saturated, the conducting
transistor cuts off and the resulting collapse of the field turns the
other one on. Does the transistor necessarily have to saturate for that
to happen? As you said, you can't model it not knowing the
characteristics of the transformer. The DC resistance of the winding
have a current limiting effect? Probably.  There seems to be more than
enuff drive from the feedback winding despite the two Si junctions. Two
things had to happen before I could get the 2N5684's to run. First, I
had to open up the holes in the TO-3 socket to 1/16" for the emitter and
base leads to fit through (fat), and second, when I powered it up for
the first time one of the MPS-U57's decided to commit suicide (not enuff
"moxie"?). I installed some 2SB1274's (because that's what I had) and
replaced the fuse and all is good. Runs like the Ge's with 550V on TX.
See attatched jpeg's. Thanx for all your help, Jim. I really appreciate
it. 
 
   Michael   KF7IYP