On Sat, 23 Jun 2007 11:56:21 -0700, wrote:


The IF is 1.7 MHz which is sampled using a 13.6 MHz clock by the
AD9874
and decimated by a factor of 48 to produce the 280 Ksamples/s output.

With 1.7 MHz IF you are going to have to construct your own LC band
pass filters in front of the AD9874, if 1.7 MHz is the first IF.

If this is the second IF and you have proper filtering (ceramic or
crystal) at the first IF, say 10.7 MHz, a very simple filter at 1.7
MHz should be sufficient. Due to not so spectacular performance of
that chip, I would suggest using a selectable IF filter at 10,7 MHz
(or whatever the first IF is), to even roughly match the required
bandwidth of the transmission.

If the IF filter would be 110-280 kHz wide to allow broadcast FM
reception, using such receiver in Europe to receive the 7.0-7.1 MHz
amateur band when there are very high power international broadcasters
starting at 7.1 MHz, might give quite disappointing results, since
those strong broadcast signals would control the AGC reducing
sensitivity and you still would get a lot of intermodulation products.

It also appears that the quoted dynamic range applies only to certain
AGC settings.

The use of 1.7 MHz IF with a very wide input filter may be problematic
at least in countries that are still using the AM broadcast band
actively, with very strong signals up to 1.6 MHz. On the other hand,
local 1.8 MHz amateur radio signals may be quite strong. In any
superheterodyne receiver, there is always a risk for signal
break-through from the environment to the IF stage at that frequency
range. For this reason, common IF frequencies, such as 10.7 MHz and
455 kHz, are usually excluded from transmitter frequency assignments.

Looking at:

http://en.wikipedia.org/wiki/Fm_broadcasting

I see that FM audio goes from 0 to 53 KHz and RDS is at 57 KHz.

While the modulation index for the RDS signal is quite low, thus
producing only the first order Bessel sidebands at +/- 57 kHz from the
carrier, the stereo difference signal S can have a quite high
modulation index, thus producing the first pair of Bessel sidebands
below +/-53 kHz from the carrier but also producing some significant
second order Bessel sidebands below +/-106 kHz and possibly even the
third order at +/-159 kHz.

Of course, this requires that there is a strong high pitch tone only
in one audio channel to get a huge difference signal.

Paul OH3LWR