Add Single Sideband (LSB / USB) demodulators

[lordmorgul]

Single Sideband is used for voice (or "phone") modulation as well as computer generated digital data signals in amateur radio. It is used in several frequency bands, but HF and VHF are most common from 1.8MHz to over 148 MHz. There are general guidelines for which sideband to use in each frequency band, but not all countries have the same frequency band limits for these guidelines.

The code would need to allow either sideband to be selected in any valid frequency range.
Being able to dynamically switch between sidebands would be helpful.
Processing both sidebands and saving the signal with the highest amplitude might be beneficial, but would also be compute intensive and reduce number of modulators that can be running.
Being able to observe and record a wide bandwidth would have value.

SSB either uses the Lower Sideband or the Upper Sideband. It is an Amplitude modulated signal with filter to remove the carrier and one sideband. This makes it very efficient to transmit, but the receiver must restore the carrier in order to demodulate it. To do that each modulator channel would need to have a carrier signal generator at the center of the modulator channel fed into it to mix then demodulate.

The SSB signal can be difficult to understand if the frequency is shifted slightly and may be hard to get working sufficiently. Receivers typically have a "clarifier rx" and possibly "clarifier tx" feature which is a subtle shift in frequency up or down to align better to the source carrier frequency. Selecting the center frequency for the modulators may not be as simple as the highest power, so this could also result in need to improve the scanner code for logic when SSB demod is selected.

[lordmorgul]

There are two common receiver designs to separate sideband:

Weaver method (freq shift, filter, freq shift)
from reddit gnuradio discussion: https://www.reddit.com/r/GNURadio/comments/1ma23tu/ssb_demodulation_using_weaver_method/

USB: Input from SDR -> Freq shift down by 1800 Hz -> lowpass filter with 1800 Hz cutoff -> Freq shift up by 1800 Hz -> Complex-to-Float -> Add Real & imag together -> Output to audio sink

LSB: Input from SDR -> Freq shift up by 1800 Hz -> lowpass filter with 1800 Hz cutoff -> Freq shift down by 1800 Hz -> Complex-to-float -> Add real and & imag together -> Output to audio sink

This has a major shortcoming where multiple signals can overlap each other.

Hilbert Transform
https://www.dsprelated.com/showarticle/176.php
Intelligibility tests indicate that a Figure 14(c) BFO -Δω frequency error of less than, say, 150 Hz can be tolerated
This indicates the binning across bandwidth may need to be on that order, which is quite small, and could be too CPU intensive. Scanner module design may be impacted.

hilbert transform block, taps set to 4k to start
ssb signal to dc, take the hilbert of the iq data
LSB = I + Q
USB = I - Q

Saving refs to gnuradio single sideband designs
https://www.youtube.com/watch?v=4fEtBS27loI
https://www.youtube.com/watch?v=3EPsCP25TeU
https://wiki.gnuradio.org/index.php/Simulation_example:_Single_Sideband_transceiver
https://www.oz9aec.net/radios/gnu-radio/simple-ssb-receiver-in-gnu-radio-companion
https://www.oz9aec.net/radios/gnu-radio/from-double-side-band-to-single-side-band-reception

[john-]

A lot of good information here. If you don't have some already it will be good to have some IQ dumps of samples that can be replayed in ham2mon when developing/testing. It might help others (like me :) contribute to the coding.