# How does FM work in the real world?September 30, 2020 3:53 AM   Subscribe

The images of frequency modulation are clear enough for a single signal. But in the real world if you increase or decrease the frequency of your signal, won't you end up interfering with other frequencies?

Just for example, if you're broadcasting a FM radio station at 90.7 MHz, how can you increase or decrease the frequency to encode information without stomping on other nearby frequencies?

This Wikipedia article says the channels are 200 kHz wide, so is it something like "FM 90.7 sends out 90.6 MHz for a 0 and 90.8 for a 1" or something along those lines?

I picked FM radio as a (hopefully) simple example, but I'm wondering how frequency modulation works in general, not just for radio. No need to explain amplitude or phase modulation, I think I understand those well enough.
posted by Tehhund to Technology (16 answers total) 6 users marked this as a favorite

...how can you increase or decrease the frequency to encode information without stomping on other nearby frequencies?

Basically, you can't. There's a reason radio frequencies are tightly controlled and allotted by the FCC. In any given area, you cannot have broadcasters operating on frequencies so close as to step on each other.
posted by Thorzdad at 4:25 AM on September 30, 2020 [2 favorites]

Response by poster: Ok, they don’t work by stomping on other frequencies. So how do they work?
posted by Tehhund at 5:06 AM on September 30, 2020

Best answer: It modulates within a range, so when an FM station is on say 90 Mhz it is really 90Mhz +/- 75Khz.

So you are right, you can't have frequency modulation on a single frequency it always needs a range to modulate within.

FM is not digital (caveats apply ), in really crude terms the amount the frequency is away from the center is the amplitude of the audio output (this is ignoring all the rf wizardry that is really needed).
posted by samj at 5:09 AM on September 30, 2020 [3 favorites]

This Wikipedia on carrier waves might help.
posted by soundguy99 at 5:11 AM on September 30, 2020 [3 favorites]

Best answer: It's important to consider the magnitudes.

FM 90.7 is actually 90.7 MHz, or 90,700,000 Hz. The channel is 200 kHz, which is 200,000 Hz. That means that FM 90.7 is actually allocated all frequencies between 90,600,000 Hz and 90,800,000 Hz. FM 90.9 would be between 90,800,000 Hz and 91,000,000 Hz, and so on. Each station varies their frequency within their allocated min and max, and therefore won't step on their neighbors.
posted by yuwtze at 5:19 AM on September 30, 2020 [6 favorites]

yuwtze has the right of it. And it isn't just FM that works like that. Say you have a pure sine wave at a single frequency. On a frequency-domain plot, it looks like a vertical line. Suppose you start turning it on and off to send Morse code. Now the line has a non-zero width (that depends on the symbol rate and how you're doing the windowing in your spectrum plot).

That's literally what "bandwidth" means -- the width of the chunk of the frequency band your signal occupies.

There are also cases (like spread spectrum) where you can get the information back out on the other end even when (possibly stronger) signals overlap with yours.

de NF3H
posted by sourcequench at 5:33 AM on September 30, 2020 [2 favorites]

Best answer: Each station varies their frequency within their allocated min and max, and therefore won't step on their neighbors.

That's true to a first approximation. But it turns out that frequency-modulating a carrier produces a signal which, if analyzed in the frequency domain rather than the time domain, does in theory contain components further from the nominal centre frequency than any fixed channel width would seem to allow.

But the nice thing about FM - actually the entire point of using FM instead of AM - is that an FM detector can reject various kinds of noise a lot better than an AM detector can. In theory, an FM detector can reconstruct the modulating signal purely from the zero-crossings of the modulated RF, and it can also ignore zero-crossings that occur too far apart or too close together.

So if you amplify the hell out of what comes out of the tuning stages and then amplitude-limit (clip) it, what you're left with is a signal whose zero-crossings are defined almost completely by the strongest available carrier amongst whatever mixture is received. You still get some zero-crossings caused by noise, but the timing of these will in general be far enough away from that of zero-crossings due to the main carrier that a demodulator can ignore them.

Once this kind of limiter circuit is designed in, interference from the extremes of the sidebands of transmissions on neighbouring channels is pretty much indistinguishable from random noise and just gets clipped off by the limiter, or ignored at the detector because its timing is too far out. In fact an FM receiver can even largely ignore transmissions on the same channel as the one it's tuned to, provided such overlapping transmissionss are weaker (i.e. arrive at lower amplitude).

This is why you will occasionally find your car radio flipping apparently at random between two different FM transmissions, only one of which will be audible at any instant: the weaker of the two signals just manifests as background hiss or roar behind the stronger one rather than being in any way intelligible. An AM receiver, by contrast, will detect multiple transmissions on the same channel as if all the audio had been combined before modulation, and you'll hear an audible mixture of both.
posted by flabdablet at 6:19 AM on September 30, 2020 [16 favorites]

Each station varies their frequency within their allocated min and max, and therefore won't step on their neighbors.

That is true, but the US has (or had) tighter rules than other countries. So in ye olden days, Mexican radio stations could overpower US stations in the same ranges.

Story about I Heard it on the X by ZZ Top, which is about a Mexican radio station that overpowers its US range mates.
posted by The_Vegetables at 7:51 AM on September 30, 2020

Mexico's X stations were all on AM, though, where FM's capture effect doesn't apply. There was an AskMe a few years ago querying how to shut down a church-owned community FM station that was deliberately broadcasting dead air on the same FM channel to block out a popular rock station from further away.
posted by scruss at 8:01 AM on September 30, 2020 [1 favorite]

In fact an FM receiver can even largely ignore transmissions on the same channel as the one it's tuned to, provided such overlapping transmissionss are weaker (i.e. arrive at lower amplitude).

As a real-world example, back in the early 90's I worked on some concerts that were live simulcasts on one of the big FM rock stations in the area, so I had a lot of interactions with the station's chief engineer. A huge part of his job was, essentially, maxing the amplitude of the station's broadcast so that it would be the strongest possible signal as far as possible - IOW, they wanted to be the strongest signal in their band all the way to the edge of their designated broadcast area, thus overpowering any possible overlapping transmissions from bordering stations. IIRC, occasionally overdoing this to the point of getting a "stop that" from the FCC was considered acceptable by his bosses. (Not often enough or strongly enough to get fined, of course, but apparently the FCC was willing to allow for a rare "accident" or miscalculation.)
posted by soundguy99 at 8:37 AM on September 30, 2020 [2 favorites]

This Wikipedia article says the channels are 200 kHz wide, so is it something like "FM 90.7 sends out 90.6 MHz for a 0 and 90.8 for a 1" or something along those lines?

Working in the frequency domain (which is the one where statements like "a channel is 200kHz wide" makes sense), modulating a carrier with another signal generates sidebands on frequencies other than that of the carrier.

If you amplitude-modulate a 100MHz carrier with a 1kHz signal and examine the result with a spectrum analyzer, what you'll see is a peak at 100MHz and two smaller peaks at 99.999MHz and 100.001MHz. AM produces two sidebands, one either side of the carrier frequency, each separated from the carrier frequency by the modulation frequency. The amplitudes of the sidebands depend directly on the depth of modulation.

If you do the same exercise with frequency-modulation, what you get is a whole bunch of sideband peaks at 100MHz ± 1kHz, 100MHz ± 2kHz, 100MHz ± 3kHz and so on. Again, the depth of modulation (which, for FM, translates to the maximum deviation from the centre frequency) affects the amplitude of these sidebands but in a much more complicated way. The higher the deviation, the more of these sidebands you'll see rising above the analyzer's noise floor. You also get weird effects on the carrier centre frequency as well: certain combinations of modulation frequency and deviation will suppress the carrier itself completely, leaving you with only a pile of sideband frequencies.

If you band-limit an AM signal - that is, if you filter the modulated carrier in such a way as to impose a cutoff on frequencies either side of the carrier - then you also band-limit the audio you can recover from the result. By contrast, band-limiting an FM signal doesn't affect the recovered audio's frequency response in such an immediately obvious way but it does worsen its signal to noise ratio, especially at the high end of the modulated audio frequency range.

So no, an FM transmitter will typically not deliberately deviate its carrier frequency to the limits of the allocated channel bandwidth; if it did, its sidebands would certainly cause major interference to the neighbouring channels. 75kHz is a typical maximum deviation for commercial FM transmitters and even then, multiple transmitters in geographically close locations will typically be assigned channels two slots (i.e. 400kHz) apart to avoid interference.
posted by flabdablet at 9:08 AM on September 30, 2020 [3 favorites]

This guy goes on and on a bit, but he does show you lots of pretty pictures of a spectrum analyzer looking at an FM signal.
posted by flabdablet at 9:11 AM on September 30, 2020 [2 favorites]

Just a note, you can get FM Transmitters | Amazon.com. They used to be really popular in the early days of MP3 players before car radios had AUX inputs or you didn't have a cassette tape player to use the other method of getting your MP3s to come out of your car's speakers. Put a sticker on your car and you could DJ for the few cars around you in that gridlock hell commute. Also handy at home if you still have clock radios and don't want to invest in whole house new-fangled tech.
posted by zengargoyle at 10:30 AM on September 30, 2020 [1 favorite]

Ok, they don’t work by stomping on other frequencies. So how do they work?

The important point is that they do work by 'stomping on other frequencies'[*], but they don't [or shouldn't] stomp on other channels.

[*] As others have already pointed out, any signal carrying data doesn't really transmit on a single frequency, but on a band of frequencies centered on the nominal frequency.

Channels are allocated at sufficiently wide frequency spacings to allow for this without interference.
posted by HiroProtagonist at 6:07 PM on September 30, 2020 [1 favorite]

>"FM 90.7 sends out 90.6 MHz for a 0 and 90.8 for a 1"

As an aside, what you're describing there is called FSK, frequency-shift keying.
posted by Standard Orange at 10:14 PM on September 30, 2020 [1 favorite]

And another way to think about FSK for a digital signal is that what you're doing with it is frequency-modulating a really heavily clipped analog signal. And the higher the bit rate, the higher the frequencies present in that clipped analog signal are going to be. Once the bit rate rises to some substantial fraction of the deviation (which would be 100kHz for a FSK signal with 90.6MHz for a 0 and 90.8 for a 1), the bandwidth of the transmission will widen substantially beyond those limits (see again Carson's Rule linked above).
posted by flabdablet at 2:45 AM on October 1, 2020

« Older Lights, Camera, Any Other Actions?   |   Asking the impossible Newer »

You are not logged in, either login or create an account to post comments