I do regret sounding as if I'm on a side here, but I must quite disagree
with many comments regarding modelling. It is very much like selecting
tools for the repair of your autobile. For some mechanics, that which
cannot be repaired with a vice-grip is unworthy of attention, and for other
mechanics, they must use a torque wrench for any tightening job. Prudence
and judgement are what make a true engineer different from a technician, and
I regret to say that the vast majority of hams are not technicians, but only
those who hope to someday be technicians. Having put the offensive bit up
front, those whose interest is piqued may read on:
Tom has indicated that modelling programs report vastly incorrect numbers in
terms of the Q of a solenoid inductor at tens or hundreds of Megahertz.
This is true. Three dimensional modelling programs were not made for that
task; they are made for modelling the electrical properties of physical
structures such as are used at tens of GHz and up, and are generally planar
or pseudo-planar. I do not know of any modelling tool that professes to
properly model that inductor. And why would I want such a tool? It is an
easy parameter to measure, and then use a macro model in simpler software,
why would anybody with good judgement attempt the task?
Bill, WA4LAV, wrote:
>My electronics instructor used to say often the problems with a
>circuit is the component you can't see. Especially with RF circuits.
Your electronics instructor was correct. Thankfully, with very moderate
test instruments, we can measure what we cannot see. The inductive effects
of long leads and long ground return paths can be measured, and inserted
into our models. Gone is the day when a grid dip meter was a useful
instrument. They belong in the trash bin next to the logarithmic decrement
meter. Or in a museum. A perfectly useful impedance measurement setup can
be had for less than the cost of a pair of 3-500Z tubes, and anybody who
professes to do amplifier work had best own such a setup or be admonished as
not serious about the purported expertise. Of course, those who are
interested in building an amplifier from time to time may be excused from
owning such equipment, but honestly, if I were an antenna experimenter, I'd
want such as well. AEA, MFJ and other offer perfectly good equiment for
reasonable figures, and the used market proliferates with used HP 8405 and
similar equipment. Quite frankly, I get far more use out of my impedance
measuring apparatus than I do out of a VOM.
Dave, AA6AQ said:
>Traditional techniques have failed to resolve the case in question --
>whether or not VHF oscillations occur, and how to best suppress them if
>they do -- in at least 3 years of near-daily interchange on this reflector.
>The combatants are personally motivated to conclusively prove their
>hypotheses; based on the continuing polarization, no one has come close.
>One must consider effectiveness when comparing techniques, not just
>convenience or familiarity.
In this case, I'm not quite certain what you mean by "traditional
techniques". However, I have used very simple programs - public domain
versions of SPICE, eCAP, and the like, to model amplifiers using tubes and
solid state devices for nearly 30 years. In EVERY case (all caps intended)
I have been able to develop a model that predicted the nature of
oscillations that the amplifier might be capable of sustaining. I consider
these to be "traditional techniques". Why the so-called combatants on the
reflector have refrained from using analytical techniques varies. Tom has
indicated that he believes modelling takes far more time than measuring. I
claim that by modelling, you improve the benefits of measuring, and vice
versa. Rich appears to lack the necessary measurement equipment to validate
any modelling he might want to do, so perhaps we are better off with Rich
not having any analysis programs available. Clearly, Rich needs a vector
measurement system. He says that hams owe him $1100 in terms of suppressor
kits not paid for (I hope that does not include ones that he offered for
free, as he has done with me), we know his finances are high enough to pay
for a good vector measurement system, and he would no longer be at the mercy
of someone else's measurements. Why he chooses not to place $200 into what
it takes to continue to advance the theories he proposes is subject to
conjecture.
It has been written that an amplifier either oscillates, or it doesn't.
That is a half truth. It is possible that an amplifier is on the verge -
that is, with a slight change to some parameter, such as valve gain, a
resistor value, or stray capacitance, it will oscillate. This is ABSOLUTELY
a measurable, and predictable condition. Rich is wrong in this matter,
parasitics are NOT unpredictable. Just because they are intermittent does
not preclude deterministic analysis.
If an amplifier analysis shows that there is a forward gain of 13 dB at some
frequency, and a reverse attenuation of 12dB at the same frequency, then you
have a potential oscillator on your hands. I once did an assessment of an
amateur amplifier which was claimed to oscillate. Having made parametric
measurements of the valves involved, two-port measurements of the input and
output networks, I then adjusted the stray elements in my SPICE model, and
the model actually showed the amplifier to be 1/4dB shy of the gain required
to sustain oscillation. No wonder that if the tune capacitor was tweaked
the wrong way, it would break into oscillation (at the predicted frequency).
This analysis, by the way, required all of 45 minutes to develop.
Again, prudence and judgement must be used, or hysteria will prevail.
You build the model, then you build the amplifier. To the extent that the
amplifier does not agree with the model, you sort out what is wrong with
your model. You add series inductance in your model to the tune cap. You
add parallel capacitance in your model to the pi-L inductors. You make the
model work. Now you can truly model the effect of suppressor components in
the input and output circuitry.
Given what the models (and my later measurements) show, I'm surprised that
nobody here waves the flag about the nature of the conductors used to mate
with those vaunted suppressors. The length and nature of them are critical;
after all, they carry all the anode current. Modelling of some
overly-compact designs has shown that it is actually stray capacitance to
ground, and not any pi-L network strays, that cause stability issues that
are resolved with parasitics.
There is no cure all. I have encountered amplifiers that required parasitic
suppressors, and I have encountered amplifiers that were destabilised by
same. I have never encountered an amplifier that a model showed as stable
that oscillated. And every amplifier that I've seen oscillate has been
predicted to do so, or be close to doing so, via modelling.
FF
_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.
Share information about yourself, create your own public profile at
http://profiles.msn.com.
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|