SSB Demodulation

SSB Demodulation Techniques

The demodulation of single-sideband (SSB) signals requires special attention, because simple mixing leads to superposition of the upper and lower sidebands at audio frequencies. The following article gives an overview of the different methods for SSB demodulation and their use in software defined radios. SSB modulators and demodulators are sometime also referred to as image reject mixers.

1.     The Filter Method

The filter method is the traditional SSB reception method in analog superhet receivers. Typically a first mixer translates the signal to an intermediate frequency (IF) first. At the IF a sharp band-pass filter (e.g. a quartz filter) simply selects only one of the two sidebands and suppresses the other one. Then the second mixer converts the remaining sideband to audio frequencies. If required, the audio spectrum can be inverted by using high-side LO injection in one of the mixer stages.

Showing the filter SSB demodulator data flow and the corresponding spectrum

The filtering method for SSB reception

The filter method requires the availability of very sharp filters at the IF and is widely spread in the analog domain, where quartz filters are used. Sideband suppression is determined by the sharpness and attenuation of the filter at the unwanted sideband.

2.     The Phasing Method

The phasing method uses complex IQ processing to resolve the superposition of lower and upper sideband at audio frequencies. The incoming signal is directly converted to audio frequencies by a complex mixer creating I and Q components. The complex mixer requires a sine and cosine LO (e.g. implemented by a phase shift of 90°). The following low-pass filters with a width of one sideband determines the final bandwidth. After filtering the Q component needs to be shifted by 90° using a Hilbert transformer, before it can be added or subtracted to select one sideband.

Showing the data flow for the phasing ssb method and its spectrum

The Phasing method for SSB reception

To realize acceptable sideband suppression, the 90° phase shift for the LO and the 90° shift between I and Q, as well as their amplitude have to be matched exactly. This is hardly practical with analog components. However in the digital domain, the LO phase shift can easily be realized by a sine and cosine waveform. Also amplitude variations are not a problem in digital processing, if the 90° phase shift for Q is distributed between both I and Q path with a 45° and -45° phase shifter. This makes the phasing method interesting for implementation in a software defined radio.

The only drawback is the exact implementation of the 45° phase shifts. Even with a digital filter it is hard to achieve an accurate and constant shift of 45° over all audio frequencies. In practice already small phase deviations lead to prohibit very good sideband suppression, especially for frequencies at the frequency borders of the sidebands.

SSB suppression diagram for phase and amplitude errors

Sideband suppression for the phasing method with different phase and amplitude errors shows the vulnerability to imbalances. Even a tiny amplitude error of only 0.1 dB and phase error of 0.1 degrees limits the suppression to 45 dB.

Further reading for the phasing method:

3.     The Weaver Demodulator (The “Third Method”)

The Weaver demodulator (also called the “Third Method”, besides filtering and phasing techniques) for SSB reception has been introduced by D. Weaver in “A Third Method of Generation and Detection of Single-Sideband Signals”. It differs a little bit from the first two methods since it does not resolve any superposition of lower and upper sideband. Instead the Weaver simply converts a portion of the spectrum to audio frequencies without any ambiguities. The Weaver demodulator uses two complex mixing stages. The first mixer stages translates the signal in order to center it at zero frequency. There the desired bandwidth is selected by low-pass filters with half the bandwidth of a sideband (f_{edge} = BW_{SSB} / 2 ) . A second mixer stage translates the signal in order to align its frequencies with the audio frequencies. Summing or subtracting I and Q selects the orientation of the output spectrum (normal or inverted).

Showing the data flow for the Weaver demod and the spectrum for Weaver processing

The Weaver method for SSB reception

The Weaver method is very well suited for software defined radios, because digital mixers and their sine and cosine LO signals can be generated with extremely high phase and amplitude accuracy. The same holds for the two identical low-pass filters. Thus no gain and phase variations occur and the circuit is perfectly balanced for very high performance.

Further reading on the Weaver demodulator:

4. Further Methods

Further methods for SSB demodulation are possible, that e.g. rely on the fact that any modulation type can be considered as a superposition of amplitude and phase/frequency modulation (see e.g.: “Innovative demodulation method for SSB technique“, The method uses no mixers, but requires a small modulation index and suffers from poor sideband suppression and distortions due to nonlinearities. Therefore its practical relevance is somehow limited).

Another method performs signal conversion to the audio band in the frequency domain. For this purpose first a FFT is executed and then the relevant sideband frequency bins are copied to audio frequencies. It seems that Pieter’s (PA3FWM) well-known WebSDR works that way.

Recent Posts