john wilkinson wrote:
Thank you all for being pacient with me and my stupid questions.

Another thought has just raised its head.

I have 2 IF amps. 1ts with approx 20dB gain, second with 100dB gain.
The first IF has a XTAL filter at 45MHz.
The second has the collins mech filter at the beginnig, then 100dB gain,
straight into a diode detector.

I think I need some form of additional filtering at the output of the
second IF, before the detector, to reduce the noise introduced by the
100dB amp, else it don't work. Except for strong signals.

Is that right?

Many thanks,
John.

Andy wries:

Exactly. The noise generated by a 100 db amp with a fairly wide
bandwidth with be great compared to the noise which is admitted thru
the narrow bandpass filter in front of it, and will probably be the
major
factor to the noise floor of the system..

Do NOT use the canned Noise Figure programs in a case like this
since they usually do not consider the bandpass of each individual
element..... You have to write your own. It ain't rocket surgery, but
canned programs have their limitations...

It is a very poorly designed receiver that has a high gain wide band
amp following a narrow filter...

Optimally, two narrow filters should be used --- one in front of
the
high gain amp to prevent interference and intermod from out of band
signals, and a second narrow band filter to keep the noise from the
high gain amp from reaching the detector.....and set the noise
bandwidth of the sytem...

In practice, filters are often distributed thru the gain of the
amplifier
to keep the noise bandwidth down in addition to stopping out of band
signals. Using two narrow band crystal filters is much more
expensive.

In my homebuilt receivers, I use the two xtal filter approach, and
never
have a problem. But I am building one of a kind, not for production,
and
don't mind spending a few more dollars in parts.....

But, again, yes you are correct in your analysis.... 100 db is far
too
much gain to put in a wideband amp feeding a wideband detector for
keeping the noise floor down...

Regarding FM versus Am ---- noise floor is noisefloor... A high
noise
floor will degrade FM as much as AM.... Limiting does NOT buy higher
sensitivity, since if the noise causes most of the limiting, the phase
noise
introduced to an FM detector is just as bad as the amplitude noise
put into an AM detector....... One small, insignificant exception is a
possible
5.6 db SNR improvement if the FM signal is wide band, but in narrow
band
FM signals the SNR improvement is negligible.....Remember , you can't
tell narrowband FM from AM on a spectrum analyzer ---- the sideband
levels are about the same..... Fm starts having an advantage only when
you start adding in the extra sidebands ( see Bessel function chart)
since
the FM sidebands are coherent and the extra noise is not --- this is
an advantage that wide band FM has......



Andy W4OAH