Most of the terms
on this page are
used with vacuum
tubes amplifiers,
but the basic
information also
applies to solid
state devices.
Review of
parasitic
oscillation cause
Parasitics are
almost always from a
tuned-plate
tuned-grid
oscillator mode. This behavior applies even in grounded grid amplifiers because,
at some frequency, the grid is no longer effectively grounded.
The
governing criteria
for oscillation is
regenerative
feedback has to
exceed losses in the
system.
Low-Q circuits will
oscillate quite
well, so long as
feedback exceeds
loss. As a matter of
fact, many
oscillators operate
with passive
component circuit
Q’s under 1. R-C
audio oscillators
are a good example
of very low-Q tuning
systems that have no
problem oscillating.
In order to
oscillate with low
feedback, stage gain has
to be quite high.
With low
feedback levels a
system generally
needs the oscillator
to be formed in a
conventional
grounded-cathode
circuit (highest
gain) with the anode
and grid resonant on
or near the same operating
frequency. As a
general rule the
anode has to be
resonant slightly
higher in frequency
than the control
grid, so the anode
load is inductive.
This produces a more
favorable phase
shift. We might
think a
grounded-grid
amplifier circuit
solves this
problem, but it does
not. While the grid
is grounded outside
the tube, it is
often not
grounded
at very high
frequencies
inside the tube.
Review of unwanted
resonances and gain
The most
problematic
resonances in a
vacuum tube are
resonances in the
grid circuit. This
is because, as we
probably all know,
the grid voltage has
the largest
influence or control
over the amplifying
system. We can say
“the gain is in
the grid”.
The grid is a
fairly large
structure and with
physical size comes
considerable
capacitance. The
grid and the leads
are also long, and
with length comes
inductance. The
combination of the
two can be a source
of significant
design problem. At
some frequency
control grid
capacitance and
series control grid
connection
inductance form a
parallel tuned
circuit. This
parallel resonance
divorces the grid
from ground near
that resonant
frequency. This is,
without fail, the
problematic
frequency.
If we put a network
analyzer on the
cathode and look at
feedthrough voltage
appearing on the
anode, we will see a
sharp peak in anode
to cathode coupling
at the frequency
where the grid is
parallel resonant.
This is the target
frequency where we
want to apply
dampening, and of
course it varies a
great deal with
amplifier
construction and the
construction of the
tube and socket.
Tubes with long thin
single grid leads
are the most
unstable at VHF,
while tubes with
thick wide multiple
grid leads provide
significantly higher
grid self-resonant
frequencies. The
most stable tubes,
contrary to what an
inexperienced
engineer or
apprentice
technician might
think, are actually
the tubes most
useful at VHF and
UHF! It might appear
contrary to common
sense, but the
easiest to stabilize
tubes are almost
always the tubes
with highest gain in
VHF or UHF service.
The reason for this
is a major parameter
limiting gain at VHF
or UHF is shunting
grid capacitance and
series grid
inductance. The very
same things that
lower grid parallel
resonant frequency
also decrease useful
external VHF gain. Useful
external VHF gain is
entirely different
than internal VHF
gain. They often run
opposite to each
other!
As we alter the
grid connection
length, the
frequency where the
grid
“floats”
changes. As with any
tuned circuit, any
series inductance
caused by long leads
(even through
capacitors or
resistors) increases
inductance and
decreases unwanted
control grid VHF
resonant frequency.
While hidden in
components, this
unwanted tuned
circuit behaves like
any tuned
circuit. If we
add inductance to
the path unwanted
grid resonance moves
lower in frequency.
The closer grid
resonance
moves to the
operating frequency,
the more difficult
building a stable
workable amplifier
becomes.
If we clamp the
leads of a 3-500Z
directly to the
chassis with nearly
zero length from the
pins, the grid
resonance can be as
high as 200MHz in a
3-500Z. This makes
suppression easy.
Other than moving
the grid resonance
higher through
shorter leads,
anything we do to
the grid reduces the
frequency where
self-oscillation
might occur. We
generally must
attack instability
outside of the grid
path, although on
rare occasions we
can modify the grid
path to aid
stability. To
suppress any
potential
oscillation, the VHF
anode path to the
chassis must be be
dampened by a
resistance that at
least nearly equals
or exceeds anode
path distributed
reactance. The anode
system problem is
generally one of
adding a dissipative
resistance that
loads the anode
enough to reduce
gain so the feedback
to the grid cannot
sustain oscillation.
This is where it
gets complicated.
There are dozens of
ways to dampen the
anode, none of them
universally better
than others. For any
particular
application or
situation the
solution can be
quite different. For
a tube in a cavity,
the walls might be
lined at a critical
point with low-Q
ferrite. For the
3-500Z example at
HF, we would try to
have the shortest
anode to chassis
path through the
tuning cap, and
insert a series
inductance with a
parallel resistance.
At 150-200 MHz we
would want the anode
path to have a low
impedance, and the
the
resistor/inductor
combination we place
in series to appear
mostly like a
dissipative
resistance. This
generally means a
wide short anode
lead or short
parallel wires
spaced some distance
apart (to minimize
inductance) and
enough turns on the
suppressor so it, in
combination with
plate capacitance,
resonates at the
unwanted frequency
(150-200MHz). The
resistor across the
inductance then
becomes a load, and
if a modest value
dampens the anode
and reduces VHF
gain.
In short form with a
3-500Z what you need
to do is make sure
the grid pins are
grounded directly
with nearly zero
length connections
directly to the
chassis. You need to
be sure there is
some shunting low
impedance reactance
at 150-200 MHz that
you can apply a
series resistance to
the path, and that
the added series
inductance does not
get so large that
excessive current
flows through the
resistor at the
operating frequency.
This is where the
problem is. Despite
what some people
say, there is no
specific universal
cure for all
situations. What
works in one
situation might not
be best in another.
There are many cases
where a suppressor
is not even
required! There are
cases where a
suppressor will make
the system worse.
There are cases
where the design is
critical. There are
cases where almost
anything will work. Without getting a
feel for what
actually happens, it
becomes a matter of
guesswork. This
really where all the
non-technical
nonsense and
arguments come from.
People don’t take
the time to learn or
understand how the
system works, and
look for magical
snake-oil universal
cures. Easy answers
are almost always
far from optimum
solutions and some
easy answers can
be worse
than no answer at all.
The only hard and
fast rule is keeping
grid grounding
short and direct
will almost always
increase stability
of a power
amplifier. What we
do in the anode
really depends
heavily on the
physical and
electrical
construction, and
without knowing
the system in great detail it
becomes impossible
to suggest the
optimum system.
What is a
parasitic
suppressor?
A “parasitic
suppressor” is
a circuit that adds
dissipative
resistance to a
circuit path in
an amplifier or
oscillator. It also alters the resonant frequency of the path where the
suppressor is inserted. Both of these effects are key to stopping or preventing
unwanted oscillations.
The
dissipative
resistance provides
a load for the circuit,
absorbing energy. This
loading
reduces available gain
over some
specific frequency
range, and broadens any resonance of the entire path where the suppressor is
inserted. Q of the suppressor itself is
not important. System
impedance,
or loading of the
system, is important. Optimum
complex impedance
varies greatly with
the circuit type, the particular components and
wiring used, and the
physical layout of
the circuit.
The suppressor’s reactance, like any reactance, changes the resonant
frequency of the path in which it is installed. Both the path impedance and
suppressor impedance, as a series connected system, must be considered in any
analysis.
We can think of a
suppressor as a
simple L/R load,
optimized for maximum effect on the system’s
impedance at unwanted frequency, ideally inserted
where maximum effect at the spurious frequency, and minimum effect at desired
frequencies, is obtained.
Where is
the suppressor
placed?
A suppressor is
normally placed in
the lowest impedance
portion of the path
involved in the unwanted
instability. In a
transistor or FET,
this can be the
collector or drain
lead. In a vacuum
tube suppression is
normally (but not
always) placed in
the anode lead. The
anode is a better
place because the
impedance from the
physical element
inside the tube to
the external anode
connection point is
almost always a very
short, heavy, direct
connection. This
means a very low
amount of additional
external resistance
has a large effect
on system gain.
Back
|