I have posted a message ABSE, with plotted response of the RF input
in a typical old tube AM radio, with a single simple air cored LC.

The graphs at ABSE plots the tested response of a typical
input circuit used in valve radios.

The test used a typical RF air cored input transformer
with loose magnetic coupling, to isolate the antenna
from the secondary and tuning gang used to tune the radio.
RF input was diect to the primary via a low impedance sig gene,
and the levels used to avoid overloading the following gear.

The output from the tuned circuit was taken to a 33k R and then
via a screened cable to an RF amp with about 470k Rin
The low Z output from the RF amp was then fed to a CRO, from
where the amplitude was plotted, and also to a signal frequncy meter,
so accurate F measurements could be made.

Fig1&2 shows the plotted response of the tuned circuit at three
different
places on the MW band.

Notice how the Q falls as F rises.

But also notice that the pass band is wider as F rises.

Fig 3 shows the more detailed response of the input LC at 1,000 kHz.

The RF attenuation at 1,040 kHz is much greater than 6 dB/octave

Once away from 1,000 kHz, at 1,400 kHz, the RF attenuation rate of the
LC is 6 dB/octave.

Fig 4 shows what happens to the AF response, assuming the AM modulated
RF
is flat from 0.0Hz to say 100 kHz.

If the tuned signals were detected to recover the audio signals,
the response of the recovered audio would follow
the Fig 4 profiles.
As you can see the widest AF bandwidth is where the Q is lowest,
and where the carrier is highest at 1,400 kHz.

But at 600 kHz, the AF bw is -3 dB at 10k.

If there are following tuned circuits with a response which is similar
to the input circuit at 600 kHz, and without a central dip in response
at Fo,
the audio bw will be cut further.

The reason for the cutting of AF bw due to the RF pass band is as
follows.
Please refer to Fig1 which shoews the RF response at 600 kHz.
There is a slight error in recording the peak in the response,
but i measured a bw between F1 and F2 of 23 kHz,
so lets round that down to 20 khz,
so the response is -3 dB down when the RF is +/- either side of
the cntre tuned frequency.

If the RF carrier at 600 kHz is modulated with a 10 kHz audio tone,
then sidebands exist in the signal at 590 kHz and 610 kHz.
These signals are attenuated by 3 dB.
Thus the recovered audio is lower, since we depend on the amplitude of
the
RF signal after passing through the filter to get our wanted audio
signal.

If we had a two two stage RF amp, each with a tuned circuit
like Fig1, the recovered af bw would be further reduced.

The attenutation well outside the pass band, say at 700 khz, would be
down
40 dB, not just 20 dB, and if the stations we listen to ate 100 khz
apart,
the mixer tube won't be much affected by cross modulation of a station
at
600 kHz even if there is a more powerful one at 700 kHz.

If the 455 kHz IF amp is designed with a flat topped response,
the usual result is even more " sideband cutting" and the audio response

will suffer .

If the IF was at 2.2 mHz, and each IFT had a 20 kHz pass band to start
with,
with critical coupling for a flat topped response,
then three cascaded IFT which contained 6 tuned LC circuits would still
allow 10 khz of audio
bw, and perhaps give enough attenuation of about 60 db
for signals at 70 kHz off the IF frequency.

To get more audio bw through a 455 khz IF,
one has to have the first IFT of two set up slightly twin peaked
and slightly over critically coupled.
Thus the arched response seen in fig 3, which is that of a
single tuned can be made considerably wider and I suggest
any lost readers poke their noses
into the text books re tuned RF or IF transformers, where there is a
tuned P and S circuit.

The RF bw has to be twice the wanted af bw, since we detect only
oe side of the envelope.

It is possible to detect both +ve and -ve sides of the AM envelope,
and combine them to create a supposedly less distorted signal,
but usually if the stations' signals have symetrical
+ and - waves on the envelope, and the receiver is aligned right,
there is no need for detecting both sides of the envelope.

Patrick Turner.



..

"Patrick Turner"

I have posted a message ABSE,


** Not visible to an Optusnet customer.



........ Phil

On Sat, 03 Jan 2004 00:12:50 +1100, Phil Allison wrote:


"Patrick Turner"

I have posted a message ABSE,


** Not visible to an Optusnet customer.


Oddly enough, the only a.b.s.es on Clara (in the UK, anyway) appear to be
alt.binaries.sounds.emmylou and alt.binaries.sounds.erotica. Somehow I
don't think Patrick has posted to either of those! My main ISP,
Blueyonder, doesn't carry anything so depraved as emmylou... :-)

--
Mick
http://www.nascom.info for Nascom & Gemini information
Also at http://www.mixtel.co.uk where the collection started.
Currently deserting M$ for linux... :-)