Hi Rik that is an interesting article. I haven't had the chance to
compare the formulae yet, but the article I was offering around was a
slightly later version of Ref 8. I think the parameters in the network may
have been optimised a little better by the time the later article was
written.

It is probably unfair to compare the output after all the filters when you
are expecting  to loose 15% of the power in the matching and filter network.
The advantage I would see is that the dissipation in the fets would be
minimised. I also think that Dave's design seems to be particularly good, as
from the comments that are made by people who have constructed copies, it
does not seem to suffer from a lot of the instability problems that can
occur with push-pull class D systems. A 'selling point' made by Sokal is
that the class E circuit is virtually bombproof, even if you change
transistors, it is very insensitive to device parameters, because the active
device is only being used as a switch. He also makes the point on simplicity
that it is an inherently single ended design, I suppose this has trade-offs
in the level of harmonics.

I will be fascinated to see some really high power design using this
technique. I believe the circuit on page 4.5 of the LF Sourcebook attributed
to Roger ZL2RX is a Class E design, but there are not enough on the
component values to check against Sokal's equations. Some of the comeonents
look as if they were 'to hand' rather than being calculated values. It has a
novel point that the choke, series inductor, and step-up transformer are
wound on the same ferrite rod core. On the other hand, bearing in mind what
has been said about the losses in some ferrite, that may be where some of
the power has gone.
All good stuff, get the blow-torches going chaps
73 de Alan G3NYK
[email protected]