> what reason is there for a specific value of resistor in a
> parasitic suppressor. ?? I have seen 50 ohms, 100 ohms, and two 50
> ohms in parallel.. ! Is there a valid reason for using any particular
> value ??
Sure, here is how it works.
The resistor is the component that mostly dominates the system at
higher frequencies, while the inductor dominates at lower
frequencies. The reason is the frequency slope of reactance of the
inductor, which shifts more current into the resistor with increasing
frequency.
The resistor value has to be reasonably large compared to the overall
impedance of the path from the tube to the tank capacitor to ground
and through the chassis back to the tube.
Since the parallel equivalent impedance of the resistor and inductor
is in series with the anode path, a larger value of resistance and
inductance will de-Q the system and add more loss at higher
frequencies where the tube might tend to oscillate.
A tube that tends to oscillate very high in frequency (very short and
wide grid leads) requires less inductance, and a tube with long thin
grid leads requires more inductance.
A tube with a long thin anode lead or a long connection to the
chassis through the blocking cap requires more resistance to de-Q the
system, because series impedance is higher.
A larger resistor value also requires more inductance, since the
"current shift" into the resistor is proportional to the the ratio of
resistance to reactance.
The inductor value has to be large enough to be sure the resistor
dominates the path from the tube to the tuning capacitor at the
frequency where the tube is unstable.
In a grounded grid amp, that frequency is right around the frequency
where the feedthrough attenuation is minimum. With a 3-500Z with
short directly grounded grid leads, that is about 180-200 MHz. With a
811A or 572B tube, it is around 60-100MHz.
With a 811A or 572B (or 3CX1200A or D7) with long thin grid leads,
you need more inductance and more resistance. The tube tends to
oscillate lower in frequency. That makes the system hard to
stabilize, because the maximum operating frequency is generally only
about 30-50% of the frequency of instability.
With a tube like a 3-500Z, you need a modest resistance and much less
inductance. The amp is much easier to stabilize, and the suppressor
(if the anode and grid leads are kept short) can have very few turns
and use a modest value of resistance.
With a tube that has no appreciable feedthrough way up into or near
the GHz range, like the 8877 or other coaxial grid tubes, you can
often not worry at all about any suppression or use a very minimal
suppressor (like a length of brass strip) since normal transit time
of electrons as well as external circuit loss reduces gain far below
critical levels without any or with minimal suppression.
You can also, in some cases, add enough anode lead of the correct
width to simply move the anode resonance far below grid resonance, IF
the tube already has a fairly low amount of grid impedance at the
parallel self-resonant frequency of the grid. That usually means a
very short fat grid is required.
You can, with a network analyzer, measure ALL of the parameters and
do a perfect suppressor without any cut and try. I just did that
with a 3CX20000A7 system that someone else could not get stabilized.
The network analyzer showed feedthrough loss at 150MHz
was less than possible tube gain, and the anode system had a
reasonably high impedance with high Q at the same frequency. Sure
enough, the tube oscillated steadily while HV was applied (no big
bangs, just a few hundred mA of grid current and a steady signal on
the spectrum analyzer).
By shifting grid resonance upwards to 330MHz through improved
grounding of the grid (lowering the socket towards the chassis and
adding multiple short wide flashings between the grid ring and the
socket frame) and the slightly lengthening the anode
lead...oscillation went away without any "lumped" suppression...even
at 50% more voltage and a lighter load than will ever be used.
While we can do this all cut and try, it is pretty nice with big PA's
with a lot of large plumbing to do it with a cold PA before the amp
is finished. It is so much easier when we measure the impedance
looking from the anode end of the tank (with tube removed) towards
the tank with an impedance test set to know what value of parallel
equivalent impedance will de-Q the system, and using a network
analyzer to measure S12 to see what frequency tube feedthrough peaks
on a cold tube mounted in the final position.
As you look at PA's, you will see long thin leaded tubes generally
have more turns on the suppressor, and longer thinner anode leads
must use more resistance (which also pushes inductance requirements
higher).
73, Tom W8JI
W8JI@contesting.com
|