The biggest advantage of push pull is that you'll have less trouble with the
output capacitance at the higher frequencies. Even so, 813's above about 21MHz
are getting a bit high on output capacity. 813's are rugged, relatively cheap,
and go fairly well with enough volts - I'd suggest 2500 as a minimum, and 3000
is better. They are rated at 650 watts from a pair with 2500volts in audio
service. If you're just building a single band amp, then there's no problem - or
use plug in coils, of course.
I disagree with Rich about Class AB2 and Class B operation. So do other
professionals, such as some little firm in Cedar Rapids - see Pappenfus, Bruene
and Schoenike, Single Sideband Principles and Circuits, McGraw Hill, 1964 et al.
What is however true, is that Class AB2 and Class B grounded cathode operation
requires good regulation of RF drive and bias supplies, and without that, the
IMD performance will suffer - abysmally. I'm assuming a regulated screen supply
is de rigeur. (Which is why the G2DAF amplifier has, in certain configurations
at least, such a bad press). Amateurs tend to have problems with AB2 or Class B
because they ignore these requirements. Additionally, negative RF feedback can
be of very great benefit. As Tom says, the easiest way to add stable RF
feedback in grounded cathode is a cathode series resistor. Incidentally, the
inductance in the cathode lead/resistor then needs to be minimised, or strange
things can happen to the input impedance - it can go negative, leading to
parasitics - of a 'different-to-the-usual' variety.
I have done RF Feedback successfully with the Bruene capacitive bridge
neutralisation circuit by using largish capacitors so that the output appears in
series opposing the volts across the grid transformer; however, the large
neutralising capacitor used for the feedback adds to the capacity from anode to
ground, so also limits the highest frequency of operation.
I suppose it's just Murphy's law that at VHF/UHF, where the best linearity is
needed, things like RF feedback are practically impossible (or at elast, highly
difficult!) Of course, if it's practically impossible, then there must be a
finite probability, but it may be better to avoid that argument. (Ian, diodes in
the left side?)
One trick that has been successfully used to gain a few more db of IMD
improvement on speech signals is the resistor in the cathode, with the cold end
well bypassed to RF and a small audio choke from there to ground, adding extra
feedback at the envelope frequency.
FAQ on WWW: http://www.contesting.com/FAQ/amps
Administrative requests: amps-REQUEST@contesting.com