AG6K spoke of the primary emission from screen grid of 3CX800A7 due to
overheated cathode, with subsequent barium deposition on the gold wire.
This certainly is a concern if, as he said, the filament voltage is grossly
high in a particular design. We should try for 2-3% of nominal, preferrably
on the low side, for long life.
We see this problem sometimes in our medium power 200 MHz tetrodes, that
supply 150 KW to the finals. Since it's a pulsed RF plant, it manifests
itself as RF power that changes across the 1 millisecond pulse. It is seen
on the screen current, which, in a pulsed system can be seen on a current
transformer with an oscilloscope. The screen current will actually change
direction as the pulse gets longer, from heating. It is dependent on RF
power and length of pulse and repetition rate.
We have seen it due to manufacturers contamination of the screen.
Interestingly, the cure seemed to be to run the thing with the filament
hot, and try and burn it off! (that's what we were told to do by the manuf).
It has worked. Over 48 hours or so, the primary electron emission from
screen to plate seems to diminish. I don't pretend to understand how this
In another of our tubes, a Machlet ML-LPT-44 (discontinued), we get grid
emission also just with age, as the filament evaporates. When it happens,
these triodes become harder to control. We use them as pulsed switches to
put voltage on the mod. anode of 1 MW klystrons at 805 MHz. The LPT-44 is
just a rep rate switch. When they have grid emission, they tend to stick
on, and leak current during the cutoff time. This creates major Xray
hazards, as well as dissipation in the system.
We have a thing called a bombarder, that we put these tubes back in when
this happens. In it, we clean off the grid contaminants by running a lot of
positive grid current (and grid dissipation) without big anode currents.
After so many hours, the tubes are clean and grid emission is back down.
FAQ on WWW: http://www.contesting.com/ampfaq.html
Administrative requests: amps-REQUEST@contesting.com