rpw3@redwood.UUCP (Rob Warnock) (07/20/85)
As several people have pointed out, I blew it talking about AGC and FM radio squelch tails. I WAS thinking about AM or SSB. (*blush*) How DOES an FM squelch work??? (I admit it, I have no idea...) Rob Warnock Systems Architecture Consultant UUCP: {ihnp4,ucbvax!dual}!fortune!redwood!rpw3 DDD: (415)572-2607 USPS: 510 Trinidad Lane, Foster City, CA 94404
jhs%Mitre-Bedford@d3unix.UUCP (07/25/85)
Blushing is definitely not necessary on the net. You are EXPECTED to make innacurate statements from time to time to keep us on our toes. If it happens naturally (as it does with most of us) then that just makes it less work to think them up. Most FM squelch circuits work by exploiting the difference in spectral density of noise coming out of the discriminator (or equivalent demodulator) on quiet or modulated carriers versus noise resulting from Gaussian noise alone in the IF strip. The most common form in my (casual) experience is a high-pass or bandpass filter off the discriminator output, followed by some kind of amplitude detector such as a rectifier and lowpass filter, and of course a threshold device such as a Schmitt trigger. The filter has to reject the speech band, and should pick up just above it and then drop off again after a couple of octaves. It is surprising how sensitive the things can be to even a slight carrier as opposed to pure noise. Apparently the high-frequency noise just above the speech band, say 5 to 20 KHz, quiets very quickly when a weak carrier pops up. This is probably due to the sudden disappearance of the "clicks" that occur in FM when the IF signal-space phasor wraps around the origin. It does this all the time when the noise is pure Gaussian with approximately equal power in any two quadrature phases. When you "bias" it toward a particular sinusoidal phase, it suddenly becomes much harder to get it to wrap around the origin. It has to overcome the rapidly-diminishing Gaussian probability distribution to get over to the origin. To understand the "clicks", you have to understand that an idealized FM demodulator (discriminator) looks at an IF wave and "computes" as its output the time derivative of the instantaneous phase of the narrowband random process it sees at its input. The phase is the arctangent of the ratio of the instanteous in-phase and quadrature-phase signals at any given instant. What is going on is that really rapid phase changes (which give large spikey outputs with lots of high frequency components because, remember, the phase is differentiated) tend to occur most easily when the amplitude in BOTH in-phase and quadrature phase channels happens to be small at the same time, i.e. the point in the plane representing the input random process is very close to the origin. When a signal is injected, this ensures that this "phasor" stays away from the origin and does not get to generate the rapid phase changes that occur without a signal, and especially not the really nasty ones that encircle the origin. The squelch detector should not be too sensitive to much higher frequencies than necessary, because these are slower to quiet than lower frequencies. The phase differentiation process leads to a nearly flat spectrum for low signal amplitudes and a triangular one, i.e. one that rises linearly with frequency, for large SNR. The best place to sense the noise is just above the speech band, i.e. at as low a frequency as possible consistent with not being fooled by the signal modulation. -John S., W3IKG