In article ,
says...
Hello fellow radionuts (radionauts?). For some reason, I've decided
to design and build an FM tuner. Just to see if I can make a good one
I guess. A few questions have presented themselves:

-PLL or quadrature (IC) decoding?

I can't speak to your own needs and interests, but for my part, coming
off a big all-mode receiver project, I would say that my next one will
definitely be... DSP-based.

I used the Philips SA604AD in my receiver for both WBFM (broadcast) and
NBFM decoding, the former at 10.7 MHz, the latter at 455 kHz. It is
quadrature-coil based, easy to use and not too hard to find. All of the
coils and filters needed for all FM modes are stock parts at DigiKey.
But the SA604AD doesn't decode stereo directly, so that may be an issue
for you.

-Stereo decoding: single IC or "discrete"?

Again, the next time I go through all this rigamarole, the final IF
output will go straight into an ADC.


-Throw in SCA! Why not? PLL?

In a non-DSP solution, the choice of PLL-versus-quadrature techniques is
usually driven by the need for stereo decoding. You're more likely to
want a PLL to regenerate the 19 kHz pilot tone for stereo decoding.
This is not necessarily the same stage (chip) that does the FM
demodulation. Stereo MPX decoding is strictly a baseband process,
regardless of implementation.

I didn't care about stereo support in my application, so a quadrature
solution was fine. I don't know much about the mechanics of stereo
pilot carrier recovery and channel separation, never having implemented
it or researched it in earnest. Google returned
http://transmitters.tripod.com/stereo.htm, which at first glance looks
like a good summary of the process.

-I'm still having a hard time figuring out AFC. I can see how I might
do it with a PLL - the DC error voltage being used to pull the LO to
the correct frequency. How was it done in the old days? ;-)

Almost the same way. You could think of AFC as a crude FLL (frequency
locked-loop). The output of the ratio detectors in the old-school FM
sets had a DC bias that corresponded to tuning error. When the radio
was exactly on-frequency, the average carrier frequency deviation was
zero, and the discriminator's DC output was centered. Drift would cause
the development of a non-zero DC bias in one direction or the other,
which was fed back to the 1st LO to steer it back on frequency. (Prior
to the widespread use of varicap diodes, this was a rather kludgy
process.)

There was no phase-locking effect in an AFC system, because ratio
detectors can only be used to compare a single incoming frequency
against a fixed tank circuit's center frequency. For a true PLL effect,
you need a way to generate an error voltage or current on the basis of
the phase difference between *two* signals.

If I
could design not only AFC, but also automatic tuning of an RF
preselector and perhaps even an antenna tuner, that would be tres cool.

My suspicion is that you can dispense with the preselector and just use
a strong (high-IP3) front end with a bandpass filter. 88-108 MHz is
considered a narrow tuning range these days.

-- jm

------------------------------------------------------
http://www.qsl.net/ke5fx
Note: My E-mail address has been altered to avoid spam
------------------------------------------------------