Single-sideband modulation (SSB) is a refinement of the technique of amplitude modulation designed to be more efficient in its use of electrical power and bandwidth. It is closely related to vestigial sideband modulation (VSB) (see below).

Standard amplitude modulation produces a modulated output signal that has twice the bandwidth of the baseband signal. Single-sideband modulation avoids this bandwidth (and power) doubling, at the cost of device complexity.

SSB was pioneered by telephone companies in the 1930s for use over long-distance lines, as part of a technique known as frequency-division multiplexing (FDM). This enabled many voice channels to be sent down a single physical circuit. The use of SSB meant that the channels could be spaced (usually) just 4,000 Hz apart, while offering a speech bandwidth of nominally 300 – 3,400 Hz.

Amateur radio operators began to experiment with the method seriously after World War II. It has become a de facto standard for long-distance voice radio transmissions since then.

Signal generation

---

Consider an amplitude-modulated signal, which will have two frequency-shifted copies of the modulating signal (the lower one is frequency-inverted) on either side of the remaining carrier wave. These are known as sidebands.

One method of producing an SSB signal is to remove one of the sidebands via filtering, leaving only either the upper sideband (USB) or less commonly the lower sideband (LSB). Most often, the carrier is reduced (suppressed) or removed entirely. Assuming both sidebands are symmetric, no information is lost in the process. Since the final RF amplification is now concentrated in a single sideband, the effective power output is greater than in normal AM (the carrier and redundant sideband account for well over half of the power output of an AM transmitter). Though SSB uses substantially less bandwidth and power, it cannot be demodulated by a simple envelope detector like standard AM.

An alternate method of generation uses phasing to suppress the unwanted sideband. To generate an SSB signal with this method, two versions of the original signal are generated which are mutually 90° out of phase. Each one of these signals is then mixed with carrier waves that are also 90° out of phase with each other. By either adding or subtracting the resulting signals, a lower or upper sideband signal results. The device that performs this task is called a Quadrature Modulator, not to be confused with Quadrature Amplitude Modulation.

Throwing the baseband signal 90° out of phase cannot be done simply by delaying it, as it contains a large range of frequencies. In analog circuits, a phasing network is used. The method was popular in the days of valve radios, but later gained a bad reputation due to poorly adjusted commercial implementations. Modulation using this method is again gaining popularity in the homebrew and DSP fields. This method, utilizing the Hilbert transform to throw the baseband audio out of phase, can be done at low cost with digital circuitry.

Since Upper Sideband is nothing more than upconverted audio, these methods can be used to upconvert any signal also.

Mathematical highlights

Let $s(t)\backslash ,$ be the baseband waveform to be transmitted. Its Fourier transform, $S(f)\backslash ,$, is hermitian symmetrical about the $f=0\backslash ,$ axis, because $s(t)\backslash ,$ is real-valued. Double sideband modulation of $s(t)\backslash ,$ to a radio transmission frequency, $F\_c\backslash ,$, moves the axis of symmetry to $f=\backslash pm\; F\_c$, and the two sides of each axis are called sidebands.

Let $\backslash widehat\; s(t)\backslash ,$ represent the Hilbert transform of $s(t)\backslash ,$.   Then

$s\_a(t)\; =\; s(t)+j\backslash cdot\; \backslash widehat\; s(t)\backslash ,$

is a useful mathematical concept, called an analytic signal. The Fourier transform of $s\_a(t)\backslash ,$ equals $2\backslash cdot\; S(f)\backslash ,$, for $f\; >\; 0\backslash ,$, but it has no negative-frequency components. So it can be modulated to a radio frequency and produce just a single sideband.

The analytic representation of $\backslash cos(2\backslash pi\; F\_c\backslash cdot\; t)\backslash ,$ is:

$\backslash cos(2\backslash pi\; F\_c\backslash cdot\; t)+j\backslash cdot\; \backslash sin(2\backslash pi\; F\_c\backslash cdot\; t)\; =\; e^\{j2\backslash pi\; F\_c\backslash cdot\; t\}$   (the equality is Euler's formula)

whose Fourier transform is $\backslash delta(f-F\_c)\backslash ,$.

When $s\_a(t)\backslash ,$ is modulated (i.e. multiplied) by $e^\{j2\backslash pi\; F\_c\backslash cdot\; t\}\backslash ,$, all frequency components are shifted by $+F\_c\backslash ,$, so there are still no negative-frequency components. Therefore, the complex product is an analytic representation of the single sideband signal:

$s\_a(t)\backslash cdot\; e^\{j2\backslash pi\; F\_c\backslash cdot\; t\}\; =\; s\_\{ssb\}(t)\; +j\backslash cdot\; \backslash widehat\; s\_\{ssb\}(t)\; \backslash ,$

where $s\_\{ssb\}(t)\backslash ,$ is the real-valued, single sideband waveform. Therefore:

{| $s\_\{ssb\}(t)\backslash ,$ $=\; Re\backslash big\backslash \{s\_a(t)\backslash cdot\; e^\{j2\backslash pi\; F\_c\backslash cdot\; t\}\backslash big\backslash \}$ $=\; Re\backslash left\backslash \{\backslash \backslash widehat\; s(t)\backslash cdotF\_c\backslash cdot\; t)+j\backslash cdot\; \backslash sin(2\backslash pi\; F\_c\backslash cdot\; t)\backslash \; \backslash right\backslash \}$ $=\; s(t)\backslash cdot\; \backslash cos(2\backslash pi\; F\_c\backslash cdot\; t)\; -\; \backslash widehat\; s(t)\backslash cdot\; \backslash sin(2\backslash pi\; F\_c\backslash cdot\; t)\backslash ,$

And the "out-of-phase carrier waves" mentioned earlier are evident.

Lower sideband

$s\_a(t)\backslash ,$ represents the baseband signal's upper sideband, $s\_\{+\}(t)\backslash ,$. It is also possible, and useful, to convey the baseband information using its lower sideband, $s\_\{-\}(t)\backslash ,$, which is a mirror image about f=0 Hz. By a general property of the Fourier transform, that symmetry means it is the complex conjugate of $s\_\{+\}(t)\backslash ,$:

$s\_\{-\}(t)\; =\; s\_\{+\}^*(t)\; =\; s\_a^*(t)\; =\; s(t)-j\backslash cdot\; \backslash widehat\; s(t)\backslash ,$

Note that:

$s\_\{+\}(t)\; +\; s\_\{-\}(t)\; =\; 2s(t)\backslash ,$

The gain of 2 is a result of defining the analytic signal (one sideband) to have the same total energy as $s(t)\backslash ,$ (both sidebands).

As before, the signal is modulated by $e^\{j2\backslash pi\; F\_c\backslash cdot\; t\}\backslash ,$. The typical $F\_c\backslash ,$ is large enough that the translated lower sideband (LSB) has no negative-frequency components. Then the result is another analytic signal, whose real part is the actual transmission.

{| $s\_\{lsb\}(t)\backslash ,$ $=\; Re\backslash big\backslash \{s\_a^*(t)\backslash cdot\; e^\{j2\backslash pi\; F\_c\backslash cdot\; t\}\backslash big\backslash \}$ $=\; s(t)\backslash cdot\; \backslash cos(2\backslash pi\; F\_c\backslash cdot\; t)\; +\; \backslash widehat\; s(t)\backslash cdot\; \backslash sin(2\backslash pi\; F\_c\backslash cdot\; t)\backslash ,$

Note that the sum of the two sideband signals is

$2s(t)\backslash cdot\; cos(2\backslash pi\; F\_c\backslash cdot\; t)\backslash ,$

which is the classic model of suppressed-carrier double sideband AM.

---

SSB and VSB can also be regarded mathematically as special cases of analog quadrature amplitude modulation.

Demodulation

---

The front end of an SSB receiver is the same as that of an AM or FM receiver, consisting of a superheterodyne RF front end that produces a frequency-shifted version of the radio frequency (RF) signal within a standard intermediate frequency (IF) band.

To recover the original signal from the IF SSB signal, the single sideband must be frequency-shifted down to its original range of baseband frequencies, by using a product detector which mixes it with the output of a beat frequency oscillator (BFO). In other words, it is just another stage of heterodyning.

For this to work, the BFO frequency must be accurately adjusted. If the BFO is mis-adjusted, the output signal will be frequency-shifted, making speech sound strange and "Donald Duck"-like, or unintelligible.

As an example, consider an IF SSB signal centered at frequency $F\_\{if\}\backslash ,$ = 45000 Hz. The baseband frequency it needs to be shifted to is $F\_b\backslash ,$ = 2000 Hz. The BFO output waveform is $cos(2\backslash pi\backslash cdot\; F\_\{bfo\}\backslash cdot\; t)\backslash ,$. When the signal is multiplied by (aka 'heterodyned with') the BFO waveform, it shifts the signal to  $(F\_\{if\}+F\_\{bfo\})\backslash ,$  and to  $|F\_\{if\}-F\_\{bfo\}|\backslash ,$, which is known as the beat frequency or image frequency. The objective is to choose an $F\_\{bfo\}\backslash ,$ that results in  $|F\_\{if\}-F\_\{bfo\}|=F\_b\backslash ,$ = 2000 Hz. (The unwanted components at $(F\_\{if\}+F\_\{bfo\})\backslash ,$ can be removed by a lowpass filter (such as the human ear).)

Note that there are two choices for $F\_\{bfo\}\backslash ,$: 43000 Hz and 47000 Hz, aka low-side and high-side injection. With high-side injection, the spectral components that were distributed around 45000 Hz will be distributed around 2000 Hz in the reverse order, also known as an inverted spectrum. That is in fact desirable when the IF spectrum is also inverted, because the BFO inversion restores the proper relationships. One reason for that is when the IF spectrum is the output of an inverting stage in the receiver. Another reason is when the SSB signal is actually a lower sideband, instead of an upper sideband. But if both reasons are true, then the IF spectrum in not inverted, and the non-inverting BFO (43000 Hz) should be used.

If $F\_\{bfo\}\backslash ,$ is off by a small amount, then the beat frequency is not exactly $F\_b\backslash ,$, which can lead to the speech distortion mentioned earlier.

Suppressed carrier SSB and VSB

---

Suppressed carrier SSB modulation is used by ATSC. DSL modems implement suppressed carrier SSB modulation as well.

Vestigial sideband

---

A vestigial sideband (in radio communication) is a sideband that has been only partly cut off or suppressed. Television broadcasts (in NTSC, PAL, or SECAM analog video format) use this method if the video is transmitted in AM, due to the large bandwidth used. It may also be used in digital transmission, such as the ATSC-standardized 8-VSB. The Milgo 4400/48 modem (circa 1967) used vestigial sideband and phase-shift keying to provide 4800 bit/s transmission over a 1600 Hz channel. SSB-SC means single side band supressed carrier.

The video baseband signal used in TV has a bandwidth of 6 MHz. To conserve bandwidth, SSB would be desirable, but the video signal has significant low frequency content (average brightness) and has rectangular synchronising pulses.The compromise is vestigial sideband modulation. In vestigial sideband the full upper sideband of bandwidth W2 = 4 MHz is transmitted, but only W1 = 1.25 MHz of the lower sideband is transmitted, along with a carrier. This effectively makes the system AM at low modulation frequencies and SSB at high modulation frequencies. The absence of the lower sideband components at high frequencies must be compensated for, and this is done by the RF and IF filters.

See also

---

• modulation for other examples of modulation techniques
• Sideband for more general information about a sideband
• ACSB for Amplitude Compandored Sideband modulation

References

---

• partly from Federal Standard 1037C in support of MIL-STD-188

Radio modulation modes

Einseitenbandmodulation | Modulación en banda lateral única | Bande latérale unique | Single-sideband modulation | Enkelt sidebåndmodulering | Modulacja jednowstęgowa | SSB | SSB | SSB