>
> > (Me)After some discussion about how oscillators work, someone said:
> .
> (Jon) I'll take responsibility cause I said it!
>.(a summary of stability concepts longer than I want to quote)
Jon, that was a good summary. One point I've been ranting on about, though,
is that the amplifier is a nonlinear system and as such one must be careful
applying linear analysis to it. Concepts such as gain margin, phase margin,
K factor and conjugate impedance matching all arise in linear system theory,
and are indeed enormously powerful. And in practice the first and sometimes
only method one applies when analyzing a nonlinear system is to linearize it
about its operating bias point. You then have a "small-signal" model on
which the tools of linear analysis can be brought to bear. But this model
will break down if the system is driven hard enough - and we clearly drive
our amplifiers into the non linear regime. The tube is cut off for a
substantial portion of the cycle. Linear that's not! The first-order symptom
is often generation of harmonics, and they're obvious in the anode current
waveform of anything but class A. But instability and even chaos can also
result. Proving unconditional stability of a nonlinear system is usually
extremely hard. Indeed, points of instability can be scattered around the
operating domain and can be separated by infinitesimal amounts from points
of stability, which is one of the really cool things that the study of chaos
and fractals has taught us. What that means, though, is that whether you
will find them during testing can be a gamble.
Note that I haven't proven that things are all that bad for us. I've only
argued that no one to my knowledge has shown that they're NOT that bad! I
truly hope the situation isn't as dire as it conceivably could be, as I hope
to built a better amp for myself some day!
Regards, Carl WS7L
--
FAQ on WWW: http://www.contesting.com/ampsfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|