When two FM signals are transmitted on the same (or about the same) center frequency, either you hear only one or you hear that noise I'm sure most of you have heard. Would it be possible for a software-defined receiver to track both signals and demodulate both of them, and play them at the same time? I'm thinking of a receiver that would make FM doubles sound like SSB doubles.
-
2$\begingroup$ It's an interesting question, and my guess (not supported by anything I can cite) is: in the general case, no, existing demodulators already do the best they can do without any knowledge of the modulating signal(s). But in specific semi-practical cases (like knowing that there are two signals, they both carry voice, and their relative strengths are not changing too quickly) then some heavy-duty math might actually be able to do something. $\endgroup$– hobbs - KC2GMay 7, 2022 at 21:50
-
1$\begingroup$ The well-known capture effect exhibited by analogue FM receivers is in part due them demodulating the FM only, and throwing away the AM component. An SDR does not need to do this (but some software may anyway), and can in theory use the amplitude information to do a better separation of the two signals. It's easy to imagine a pair of signals that could not be separated however. It will depend on the specific signals, and the specific SDR software. $\endgroup$– Neil_UKMay 8, 2022 at 18:11
-
$\begingroup$ Perhaps apropos: github.com/pvachon/tsl-sdr (the multifm part) $\endgroup$– user21417May 10, 2022 at 15:25
-
$\begingroup$ @JohnVE3WNA does multifm do what I was asking about? $\endgroup$– SomeoneMay 10, 2022 at 16:22
-
$\begingroup$ @Someone no idea, but it might be expanding on some of the ideas in the accepted answer. The author says "This is supposed to be of interest to people who might want to demodulate multiple channels to feed different parts of a monitoring workflow." Might be worth a look. $\endgroup$– user21417May 10, 2022 at 16:24
3 Answers
Yes. A high sample rate (much higher than the audio frequencies modulated) SDR plus an FFT based demodulator could play both as a mix, but not always separate the two. You can clearly see the wiggles from two separate FM signals in a high sample rate high resolution waterfall, except where they overlap. You don’t have to be “captured” by only one frequency deviation track if you capture the entire possible bandwidth as IQ samples, and not just make the incorrect, but common, assumption by locking onto just one deviation track. The problem becomes tracking which signal is which when the original audio amplitudes, thus FM frequency deviation tracks, match, or criss-cross. But if you simultaneously tracked two sets of frequency deviations, both the high one and the low one, then demodulate both of those tracks, an STFT/FFT based SDR demodulator could play a mix of two FM signals in the same channel, assuming both are clearly above the noise floor.
As an alternative to a high res STFT/FFT 2D matrix for FM deviation tracking, a parametric best fit of two sinusoids to the captured IQ data stream might work even better, assuming only 2 FM signals of near constant RF amplitude in the same channel. The problem then converts to deciding when there are actually exactly 2 signals worth demodulating and mixing, and not just one signal plus a track of noise that should be ignored.
In theory, it would be possible for an SDR to capture the primary signal, demodulate it, and subtract it from the total signal and extract the second signal from the remains and demodulate it.
Practically, this is a bit more difficult, and extracting the primary signal won't be perfect (even when you can hear the primary, it still gets noise from the secondary), and any defects in extracting the primary signal will significantly degrade the second signal. Also the likelihood of success of that heavily relies on your AGC and the bit depth and sensitivity of your ADC and how much of the secondary is left after subtracting the primary. If the AGC turns down the gain to the point where the secondary is under the noise floor of the ADC, it's just not going to work.
Also, if you are hearing it as buzzing, that means that the radio failed to capture either of the signals because they are too close in power. In this case, it is unlikely you'll have any luck separating them.
Having said that, who knows if someone could get extra clever and find new ways to separate the signals and improve the results. DSP leaves a lot of room for possibilities. Unfortuantely, while there has been some work on this, it is not a well studied problem and probably nobody has tried to commercially exploit it or is likely to try.
-
$\begingroup$ Note: I've thought about this a while, and it is a pet project I'd like to see done, but I don't have the DSP expertise (or time) to do it. $\endgroup$ May 8, 2022 at 2:24
Even SDR receivers can be subject to the capture effect. Depending on how the limiter or demodulator is designed often competing signals will cause the weaker one to be attenuated. This means that trying to demodulate two signals near the same carrier isn't any easier with SDR than with traditional RF design.
-
1$\begingroup$ A lot can be done with DSP software, but I tend to agree. Still, I was going to post a link to this in dsp.SE Chat, but it appears all their rooms are either frozen or specific. $\endgroup$– Mike Waters ♦May 7, 2022 at 22:17
-
1$\begingroup$ @MikeWaters I did brief search before I posted my "probably not" answer, but clever demod design in software might offer a solution, so let me know if you run into contradictory info. $\endgroup$– user21417May 7, 2022 at 22:19
-
$\begingroup$ Yes, "probably not" is a good answer, John. :-) Upvoted. $\endgroup$– Mike Waters ♦May 7, 2022 at 22:22
-
$\begingroup$ I'd go with "sometimes possible" but "probably not" isn't wrong. :) $\endgroup$ May 7, 2022 at 22:28
-
1$\begingroup$ Capture has nothing to do with a receiver being an SDR. It has to do with which specific DSP algorithm is used to demodulate the digital samples. An SDR doesn't throw most of the information away, but samples the full channel bandwidth. Makes lots of thing easier. $\endgroup$– hotpaw2May 8, 2022 at 4:34