Hi Dave,
> Determining by measurement that one instance of a circuit fails to
> oscillate does prove false the assertion "every instance oscillates". But
> the domain here is production, where there are hundreds or thousands of
> such instances; one cannot measure them all.
One can not model an RF system without measurements.
> The question is whether, with some combination of reasonable parameters,
> some instances oscillate. Modelling is a pragmatic way to answer that
> question -- it allows evaluation of the multiple combinations of
> parameters likely to be encountered during a circuit's production.
You don't know those parameters without measurements.
> Models and modelling results can also be shared via the net, permitting
> independent verification.
You won't know the models are valid unless you make
measurements, and analyze the model step by step to be sure it
agrees with the measurements.
With measurements and calculations, you can determine the
feedback and gain limits. From that, you can tell if it's possible to
oscillate.
You can't do a model without those measurements and some of
the calculations needed to make the model. All the model will do is
handle the "math" as you change various parameters.
You can do that without a model just as easily as with a model.
Let me give you an example of how people assume models are
correct. Look at the Q of solenoid inductors on modelling
programs, and then measure some inductors on an accurate
instrument ( I have a mid-80's HP Impedance test set that
measures Q on any frequency from 100 kHz to 1 GHz).
The highest HF Q I've ever measured for a large solenoid inductor is
in the upper hundreds, about 800. If I walk near the inductor, the Q
drops.
To measure that Q, I have to put the inductor in a shielded fixture
that is about three feet square.
I can model the same inductor on software, and easily get Q's of
many thousands. Not only is Q greatly over estimated, Q in many
modelling programs increases with frequency right up to self-
resonance. That does not happen in a real inductor. Q peaks below
the frequency where the inductor is self-resonant, and then rapidly
decreases as the self-resonant frequency is approached. The
reason for that is circulating currents in the inductor increase
rapidly as the inductor approaches self-resonance, and losses also
increase. Apparently the models ignore many effects.
IMO, if modeling programs can't handle a simple inductor
properly...they need to be carefully watched with other components.
73, Tom W8JI
w8ji@contesting.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
|