OK.  The first attempt at a Class E PA using all junk box bits has been a bit more sucessful as reasonably expected, and has flagged up exactly what several people did mention on this reflector.    Using a the somewhat old and vernerable IRF462 with its Rdson of 0.4 ohms, even with two in parallel, the efficiency achieved never exceeded 84%, and looking at the scope trace the voltage drop across them at peak was in the region of 7 - 9V.   So its clear where the power was being lost.   

 

All the caps were made up from parallel combinations of the 3.3nF things I had so many of - and knowing the right scope trace shape to tune for made setting up straightforward.   The lower efficiency meant the 500 Watts aim couldn't be achieved,  but at a 50V rail I did get a smidgen more than 410W, and felt this is about the limit I really ought to go for with this sub-optimum choice of device

 

The additional losses in the FETs made tuning flatter than expected, and there were several combinations of Caps (all with the same fixed L) that gave a similar overall efficiency once the trace had been optimised, Just max power was different.

 

The total cost in real money spent on components used in the final breadboard was absolutely zero, everything came from the junk box.  (Although the 3.3nF capactors were all originally bought new back in 1997 when I was resonanting a 3m diameter two turn  loop of water pipe on 73kHz)

 

I'll do a write up and post on my web site for anyone who wants to build similar PAs..  There are a couple of minor gotchas its worth looking out for - not least burnt fingers on an inadequate DC feed torroid.

 

Now I know the desgn is so easy to get going, may spend a bit of money on some more modern devices and aim for a better efficiency. 

 

Andy