Patrick Turner wrote:
Jon Noring wrote:

Note again that Patrick himself said that IF is not needed when one is
building a single frequency receiver -- and from his comments he is a
very strong advocate of superhet design for a tunable receiver. That's
all the channel TRF is: a single frequency receiver, duplicated n
number of times (where n is the number of channels one wants to tune,
which are switched in and out.)

120 separate AM channels with perhaps 480 discrete LCs and two tubes
each is an entirely overcomplex and impractical idea.

Sigh.

It was a bad choice of wording on my part, since I assumed from what I
previously wrote that what I intended was obvious: that the tubes and
RF transformers remain the same, but the rest of the LC components of
the bandpass tuning stages will be swapped out from channel to
channel.

This is functionally *equivalent* to having 120+ independent and
optimized TRF circuits (one for each channel) -- that's what I
intended to say.

This is more than obvious, since in traditionally-tuned radios, nearly
all the components remain the same except the tuning capacitor (or for
a few radio designs, a variable inductor.)

Same with the channel TRF: all channels use the same common components
except those whose values/properties must change as a function of
tuning frequency, which are the bandpass filter components. Thus the
same tubes and RF transformers (as a matter of practicality) are
intended to be commonly used for all the channels.

Now, again, why use the channel approach when one can use either a
single or double tuned bandpass filter? It's a matter of the degrees
of freedom one is given in optimizing the bandpass characteristics.
In the channel TRF we should be able to, in principle at least, assure
that for each channel, from 500 khz to 1800 khz, we can have
essentially the same exact bandshape: bandwidth, shape factor, etc.
And higher order filters are definitely a possibility (if it makes
any sense to use them -- delay/linear phase is an issue.)

This degree of bandshape control cannot be accomplished by tuning one
or two capacitors (or inductors) in the bandpass tuner. In the TRF as
John Byrns is studying, simply adjusting the capacitance for tuning
has the downside of increasing bandwidth for higher frequency (in a
simple parallel RLC circuit, BW=(1/RC).) There are tricks that can be
done within the limited parameter space to keep the bandwidth more
constant, but it probably has a negative effect on the shape factor
and degree of linear phase, and still does not give the degree of
control preferred (of course, a variable inductor suggests itself.)
Thus was born the "channel TRF", taking advantage of the fact that,
for the BCB at least, all broadcasts are on pre-assigned frequencies
(channels), so why care about being able to tune in-between these
frequencies?

(Now, as I think about it, it would be possible to build a continuous
tuning system for these higher-order bandpass filters, optimally
varying each of the LC components to their best values as a function
of center frequency. It could be a "true quintuply tuned circuit" (or
higher order.) But mechanically accomplishing this would get
extraordinarily complicated: having to increase this capacitance a
certain nonlinear way, decrease that inductance by its own function,
slowly increase another capacitance, etc., all at the same time. This
is much much much more complicated than even the all-channel TRF.)

Jon