In 1994, working with a tube rebuilder, Econco, we autopsied several
4CW250,000B tetrodes, all Eimac builds from the 1980s. These tubes
were used in our plate modulators as the series tube, a parallel pair
of them used, with 30 KV DC on the plate, and about 22 kV pulsed at
the cathode used for the output to the plate of the RF final PA tube.
This floating deck hookup is sometimes called totem pole
configuration.
Anyway, the 250K's had started to breakdown frequently, with internal
arcs, similar to gassy tubes. The screen to cathode spark gap was
breaking down often, indicating that the screen was rising to high
voltage. Also, the grid bias power supply would get killed sometime.
Upon examining the internals, (i have the photos somewhere - it was
before digital cameras were available) we found that the grid cage
was warped in the worst tubes. More closely, we found that the inside
of the plate (anode) had much cutting from electron beam. It was cut
with tiny slots, that mirrored the beam pattern through the slots in
the screen and control grid cages. Since then I have found that this
is fairly typical in high power tubes, although it is worse in some
conditions of bias and classes of operation.
We surmised that:
1) the grid warpage was not a manufacturing defect, but a result of a failure
2) E beam cutting was the cause, as it was liberating a lot of copper
material into the vacuum, and arcs were bombarding the heater and
grids.
3) The beam cutting was causing hot spots, the grid secondaries would
make things worse, eventually it was a runaway until the grid warped,
which REALLY caused a problem - with current hogging at that point.
4) the tube is not perfectly symmetric, radially speaking. It only
takes a tiny misalignment for the focused beam to be nonuniform
around the tube. It can be seen in the cutting pattern of the copper
plate.
5) inadequate cutoff bias was the reason, about -350 Volts DC from
the original power supples. Focusing of the beam during 'cutoff' is
possible, and can also generate substantial Xrays.
Looking at the tube datasheet for 800 volts on the screen (there is
no zero screen voltage curve, although that is the condition that we
have with cutoff), with -350 VDC on the grid, about 1 mA will flow
with 10 kV across the tube. With 30 kV in our case, the current is
even more. Taking that minute current x the plate voltage, one can
understand that there is significant power being generated in that
beam to do some damage.
Redesigning the grid power supply to operate at -600 VDC, and buying
new tubes, seems to have fixed this mysterious failure problem.
Haven't lost a tube that way in the past 6 years - knock on wood.
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|