[Amps] Input VSWR of common grid amplifiers

[John Lyles]

In 2009-2012 I completed the design and testing of a very large tetrode 
amplifier that ran in grounded grid configuration. Grids not tied 
directly to ground, but grounded for RF via cavity circuits that have 
built in blocking capacitors to allow normal bias voltages. Like their 
amateur siblings ("tubes with handles" as R. Measures used to say), the 
input match was highly dependent on the electron beam, specifically the 
level of DC cathode current. I have this cascade of amplifiers, one 150 
kW tetrode driving the final stage of several MW. Being pulsed, the grid 
bias is switched from cutoff to conduction just before RF drive is 
pulsed on, to avoid 100 kW of pulsed plate dissipation in the final 
(about 12 kW average power). It took a while before we figured out best 
way to protect things, as if the final power supply tripped offline, the 
IPA would be driving into a horrible mismatch. So fast electronics are 
involved, in FPGA logic, that shuts off the RF drive and so forth. Also, 
timing of the conduction bias pulses had to be right, so that the RF 
drive didn't come up too many microseconds before the beam was there. 
Sweeping the input match of these amplifiers while they are pulsing is 
not easy. I used a Hp8753 VNA, at 10 mW power, and sweeped across the 
bandpass of the amplifier cavity. As Bias is cranked up, and a few 
amperes of quiescent (but pulsed) plate current flows, the input return 
loss goes from < 1 dB to 10 dB or more and a definite dip occurs. It is 
a chopped waveform due to pulsing, as I don't have the ability to run 
these things in CW key down. I bring this up as it parallels what we 
hams have to deal with using GG amplifiers. Not having cavity circuits 
for HF, we use L and C and try and get some Q, without excessive 
narrowbanding the thing - broaden the input tuning at least a ham bands 
width.

You drive the thing with the normal level, and get the current up to 
measure the match. Lots of compromises can be made, and those amplifiers 
that require a length of cable, have input matches that were fine for 
tube exciters but these days not so good for solid state drivers. Much 
of the problem has been alluded to here, the cyclic fluctuation of the 
beam from cutoff to conduction across each cycle of RF voltage, as 
others have suggested.

For HF ham stuff I have the trick now. I use a Delta Electronics OIB-2 
operating impedance bridge. Same thing that broadcaster engineers use to 
measure their antenna Z under power, but the HF version. This thing can 
take 1 kW through power, and will read the input Z of an antenna, line 
or amplifier, while driven. It reads the R and X component to 5%, and 
from this some suitable matching can be developed to get it to 50 ohms. 
I now have two of them, so one will be made available FS.

73
John
K5PRO