My project right now is a push-pull amplifier, to deliver 10 Kv peak (not
Peak to peak) on each side, 180 deg phased. At 2 to 6 Mhz. Combined into a
transformer or combiner, it will deliver 20 Kv Peak.
I am using two tubes and push pull, instead of a single tetrode and an
output transformer
to go single ended to balanced. Reason - the load is a pair of triodes,
push-pull. The additional cost of another tube and socket and all the power
supplies is high, but I can get two tubes that each deliver the desired
plate current peaks easier than a single bottle. The bigger the tubes get,
the more prone they are to internal resonances due to higher frequency
cylindrical mode conditions in the envelope (structure).*
A real plus to using Push Pull with an output transformer, is the DC feed
for plate voltage can come into the center tap of the transformer, at the
node. RF choking is greatly simplified (Series fed DC). I don't have that
luxury, as each side of my amplifier looks like an individual single ended
tube. The combining is done in the load of the final stage. So I've built
two identical plate chokes, to shunt feed the DC to each tube.
With push-pull, the tube output capacitance is not doubled (to the tank)
when the power is doubled with two tubes. This is a great way to achieve
higher power without sacrificing bandwidth (R and C). Paralleled operation
will double the capacitance. I believe that in the early days of FM
broadcast transmitters, the output stages were push pull, to get more power
without hitting bandwidth limitations. They used glass tubes then. Only
through the advent of larger ceramic triodes and tetrodes have the
single-ended circuits come around for these rigs.
And the aforementioned even harmonic suppression. A good Pi or Pi-L, of
course, can achieve such suppression, as well as the higher harmonics from
third on up. These days most folks stick to single ended and Pi networks.
Dealing with the symmetric construction on 6 meters can be more of a
headache, but a very beautiful layout if carried out well. Single-ended is
much simplier in this case.
Building the output transformer for the push-pull network wasn't trivial,
so i am using a pair of Pi networks, Q = 5-10. They are identical, and have
linked tuning shafts on the caps. I would have loved to use an H, which is
really a balanced Pi. The vacuum caps would unfortunately have had high RF
voltage on the outer housing and shaft. One of the important things to try
and do in push pull is make everything mirror image, symmetry. Otherwise
you have some extra L or shunt C on one side, which can make things tune
oddly and is hard to build physically. Somewhere hams came to the same
conclusions, as seen in Handbook amplifiers all the way back to before
WWII, in which the tuning capacitors were such beautiful symmetrical
things.
Using two Pi's in the output, I have 4 caps, but the cases are grounded
(also good for cooling purpose). My input tuner is a true balanced H. It
has only two caps and two inductors. Each side looks like a Pi, but the
caps are floating in insulators, and tune both grids simultaneously. It
works great. The solid state driver sees 50 ohms.
*The Eimac 8974 (also X2159) megawatt tetrode is a prime example. It is a
BIG tube, physically. And it needs to be tamed in a well engineered
circuit, with lots of bypassing, and maybe even lossy mode suppressors
around the screen connector. We have a machine at Los Alamos using these
things at 55 Mhz, and they are extremely touchy in grid driven service at
that high freq. But it works.
I saw a couple of these tubes in a surplus store in crates. I'll bet you
could buy them for $500 each if you work on the owner. Would make a dandy
HF afterburner, one for 1,000,000 Watts. You need a crane or forklift to
change the tube, however.
John
K5PRO
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|