Perhaps we should look at what we want in a suppressor.
I hope this ascii drawing looks ok after going through mailers.
The system looks like this:
La Ltank
___sup____(((((((___________(((((_____
__I__ _I_
g --- - - - - ---- C tune
^ I
_^_____________chassis_____________________________
Between g and chassis we have an impedance that varies from being
very low to fairly high. This impedance is determined by the grid
lead and grid structure stray capacitances, inductances, and
resistances.
Ltank impedance is a very high inductive reactance at VHF, while C
tune is very low. The result is the primary impedance of the anode
system (La) at VHF is inductive, and is parallel tuned by the anode
to chassis capacitances.
That makes the anode INSIDE the tube see a very high impedance
(parallel resonant) at some VHF frequency.
The grid also has the same effect, with the grid to chassis
capacitance parallel tuning the grid lead and structure inductance to
resonance at some very high frequency.
The combination of a parallel resonant grid system and parallel
resonant anode system, and grid to anode capacitance forms a TPTG
(tune plate tuned grid) oscillator.
Of course at other frequencies the anode and grid leads are SERIES
resonant and present a very low impedance inside the tube. In this
case the tube will be more stable. The problem with using a grid dip
oscillator to look for "problems" is it tells us NOTHING about the
impedance. It might indicate a strong dip at a good resonance that
lowers grid and anode impedance, just as easily as it indicates a dip
at a "bad resonance" that increases anode to ground impedance and
grid to ground impedance. We'd have no idea which is which.
Only passing a signal through the tube from cathode to anode would
tell us whether the impedances inside the tube are high (bad
parallel resonances) or low (good series resonances).
For example, a 3-500Z's "bad spot" is about 180-220 MHz in most
good layouts with short direct grid leads. The bad spot becomes
lower in frequency with longer leads (even if they are through 200 pF
or larger capacitors). The 3-500Z, in a good layout, tends to
oscillate between 150 and 220 MHz. In the AL-80B, the oscillation
occurs somewhere around 200 MHz. This is repeatable and can be
measured by SHORTING the suppressor (in some cases HV needs to be
raised to get the tube to oscillate) and looking at the frequency of
oscillation on a spectrum analyzer.
Feedback occurs from the grid to anode capacitance. This feedback is
degenerative, and so requires appreciable phase shift to become
regenerative. The feedback capacitance itself adds a phase lead
somewhat below 90 degrees to the feedback path.
The main problem is resonances in the grid and anode also add
additional phase shift, and on some frequency the total phase shift
of ALL these effects might combine with sufficient feedback and tube
voltage gain to exceed the circuit loss. If that is the case the tube
oscillates, if that is not the case the tube will NEVER oscillate
(even if "rung" by a transient).
What we generally want to do is add enough series resistance at "sup"
to de-Q the anode path impedance. This resistance loads the system,
reducing phase shift change with frequency and VHF gain. The reason
it lowers gain is the system is left mostly with the anode to ground
capacitance in parallel with the tube's internal resistive losses.
This reduces VHF gain, and makes the phase shift less critical.
Remember even low values of reactance don't absorb energy unless there
is a resistance or load someplace in the system, so we don't want
infinite resistance either. That's why we generally want something
between extremes.
As Peter pointed out, and as I have on numerous occasions, there is
an optimum value of impedance at "sup" that maximizes stability. It
is not when the parallel equivalent resistance of the suppressor
system is zero ohms, because in that case the equivalent series
resistance of the suppressor as a two terminal device is ALSO zero
ohms.
A suppressor is more likely to be effective if it's two terminal
impedance presents a high value of resistance in parallel with a
very high value of reactance, or presents a high value of resistance
in series with a very low value of inductance (either is the same).
What we really need to do is consider the two terminal impedance of
the suppressor at every frequency, and try to "present" a load
resistance to the anode that properly dampens or loads the system.
The worse possible impedance for the suppressor is a very low
resistance, because it starts to look like it isn't even there.
Another problem if the suppressor's loading resistance is too
high the suppressor's inductor just looks like an extension of the
anode leads, LOWERING the anode parallel resonant frequency.
That could move the anode resonance near the grid resonance in some
cases, and form the undesired TPTG oscillator.
A nichrome coil in parallel with a resistor looks more like two
resistors in parallel over a wider frequency range. That improves low
frequency stability (where the amp is generally stable anyway, unless
it's 811A's, 572's or other tubes with long internal leads). But at
higher frequencies (where most amps oscillate) the Q is essentially
the same with either suppressor!
Unfortunately the hairpin suppressor (as used in the Titan
modification kit) actually INCREASES VHF Q of the anode SYSTEM,
because the suppressor simply doesn't have enough inductance and
its two terminal resistance is a lot lower than Ten Tec's
suppressors. The moderately high anode lead reactance combines with
the lower series resistance of the nichrome suppressor at VHF to
increase system Q.
It's interesting to note that even if the Q of the suppressor, as a
separate device, is very low... it can raise the Q of the anode
system when compared to a suppressor with higher self-
Q. For example?
A one ohm resistor in series with an inductance of one ohm
has a Q of one. A fifty ohm resistor placed in series with a
inductance of twenty five ohms has a Q of two.
When they are connected in series with an anode inductance of 200
ohms, the system Q becomes 4.5 with the fifty ohm (Q=2) suppressor
and Q=201 with the lower self- Q suppressor.
That's the danger of looking at the Q of an individual part of a
complex system, and deciding you have saved the world from
all potential oscillations.
73, Tom W8JI
--
FAQ on WWW: http://www.contesting.com/ampfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|