For high power systems, I prefer to put an intermediate dummy load with
a coaxial patch panel to select load or driving the final stage. This is
done as it is time consuming to have to break loose a 3 inch or 6 inch
coaxial line to then insert a water cooled load. By selecting the
permanent mounted load, I can quickly switch over and test the driver
and tune it up. Its a big dummy load, like a 50 kW Bird or a 200 kW
Altronic Research water cooled resistor. Every stage is tuned up to 50
ohms, so that when combined, the various directional couplers indicate
forward and reflected power and you know which knobs to adjust to get a
match similar to the dummy load operation.
As discussed by Donald Fox, Joe Subich, and Kim Elmore, the drive power
has to go somewhere. It goes into a dummy load if you are lucky to have
that as I discussed, or it can also be driven into the tube of the next
stage, if the match is good. But one has to be careful of grid current
excesses. One system I worked with had a water cooled grid (the RCA 7835
grounded grid triode) so you could send as much there as you wanted
without concern.
The amplifier input and output tuning controls are very similar to ham
amplifiers, in a well-designed high power system. Sometimes the
loading/coupling control is left out, assuming a constant regular
impedance (not an antenna) but then you are stuck with optimized
operation at only one power point. Sometimes it is a fixed impedance
transformer for input matching.
For cathode-driven (grounded grid/screen grid) operation with tetrodes,
the input impedance of an amplifier is highly dependent on the cathode
current in the tube. This means the beam must be on, the tube drawing
idling plate current at least. So various logic is interlocked with the
coaxial patch panel, so as to allow the pulsing of the grid voltage on
the final stage, so that it is idling in a pulsed way. Quiescent current
in the TH628L final tetrode is about 5-10 amps for class AB2 or B. With
23 Kv DC of plate voltage that is between 100 and 200 kW of plate
dissipation alone, so one can see why pulsed is much easier for testing.
During normal operation, the bias is pulsed on for the driver and final,
as well as the solid state predriver (5-20 kW stage). After a few
microseconds, RF is ramped up to full power and stays there for a
millisecond. This is a typical pulse for a particle accelerator. Then
shut back off again, to wait for 7.3 milliseconds until the next pulse.
For continuous operation such as shortwave or the ancient WLW 500 kW, it
is all dissipating a lot of power so you have to be extremely careful
when tuning. Also watching for parasitics which can be devastating. In
the days before having continuous oscilloscopes, it must have been
challenging, watching the meters for sudden changes.
As for using network analyzers, transmitters were developed using the
estimated impedances at the plate of the tube, at the grid, just as is
done now. But there was no easy ways to measure it. Grid dip meters
crudely showed the frequency of various resonances. Neon bulbs showed
where the RF voltage was rising in a standing wave. RF detectors with
diodes work. It took a bit of work, but could be done. We take VNA's for
granted now, but using them in a high power tube circuit is still not
trivial as they are 50 ohm instruments, and impedances may go up into
the K ohm.
73
John K5PRO
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|