Hi Wynand,
> I know I have to look at the capacitance of the four valves in parallel as
> this will impact on efficiency especially at 10 meters. I AM JUST NOT
> SURE WHAT I AM SUPPOSED TO BE LOOKING AT.
Remember the input system can also greatly affect efficiency. As a
general rule, the tank Q can change from anything from just over
the square root of the ratio of anode to load impedance, to very
high Q's, with little change in efficiency. The efficiency actually is
maximum with the lowest possible loaded Q, and gradually
decreases.
Say the tube combination requires a load impedance of 2000
ohms, and the load is 50 ohms. The maximum efficiency would
occur with a tank Q of 2000/40=50....then take the square root of
that number (7) and just add a tiny bit to it to be sure the network
behaves like a true pi-network (phase shift more than 90 degrees in
the network).
With decent tank components, even Q's of 40 or more produce
excellent efficiency, so you have (with good components) a wide
range of latitude. Many VHF and UHF amplifiers give excellent
efficiency with tank Q's in the hundreds, because the components
are good.
An error in input system design can have monumental effects on
efficiency, if the input circuit at the cathode-driven tube presents a
high impedance at harmonics of the operating frequency. This
allows the cathode to "move around" at the harmonics, much like
you are driving the tube with off frequency signals. It doesn't matter
of those harmonics are created in the tubes grid-cathode and
anode-cathode currents, or if they are from a harmonic-rich exciter.
The results are the same. You have harmonic energy driving the
tube, and that can greatly reduce efficiency.
So by all means, besides watching the layout as suggested, watch
the INPUT circuit. Make sure it is near the cathode, the leads are
short, and it is a LOW-PASS type of network that provides a LOW
shunt impedance at the cathode for all harmonics up to perhaps
the 5th harmonic of the maximum operating frequency.
Never use a L-C-L type of input, or a C-L-C "T" type. Use a parallel
tuned network, or a C-L-C network in pi configuration, and keep it
as close to the tube as possible.
The Q of the network on the TUBE side of the network should be at
least two...or the impedance at the cathode will not be low enough
at low harmonics. The OVERALL Q of the input network is
meaningless, since the series L divorces the exciter half of the
network from the tube at harmonics. What you want is a Q of two
or more at the output capacitor to tube junction.
That generally places the overall network Q at more than 3 or four,
but it is advisable to always have a look at R/Xc at the tube side to
be sure it is at least 2 or more. For a 50 ohm tube, you'd need 25
ohms or less Xc on the tube side.
> I also now want to add a tuned input circuit as previously I just put a
> ATU beween the rig and amp input.
Bingo. That's a very poor idea, and can kill the efficiency under the
right cable lengths and network impedance no matter what exciter
SWR is.
73, Tom W8JI
w8ji@contesting.com
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|