Owen Duffy wrote:
(Richard Harrison) wrote in news:23000-
:
Jim Lux wrote:
"in a linear system"
It produces no significant harmonics, so the system is linear.
That is a new / unconventional definition of 'linear'.
The term is usually used in this context to mean a linear transfer
characteristic, ie PowerOut vs PowerIn is linear.
Or, as I used it, that superposition holds.
One can build an amplifier or other device where the Pout(Pin) =straight
line, but is not linear in the formal sense. Say you built a widget that
measured the input frequency and amplitude, then drove a synthesizer at
that frequency and amplitude = 2*input amplitude.
Considering a typical valve Class C RF amplifier with a resonant load:
Conduction angle will typically be around 120°, and to achieve that, the
grid bias would be around twice the cutoff voltage.
If you attempted to pass a signal such as SSB though a Class C amplifier
that was biased to twice the cutoff value, there would be no output
signal when the peak input was less than about 50% max drive voltage, or
about 25% power, and for greater drive voltage there would be output. How
could such a transfer characteristic be argued to be linear?
It would not be.You're right
The active device isn't linear.
neither is the whole assembly.
I think, though, that sometimes we take a more casual view of linear
(e.g. people talk about the linearity of a log detector.. referring to
the deviation from a Voltage out=dBm in straight line.)
And, some confusion about nonlinear devices in a building block that is,
by and large, linear (e.g. a power op amp with an AB2 output stage and a
fair amount of negative feedback) with some constraints on frequency and
amplitude.
Owen