>I recall an article in Ham Radio many decades by Bill Orr,
>with a super cathode circuit. As I recall, a good portion
>of the drive was to the cathode and the screen was
>connected down on the cathode choke. I have never seen a
>complete amplifier design, but as I recall there was some
>degeneration in the circuit.
Colin,
Orr lifted that design from Collins, but like Collins in
later years Orr made some mistakes in the application of
negative feedback. The problem is two-fold:
1.) In some systems is the phase changes quite a large
amount with tuning or over wide frequency excursions.
2.) In some systems the grid to cathode impedance varies
over wide excursions over the RF cycle, and with drive
level.
3.) A system cannot depend on the feedback being applied to
the same element that shields the cathode from the anode.
This is because the feedback circuit effectively lifts the
grid from the chassis for RF, so you no longer have a good
shield. It destabilizes the PA.
Imagine you have a system with tuned circuits or portions of
tuned circuits in or at the feedback path. As those circuits
are adjusted the phase of the feedback can change quite a
large amount. It is very easy to change from 180 degrees to
something less than 90 or more than 270 degrees as the tuned
circuits are moved through their range. That's one problem.
Another is the cathode to grid impedance. In a sub-class 2
amplifier, like AB2, the grid to cathode impedance varies
greatly with conduction angle. This makes it impossible to
have stable amounts of negative feedback by lifting the grid
through a divider unless the divider has many times LESS
impedance than the lowest grid to cathode impedance. Take
for example the poor idea of using a 3-500Z with the grid
floated through a small mica cap ala SB220 and clones. The
theory is the grid cathode capacitance forms one leg of a
divider with the grid to chassis cap as the other half of
the divider.
The problem is the G to K capacitance of the 3-500 is only
around 15 pF depending on socketing and layout. That
capacitance at the socket terminals also varies considerably
with frequency because of series inductance and transmission
line effects. It is a bit less on lower bands and
significantly more on higher bands.
The time-varying (due to drive level) grid to cathode
impedance shunting that capacitor is only a few hundred ohms
average, and goes from nearly an open circuit during
negative grid excursions to just a few dozen ohms on the
peak of positive excursions.
Now the misplaced notion is a capacitor from grid to chassis
makes a nice capacitive divider in conjunction with the
grid-cathode internal capacitor as the other leg, but in
order to divide voltage without shifting phase the FIXED
capacitances would have to totally dominate the voltage
division system. It doesn't come remotely close. The upper
capacitor in the divider has considerable series inductance,
and it has a reactance of 3k ohms on 3.5MHz and much less
than 200 ohms on 28Mhz (because of the series inductance and
frequency change).
So on 3.5 through 28 the tube has a lower grid divider leg
of around 38 ohms down to 5 ohms on ten meters. The upper
capacitor has a reactance of 3000 ohms on 80 down to less
than 200 ohms on 10 meters. The exact values move around a
bit from lead reactances.
That 3000 ohm reactance on 80 is in parallel with the
average grid impedance of a few hundred ohms!!!! This causes
a HUGE phase error in the divider. By the time ten meters is
reached the grid to cathode reactance is down to 200 ohms.
This is in PARALLEL with an average grid to cathode
resistance of 200 ohms, but that resistance varies over the
RF cycle from much less than 200 to nearly infinite. This
not only greatly changes the amount of feedback from band to
band and over each RF cycle, it changes the feedback phase a
considerable amount.It causes an increasing phase lead as
frequency is lowered.
The idea a grounded grid class triode can use a floating
grid for negative feedback is just silly. We not only can
calculate the errors, the errors show up in testing. Making
matters even worse, it is applied to tubes that draw grid
current!! Harmonics that contribute to IM have uncontrolled
phase, and the stability is compromised.
This is one reason why the 30L1 demands a long driver cable
and has a long history of field modifications to improve
stability, and why amps like the SB220 are improved by
directly grounding the grids.
The partially floating control grid has a place in a GG AB1
tetrode where the grid NEVER draws current and the screen
shields the output from input, but it is very poor
engineering in a AB2 amplifier or especially a triode!
A resistor in the cathode, if the stray C across the
resistor or series inductance to the tuned input or cathode
is low, does not do any of this. It is always a better
system by far.
But then, it was never exotic enough to be anyone's
exclusive pet idea, was it???
We get in trouble with these agendas where something that
works in one specific case is taken outside all boundaries
and applied to systems where it falls apart, just because it
is someone's pet idea or cause in life. This applies to
nichrome as well as that souper cathode drive.
When someone thinks they have the single universal cure all
for all applications look out!!!
73 Tom
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|