Absolutely right Tom.
If whomever is dealing with passband stability, the suppressor is almost out of
the circuit. It is some small series jXl, where it will matter little with the
high operating impedance of the tube, and the Pi L or Pi net it is feeding !
Without going into S parameters and analytic stability analysis, I will give
the reader's digest version (math is easier, but does help explain the physics
of the problem).
Inband stability is probably much more a function of phase shift from input to
output because of the input and output tuned ciruits, gain and isolation where
enough feedback in an amplifier with little phase margin (the input and output
are almost in phase, when instead, they should be nearly out of phase) cause an
oscillation (isolation does not overcome gain). Note that each Pi or L network
will introduce large phase shifts, and the sum of these thru the amplifier can
leave it such that the amplifier output is in a non benign phase at the input
such that poor isolation will make it sing inband.
I had an amplifier from a manufacturer recently that was not unconditionally
stable. That is, some setting of tune and load settings would cause the thing
to take off on 10m. The amplifier has a pair of 4cx800s and is high gain
(about 17 dB). I reviewed the schematic and modified the input circuit to
improve phase margin by changing the tuned input to include some zeros instead
of poles (series L and shunt Cs are lowpass or poles, series Cs and shunt Ls
are highpass or zeros).
Wa-La ! Amp fixed, it did not take off anymore. I trapped the oscillation on
the scope, and it was in the low 30's of MHz, meaning it was inband.
My point is this: if you have an oscillation, you must first determine where it
is: inband or outband (VHF/UHF). If it is inband, suspect bad phase response
because of too many poles (or too many zeros) in your matching systems. As an
example, here's a rough count:
Input:
- Coupling Cap: Outband Zero and essentially non inband phase affecting.
- Input Match: 3 poles and inband affecting. Consider changing this to a
highpass design (series C, shunt L, series C) to balance poles and zeros.
Output:
- Plate Blocker: Outband Zero, non inband phase affecting.
- Suppressor: Outband Pole at least until 10m, non inband affecting.
- Output Match: Pi or Pi - L almost always 3 inband poles for the Pi, and 4
for the L. Definately affects inband phase response.
The low hanging fruit is to change the input circuit from low pass to high pass
which will probably improve the phase margin such that the amplifier will be
stable again inband.
Having said all of this, a relatively low gain grounded grid amp (8 - 12 dB)
should not take off inband regardless of what tuned input is used; the output
circuit should be sufficiently isolated such that even -20 dB isolation
overcomes the gain of the tube provide there is some phase shift from in phase.
But, in the case of a grid driven Tetrode, or high gain triode, you may not
have enough isolation, and if the poles and zeros in your amp (plus the 180
shift of the tube) cause and in-phase relationship at some frequency...that amp
is on its way to the moon.
73,
Steve, K0SF
----- Original Message -----
From: "Tom W8JI" <w8ji@w8ji.com>
To: g3rzp@g3rzp.wanadoo.co.uk, amps@contesting.com
Sent: Friday, July 13, 2007 11:14:45 AM (GMT-0600) America/Chicago
Subject: Re: [Amps] suppressors
> Steve said:
>>In every case so far, copper gives
> better 'goodness'.<
> Which, from purely a theoretical viewpoint, is exactly
> what should be expected.
> 73
> Peter G3RZP
N7WS measured identical magical nichrome and conventional
Ameritron suppressors, and found at VHF near the frequency
of suppression there was essentially no difference at all.
As frequency was reduced to HF and lower, the nichrome
suppressor remained lossy.
A nichrome or other lossy resistive conductor suppressor is
NOT a "low-Q VHF suppressor". It is actually a low Q low
frequency suppressor.
This makes perfect sense and follows conventional wisdom,
since the primary path for lower frequency signals is
through the inductor. The primary path for VHF and highest
frequency signals shifts into the resistor. Thus the coil
loss affects dc the most, and UHF the least. The resistor
affects UHF and the highest frequencies the most, and dc the
least.
By adding a resistive conductor we lower HF Q the most,
while barely changing VHF Q.
If I had an amplifier that was unstable at or near the
operating frequency and if I couldn't neutralize the
amplifier, adding resistance that swamps the signal
frequency might be a solution. If I really had a VHF
oscillation in an HF amplifier then the solution would be
increasing inductance and SHUNT resistance.
It's amazing how such a very simple system has become so
misunderstood, almost to the point of being voodoo.
73 Tom
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|