ca3189 am receiver circuits
(ww1.microchip.com/downloads/en/DeviceDoc/70090a.pdf) shows a 10.7MHz ceramic discriminator needed for FSK demodulation of signal. As I read, a discriminator is essentially a resonator.
Can I use a standard crystal in parallel with a resistor in place of the discriminator? Is there a difference between a crystal and a ceramic discriminator (eg bandwidth) ?
For FM demodulation you need 75kHz bandwidth. Ordinary crystal is too much narrow. Ceramic discriminator is designed for needed bandwidth and you can preciselly tune it. It is actually paralel resonance circuit.
But I can reduce the Q and widen the bandwidth with a parallel resistor. Am I wrong?
These types of discriminators use the delay and multiply method. The ceramic resonator forms a one pole band pass filter with delay in the center frequency. Whatever method you use, you must have relatively constant delay across the signal bandwidth or there will be severe distortion. If you do not want to use a ceramic resonator, your next best choice is a LC tank circuit which will have labor cost of tuning each unit. This may not be a problem on a system which is not manufactured in large quantities.
Usefull bandwidth with crystal is limited to few hundred Hz. For linear demodulation at 75 kHz bandwidth and more using LC tank circuit you may look at datasheet for CA3189. It is an old FM IF discriminator but uses the same demodulation principle. There is an example of double tuned tank circuit for low distortion which might give you an idea how to realize it. Using single tank circuit and reducing it's Q for greater bandwidth leads to reduced sensitivity of discriminator.
Thank you for the answers.
Actually linearity is not important, because it is an FSK not an FM demodulation. I could use LC tank as well, but I wanted to design the receiver with no tuning needed at all. Unfortunately I don't find a source for a 10.7MHz discriminator, so I have to experiment with finding out the parameters for a suitable tunable coil or tune the capacitor in the LC. But the capacity range needed is too large, 0-60 pF, I can realize it only with a fixed-value chip cap in parallel with a tunable one.