The full quote is "Just the Facts, Ma'am."

A tip of the hat to Sgt. Friday, who always solves the case with a least waste of words. An admirable model.

But caution: this is me, not Friday, and economy has been thrown to the wind on these tabs!

 

 

About that diagram ...

You graciously went this far. You deserve an explanation. But there's so much more (at the tabs to the right).

 

Just one complex multiplication! Many smart people find this ridiculous, until thinking about it a bit. In the end, it's just like the "full" phase method (weighted everywhere by m(t)'s instantaneous magnitude) of generating SSB, as used in the late '40s and early '50s ironically enough (before affordable, sharp IF filters)! Recall the double balanced mixer.

A quick word about magnitude. It's not the same as amplitude! Amplitude applies to a single scalar thing, like a voltage read on a VOM, or, in math, a function of one single-dimension variable. Changes in amp(litude) makes your speaker cone move. Magnitude, on the other hand, applies, in any way that we'll care about here, to two single-dimension variables taken together, but each of a different dimension. Mag(nitude) is like a right triangle hypotenuse, i.e. mag (z) = sqrt [ amp^2 (x) + amp^2 (y) ], where z is a complex number made up of two simple scalars, x and y, at "right angles," or orthogonal, to each other. (Orthogonality is the fundamental thing about dimensions!) Now, onward.

The Real part of a (complex-valued) analytic signal contains all the information. You don't need the Imaginary part for moving information!* The analytic signal, s(t) - j Hi [s(t)], is merely the essential (seemingly redundant, but not!) expression allowing a "complete" multiplication of two phase-varying "sinusoids," also called phasors. Hi [ ] is the Hilbert transform, in which +cosine becomes -sine, and +sine becomes +cosine, either a step forward by 90 deg (by that "imaginary operator" j). Note that by the very definition of j, the two halves of the analytic signal are orthogonal. Finally, since linear signals, e.g. voice but almost anything else as well, are just sums of weighted sines (i.e., a Fourier series), you have your RF signal. QED! So it's multiplication of two phasors producing a product phasor.

Now to get SSB this way, your rig's method may vary, but many if not most SDRs do exactly this at some IF a bit above voice, or above a preferred CW tone, or above an arbitrary digital (or analog!) "audio." (One example: "AFSK," or Audio FSK, presents inputs centered at some fc = 32/n or 48/n kHz, n = 2 or more to allow for a rig's digital IF of 32 or 48 kHz, in which case it's actually not even AFSK as it's no longer audio! It will become an SSB in the rig's DSP at digital IF using phasor multiplication.) Then, heterodyne the result, often if not usually, digitally up to a radio IF. The unwanted RF image at that point is straightforward analog filtering, perhaps using the rig's roofing filter in the opposite direction as receive. Hooray! (You could even go all the way to final transmit f digitally, using an anti-sampling, or interpolation, filter just before the amplifiers. This is exactly the idea behind the direct digital synthesizer, or DDS.)

Now, let's cement this whole concept by way of an accurate analogy to something most of us know cold. Consider it to be the same idea as Nyquist's sampling theorem. If you have only a time series of sample amplitudes, but no phase information about each sample, then you need two time amplitude sequences, say A and B, each of length N, "interleaved" together, one sample A followed by a second sample B, etc., in order to resolve frequency unambiguously in the same time as either A consumes or B consumes. Of course, that means that the interleaved time sample rate will be twice the resulting unambiguous frequency sample rate. That's Nyquist, at the limit. So you can immediately see that any two adjacent samples of A and B, as an "effective analytic sample" in this analogy must provide phase at that point in time, and that the "size" of this effective analytic sample must linearly reflect the sizes of the two alternate simple samples. And that's what a discrete Fourier transform does. This alternation that's the essense of Nyquist is just a little bit unequal to simultaneously-obtained complex samples, but it's very close to perfect and the difference is another digression that can wait. In the end, Nyquist is just another example of analytic signals. He probably knew that!

Now, whew, back to that diagram! That's an Upper Sideband (or USB) illustrated there. For the Lower Sideband (or LSB), just change the sign of j Hi [m(t)] from - to +. Yes, amazingly, to obtain the LSB the modulating signal m(t) goes in angular progression as if backward in time! This is exactly why the LSB spectrum falls below the fc of c(t), while of course the USB falls above! But don't change that - j Hi [c(t)] sign! The carrier sine (or cosine) progression of a "real" signal (the only thing transmittable as real power expressed by a propagating EM field) is always forward in time from zero (0) in that diagram above. Causality rules. No time travel allowed! Unless of course, it's Back to the Future!

 

* This is a critical fact about amplitude modulation, or AM, and even about FM as well. You don't need complex arithmetic for either of those. The Imaginary part of an analytic signal s(t) is identical to the Real part but offset at every instant in time by pi/2 radians in phase, i.e. in quadrature. This two-part "scheme," whereby we mean a real function and its negative imaginary Hilbert transform (function), can carry, as the content of an SSB, any kind of modulating signal, so take your pick from analog audio (the usual you hear on HF) to FM, to digital, to whatever. But always be aware that only real signals can move power, in a circuit or as an EM field, so that continuous real power must represent the magnitude, instant to instant, of a series of complex number instances, each such magnitude defined as the product of a compex number and its complex conjugate. Amplify as desired. Rinse and repeat!

 

 

So, for all that arithmetic in the previous tab, SSB is, in short, just a kind of linear magnitude- plus phase-modulated carrier. In other words, SSB is a phasor. And you barely even need the magnitude.

 

If you ask an AI chatbot to define SSB, it will reply that SSB is a radio method to transmit information in half the bandwidth of any other type of modulation and using half the power. I know, I did this. This AI reply is first, a description and not a defintion, and second, is incorrect about half the power. Well, not bad for poorly-written software!

So starting like AI with a description, "Single Sideband" enables the minimum bandwidth radio signal containing all of an audio information's bandwidth. This is what the ordinary guy on the street wants to know. But you are not him.

The FCC calls it emission type J3E, and identifies it as a form of AM. Well, there is an AM-like component in a "fullsome" SSB signal. But like those late night TV commercials, wait, there's more! And how!

You may have noticed that several ultra-low-cost "QRP rigs" do a simulacrum of SSB in which no audio amplitude information is transmitted at all! This cheaper form of SSB is all, 100%, phase information. And many listeners could not hear much audio intelligibility difference on SSB between such a $200 QRP rig and a $5,000 super rig!*

So we reach an important digression before we actually answer the question! It turns out that phase information is much more important than amplitude information to the comprehension - the effective understanding - of a voice. Whether in radio or in plain air. And thus, a phase-modulated sideband is the essential thing to move up to a radio frequency. It may sound amazing, but you can do "sound" SSB at ultra-high audio frequencies easily understood by, say, your dog. That said, this should not be confused with phase discrimination in human hearing. The ear's ability to track a slow-moving phase (at any frequency) or even a small jump in phase at the origin of an acoustic (audio sound pressure) signal is virtually nil. And it must be added that this has nothing to do with "stereo" direction-sensing, which exploits differential phase and which two good human ears can be quite good at. But verbal information via the ears simply integrates the phase of the signal into a frequency (and, in parallel, into a spectrum) of arriving acoustic energy! This is mostly done in the acoustic lobe of the brain adjacent to the ears, but that's another story.

Of course, in real life audio, emphasis of expression does involve amplitude information of a low bandwidth, i.e. across verbal plossels, words or sequences of words. Music provides proof of concept. A phase/frequency-only "version" of SSB (that of the clock chip QRP variant) can't express this subtlety of sound, even though it can get words across. But any conventional SSB rig with a good microphone (and it's almost all mic) can certainly make that subtlety. Conversational quality SSB has some amplitude as well as the essential phase/frequency information in it. I.e., cheap QRP SSB is communications grade, maybe not strictly short of but lacking that visceral emotional punch! In the end, what you need in an SSB signal depends on what you, and your wallet, are willing to settle for!

Since we're digressing on audio quality, there's lately a fairly new thing, "extended" SSB, or eSSB, showing up in ham radio. You could call it "SSB HD" or maybe "HiFi SSB." This being just more audio bandwidth than conventional ham SSB, most folks thinking 2.8kHz of total modulation width (nominally 3kHz, but typically hard filtering out the bottom 200 Hz or so to minimize "bleed" into the other sideband). Alas, we might be seeing hints of "eSSB Wars" (of nominal 3kHz SSB vs this wider SSB) already being whispered among our community. This might, hopefully, go the way of the bad old SSB Wars and become a strictly non-confrontational thing among us hams. Like really old-fashioned AM has become. Kind of live and let live.

Now please note: the analog phase-modulated component in SSB is not the same as in either "classic" PM or FM! These latter two both involve integrals of the analog baseband phase, translating phase to frequency shortterm, whereas SSB uses the same phase information directly applied to the phase of reference carrier cosine, sans the integral. A really big difference. Especially when bandwidth is considered. PM and FM are double sideband signals, just like DSB with (-C) or without (-Suppressed C, or -SC) that (unmodulated) carrier. SSB is, of course, a single sideband signal. See the Mod 100 tab.

Also please note: SSB is used primarily for its (factor-of-2) bandwidth reduction relative to either DSB(-SC) or DSB(-C) aka AM, the latter of which would have the same information energy efficiency (and SNR!) as SSB or DSB(-SC) were it not for that unmodulated carrier component (that C), the culprit robbing from the signal's total energy budget a huge percentage 66.7%, and no less. The energy plowed into this unmodulated carrier "substrate" exists solely to avoid overmodulation and the nonlinear catastrophe which follows. Of course, if you value the "quieting" of the bandwidth between spoken words, a la commercial AM or FM radio, this unmodulated substrate can be of considerable listening value. More on that below wrt SW stations. No US ham has any realistic reason to ever use DSB-SC, and there never was a DSB-SC "movement" in ham radio. But US hams do have a legit arthouse or antique shop right, if you will, to use "historical" AM if it does not unnecessarily interfere (the FCC's key words) with other ham radio use of the bands. This is why AM "meet-ups, nets or groups" happen in local daytime when the MUF is lower than that of the HF band in use, and usually at the extreme top end of that band, the traditional territory of international Shortwave (SW) Broadcast above about 30m. Now you ask another question: so why do high power HF SW stations all over the world still use this archaic DSB(-C), especially when the band is well inside the MUF, like early mornings? Simple. First, because they can, having the approval of the ITU who treasure their licensing fees and political comity. But second, and likely more importantly, because the receiver for a full DSB-C signal is the simplest of all, an envelope detector requiring no precise tuning and almost no circuitry (or software) to extract the audio. So if you're trying to reach listeners at all levels of receive capability, go DSB-C, and not DSB-SC, and suffer the ~ 7 dB power inefficiency. You just make it up with huge transmitters and antennas. At government expense, I promise you. Just like VOA does it.

Back to SSB. You cannot improve on its minimum bandwidth (for any fixed analog information content)! If you attempt to gain back SNR using a "twisted" modulation like FM, or even better, pre-emphasis FM, or PM, you will totally lose the bandwidth contest due to a rapidly expanding RF bandwidth resulting from the several significant orders of Bessel functions of the modulated carrier that's the essential characteristic of these latter two formats.

If you attempt to improve on SSB's minimal bandwidth, but still - and this is important - insisting on retaining the full analog "fidelity" of the original audio source, by using a digital modulation (even forward error correction encoded, i.e. with FEC), you cannot do it at a very low SNR. This has to do with the dichotomy of bandwidth-limited versus energy-limited signaling domains. However, and this is a key point, digital modulation can dramatically exploit the compression of the original audio signal's digitized (sampled) bandwidth, by applying "source coding," but always incurring a degree of "rate distortion," according to information theory. For high fidelity buffs, this describes well the difference between MP-3 audio and so-called "lossless digital audio" (which is really just very low loss, meaning a loss indistinguishable to the human ear for a given bandwidth). Source coding is perhaps the lesser known of the two basic principles in information theory, the other, being the limit known as channel capacity, but that's another story. State-of-art digital audio compression can lead to a significant bandwidth reduction, exploited nowadays everywhere, and quite noticeably in US consumer "HD," which are digital subcarriers astride conventional broadcast AM and especially FM. Individual ears may or not may agree that this distortion, a reduction in fidelity compared with the fullest hi-fi analog, is worth the narrower signal (and so more bandwidth efficient, as long as the received digital bit energy remains large enough to avoid catastrophic FEC decoding failure, a loss unique to digital transmission). But if they do agree, and they're hams, they become aware of FreeDV (free digital voice), a ham radio effort (mostly for HF) similar to the vocoder effort in the M17 Project (mostly for VHF/UHF). FreeDV compresses communication-grade voice audio into 1kHz total occupancy RF, a huge gain for occupancy and power usage, but also a pretty noticeable loss in fidelity, yet enthusiasts seem to love it. FreeDV uses FEC to help mitigate against QSB and "light" QRM decoding threshold failure. The viability of FreeDV in a typical fading HF channel, for a given transmitted signal power, is a subject complicated by this decoding threshold effect, but that is another story.

Bottom line: If one must have "luxury voice audio," yet occupying no more than 3kHz of RF, then it's good old SSB, which, with a good microphone, has been the ham radio standard of quality for over a half century now. To which, of course, the ESSB folks may disagree.

 

* Afficionados may complain that of course there's a difference in quality of fidelity. And there certainly could be in strong link conditions. But most of these super rigs have been "tuned" by their users for high audio compression, which gives them the most "punch" in tenuous conditions or DX pile-ups. When highly compressed, an SSB signal has very little transmitted amplitude variation, bordering on that of the phase-only "cheap" SSB. This is an odd case where really good cheap is more prized than luxury.

 

 

The four most utilized "linear" "analog" (i.e. not digital) modulations in ham radio are DSB (AM), SSB, FM and PM.

CW is not in that list. But it could be, as CW is easily realized, and likely by default the way it's done in high end rigs, by nicely-timed, shaped tone-burst baseband input to the SSB modulator of an RF signal. And CW can also be done as (shaped) pulse duration-coded modulation (we really don't have a nice acronym for it, except CW!), also not in the list. Such is the ancient way CW is done, strangely enough. But this is about SSB. See CW back at Home.

Using a simple single tone (a sinusoid) for the analog baseband, we can contrast these four in trigonometric math. Because all these modulations are linear in the math sense of the word, then real-world combinations of tones, i.e. ordinary audio, are just weighted sums of these single tone expressions, a Fourier series expansion (or "analysis") of the overall audio. So long as the system stays linear, showing everything in terms of one single tone is sufficient.

Now we encounter a subtlety. In math, if a function G handles two arguments, say f and h, as G[ a f + b h ]= a G[ f ] + b G[ h ]. The arguments must appear as a sum; if they appear as anything else, say as a product, that is a behavior in the nonlinear realm. Thus, applying a (voltage or power) gain G to, say, two tones f(t) and h(t), where the arguments vary with time t, of relative unboosted amplitudes a and b, the result is the same two in the same original proportion, just larger if G > 1. This is one way radio amplifiers are measured for linearity, the classic "two-tone test" where f(t) and h(t) are two closely-spaced tones at RF. But there's a rub with the term "linear" having nothing to do directly with the above expression. An alternative use of the term is what we mean when speaking of a linear analog modulation, or linear analog "information signal" source to be modulated onto a carrier, then transmitted as a radio signal. And that word "signal" is also daunting in its manifold meanings! But back to modulation and SSB.

"Linear modulations" are conventionally understood to be continuous in phase and magnitude. Continuous means that there are no abrupt changes in either the phase or the magnitude of the overall signal. The detail of this has to do with a signal being differentiable everywhere. See a text on calculus. Voice is an example. Because signals, voice or anything else, can be decomposed instant-by-instant by Fourier analysis into a sum of (many) weighted tones, the idea of continuity in phase and magnitude passes on to each of the tones. If a modulation or source (controlling a modulation) is of this type, it's called "linear."

Now the two uses of the term "linear" can get conflated. Experts might say that CW or some digital modulations, even AM, can operate with a non-linear amplifier. The latter can be true! Some older commercial AM transmitters used full power Class C amplification of the RF carrier, but applied the modulation onto the amplifier output circiuit's supply current line via a large transformer coupling the audio signal in at the same or nearly same power level.

Some modulations are "constant envelope" (CE), and some are not. FM and PM certainly are. AM is certainly not! In the digital world, unfiltered m-ary PSK and Continuous Phase FSK (CPFSK) are basically both CE. So is Minimum Phase Shift Keying (MSK). These are all used in ham radio, and they can all survive, if that's the word, non-linear, say Class C, amplification to the final output. And also FM and PM. Obviously, non-CE digital signals have, at least, amplitude variations and cannot survive. Modulations need to be "fit" to the equipment and spectrum to be occupied. Some digital ham waveforms can be used with Class C amps.

CW would seem to be another such CE, no? Well, not really. ON and OFF is not CE. CW must be "shaped" or else create key clicks, which are the keyed ON-OFF square-like wave (odd order) harmonics in the audio passband around fc. No one does that. Unacceptable! So why is a Class C power amplifier still a possibility for CW? It is because it's not really operated as "pure" Class C. It's "filtered" Class C using some technique to dampen the rise and fall times of the output, even a simple RC circuit on the grid of the final power tube. That's old school. Virtually all modern CW is tone-on-SSB with quite linear final amplifiers. Same for digital. See CW or DIG back at Home.

All this said, full-fledged SSB, operated as it should be with linear amplification up to the transmitted power level, can carry any real-valued signal known to man. You just supply the signal to it as an audio input. A digital example is FSK over an Audio interface to the rig, that being called AFSK. (But once at RF, it's just FSK again!) Because this input signal (to the SSB modulator) is real-valued, it is physical and can be implemented with a single modulated voltage or current source (driven by, say, a microphone or a computer sound card). So can AM! But with energy and bandwidth disadvantages. Same for FM or PM, with their energy and bandwidth efficiencies falling in between. SSB really is a most flexible, multipurpose, bandwidth efficient tool for transferring modulated information from one point to another.

 

 

This is the most uneconomical page in this entire website (at least, so I think).

 

The simplest PSD (power spectral density) view of SSB relative to the audio (Baseband) and unmodulated carrier (cosine at RF) is what everyone knows from the start, Figure 1. How to get it mathematically is less well known.

 

Figure 1, Relative Power Spectral Densities of a DSB, DSB-SC and SSB Signal, from "Audio" Baseband to LSB or USB

 

The content of Figure 1. is power, or more accurately power density (W/Hz). Every one of these six possibilities, including the baseband trace, and meaning (importantly!) the entire line trace inclusive of both positive and negative frequencies, can be generated and transmitted as a real signal. But the top baseband would only be - in a pragmatic sense - an audio signal, real enough to hear acoustically, although you could transmit an ultra-short-range electromagnetic field that could couple to a receiver, done all the time at AF by transformers, but it would not be "radio" by convention. The remaining five traces are all RF by convention.

The next important thing is that all six power spectra above are (mirror) symmetric around 0 Hz. This is critical. Only symmetric-around-0-Hz signals can be real-valued! Anything else can be either mathematical or a mutliplicity of signals expressed relatively, and hence complex-valued yet very useful, but asymmetric around 0 Hz. You cannot transmit an asymmetric-around-0-Hz signal as a power-carrying electromagnetic field. It violates physics.

Some readers may object by noting the Vestigial Sideband (VSB) signal used in early TV. Well, VSB transmitted over the air is exactly symmetric around 0 Hz, or it couldn't be transmitted, period. VSB is only weirdly asymmetric around an RF fc, but that's OK as will be noted for both flavors (and there are only two!) of SSB, LSB and USB.

So of course, the RF sidebands of the bottom-two traces, i.e. the SSB signals, are asymmetric around fc. But this isn't at all the same as asymmetric around 0 Hz, and suggests an introduction to the modulation theorem of Fourier analysis, which shows how you can do this either mathematically or physically in two quadrature phases, say an I and a 90-degree-different Q, of an underlying RF carrier. Engineers tend to see it as math, while non-engineers tend to see currents and voltages. But they are the same thing.

Fortunately for modern ham radio, the math can be expressed in digital sample sequences and realized in practical software, convertable to continuous electric signals. This is the "core" of digital radio. The software-defined radio (SDR) includes a higher set of "layers" above the digital radio (which is also part of the SDR) that adds other stuff beyond the signal math, like display and control among many others.

You can do this SSB math in all-real-valued trigonometry, but it gets messy fast. The all-real-valued approach is close to (really, exactly) the phasing method of generating SSB, known for at least 100 years now. (The popular phasing-plus-filtering out of the unwanted sideband method was dominant for over 50 years, but the all-phasing method, no filter, is "perfect" if you will, just not as convenient in circuitry.) But to understand what's going on, it's easier to use phasors representing the three complex-valued analytic signals of the baseband and the unmodulated carrier multiplied together to produce and a resulting SSB, the latter being a purely real value you can transmit as a EM field around an antenna. See Phasors and Analytic Signals elsewhere in this blog for an intro or a refresher.

 

 

Again, what the survivors say.

Note 1: Alas, the reader must hunt down their own copy of copyrighted articles.

So we live with a ruthlessly reduced summary here.

 

We must start with Carson. So many things start with Carson, except waves, codes, and CW, which start with Hertz, Cooke and Wheatstone, and Marconi. Carson's bandwidth rule is still applicable to most every analog modulation.

And again, it all happens in the USA, strangely enough. Love that German, the Brits and one Italian guy for the great stuff they did in early radio. But SSB, like Morse(-Vail) code, requires an Atlantic crossing!

Carson, at AT&T, did the math and filed for a patent on true SSB as a way to increase the number of channels of long-haul radio telephony circuits, which was granted in Britain but litigated in the US until the first one-way SSB transmission was actually made in January of 1923, New York to London. Regular SSB radiotelephony went commercial in 1927 at a going charge of about $1,500 (the equivalent 2022 USD) for a three-minute call. Parse that!

This tab cannot improve on an early history of SSB in a 1956 IRE article by Arthur Oswald! (Note 1). Carson figured it out and tried to patent it, and the (US) Navy tried it out too. But it was just too impractical a technology in 1915. It would have to wait three decades, even though hams tried doing it during the depths of the Great Depression, as reported in R/9 magazine in 1933-34 and in QST in October 1935. This seems incredible given the fast development pace in FM (Armstrong, alone) and television (Sarnoff, RCA) in that same period, but times were rough. Get in line for soup. Analog TV eventually started using "vestigial sideband," a distant relative of SSB, to trim channel bandwidth to 6MHz to get more channels shoe-horned in to the total TV allotment. Then WW II ended, and hams got back on the air (they were totally off from December 1941 to late 1945). And did their ham thing, experimenting fast, making up for lost time. Late to the game, the inimitable (take that loosely) Gen. Curtis LeMay, K3JUY/K4RFA/W6EZV (can you believe, they changed call signs as they moved their household), tried out SSB in a couple 1956 test flights and declared it the future of US Air Force voice comms.

Meanwhile, hams restarted playing with SSB right off. This tab also cannot improve on an SSB article by Oswald "Mike" Villard, W6QYT, in the January 1948 QST (you could search it). Villard was at Stanford University in radioscience and reported on, what seems to be, the first ham SSB "QSO" in September of 1947, between a 20W transmitter at the club station W6ZTE on campus and received by W6VQD, presumably somewhere off campus. Soon another "SSB QSO" using a higher power transmitter at the Stanford club's W6YX, also on campus, connected with W0NWF, as recounted in a January 2003 QST article by Gil McElroy, VE3PKD (if you're in ARRL, you can just download it).

The old articles cited here generally called SSB "s.s.s.c." for single-sideband, suppressed-carrier. This awkward abbreviation was an anachronism even for its time. Mathematically, SSB arises unmodulated-carrier-free, nothing to suppress in the first place! And the unused, "other" sideband disappears completely in the math as well, nothing physical to get rid of, also in the first place. Carson possibly knew all that in 1914, even if all he had to work with were RF filters for literally stamping out the unwanted signal components. But the practical method of generating SSB by 1956 was phased suppression of the unmodulated center, the balanced mixer approach, tightly following the math, then brute-force filtering of the unwanted sideband. It was what the technology of the times was friendly to. The double-balanced mixer approach, the unwanted sideband killed off filter-free, supported pure math-like generation of SSB, but the (unwanted) sideband filter method, apparently cheaper, would hold for another 40 years!

In any event, by 1956 Collins Radio, always close to the USAF, decided to ditch all their AM product for SSB, in the government and ham gear markets, and the rest, as they say, is history.

Soon would follow many manufacturers, from a little firm in Benson AZ (Swan) to a radio start-up in Tokyo (Yaesu). Nearly everyone using phone below 30 MHz would be on SSB within a decade, that is, after the SSB Wars.

 

 

Definitively known as the Single-Sideband Wars, or Sideband Wars, or just SSB Wars. Take your pick.

Note 1: Alas, the reader must hunt down their own copy of copyrighted articles.

So we continue to live with a ruthlessly reduced summary here, beautifully expanded in those articles.

 

Single sideband, emission type J3E, just went by "sideband" by the early '60s. And even now, we suppose.

Picking up with that January 2003 QST article by Gil McElroy, VE3PKD, titled Amateur Radio and the Rise of SSB, you get the picture. SSB was fanning out across the bands, and AM was facing a formidably efficient competitor for the available bandwidth and chasing DX. Occupancy bandwidth was mostly the issue, not even the roughly 4.8dB power advantage of SSB over DSB(-C)*. In good solar years, the late '50s' Cycle 19 being the best in living memory, the HF bands were heavily occupied. There may also have been more active HF hams back then, possibly. The combination of all this, where SSB found itself cheek-to-jowl with AM, was unpleasant, for both kinds of enthusiasts, since neither wanted interference from the other. The "double necessary bandwidth" of AM vs SSB was the first shot fired over the transom by the Sideband guys. The AM guys fought back shooting at the "ducks," as in "quacking voices," when they bothered to turn their BFOs on at all. SSB just didn't "sound right" to them - was it possibly the smoothly constant "quieting" of a nice strong AM signal with that "carrier" versus a splashing background of SSB on fast AGC? In any event, the AM guys called it a cartoon character voice quality. Yeah, you know which character.

All this began in the mid '50s, and didn't cool down. Maybe the Cold War had a psychological role in all this. It went on long enough, with ups and downs, that the "war" element of it came to be "wars" in the lingo. Mere war wasn't supposed to last this long.

To get a feeling for what emerged, there was apparently, eventually, almost unbelievably, a letter to QST magazine in 1963 asking ARRL to "get on the ball and ask FCC to give the a.m. boys six months to go s.s.b." That was a siren call to action for the AM crowd, and reflected a sad polarization of what was once a quite congenial ham community. AM op(erator)s were now a shrinking minority, literally under an occupancy siege within the walls of the HF bands. Worse, the FCC actually considered doing exactly that. SSB seemed to hold the winning hand to force unconditional surrender. But fortunately, a compromise was reached. Obviously, we still have AM today, within the current Part 97 "necessary bandwidth" and "no unnecessary interference" rules, and FCC's formal recognition of emission type A3E (AM, or DSB-C). This roiling decade of ham history is a subject for more future research by your reporter here.

 

* Two-thirds of any straight AM signal is unmodulated carrier (keeping the signal linear in the audio range). So only one-third is useful information, that's the ~ 4.78dB "gain" offered by SSB. Some folks challenge this, saying the deleted unwanted sideband of SSB bestows another ~ 2.2dB of gain; not so! That extra gain simply doesn't exist as a fact, because a coherent AM detector (vice the simpler, non-coherent envelope detector) can recover both AM sidebands with the same efficiency as a product detector (used for SSB). It is true that the envelope detector loses about 2dB of SNR capability relative to the coherent amplitude detector, but that's a choice of the receiver designer, not an intrinsic limitation of the modulated transmit signal.

It is SNR that determines performance in radio. For non-coherent signals, this is just signal power over noise power, and the 2.2dB challengers win. But for coherent signals, this is signal energy over noise power density, or Es/No (where No doesn't change over the increased bandwidth!), and the challengers are set straight. Just say'in!

 

 

It's virtually all digital. Even in software (or firmware). Except, of course, for the boat anchor gang.*

It's simple to generate SSB in one multiply at a final RF frequency. (See SSB101.) That isn't how it's usually done, though. Because half of that multiply is a complex-valued sinusoid at the final RF or IF frequency. This is still a tiny bit challenging in floating point math today, though easily done by the fastest chips up to about 30MHz. But things get hotter and $$$'er further up. What usually happens is that the SSB gets generated at or near baseband, then heterodyned up, that even digitally, at least at HF. There are other ways as well. Like "direct" SSB via a clock generator chip (with a compromise, but cheap!) or via a direct digital synthesizer (DDS) with magnitude and phase controls for a "full SSB," the way to go in this ham's mind.

Because it's software, or an FPGA at very worst, even QRP rigs do SSB. Add a PC interface, and you can "house" virtually any modulation on an SSB. It's the go-to carrier. SSB is "all carrier," just like your best burger is "all beef." (Howls from the peanut gallery on 40!) And so, of course, AM can ride it. You can even put FM (also a double sideband signal) onto an SSB, although the SSB bandwidth must be wide enough to accommodate the much larger modulation index (typically 3, or three times that of AM) of "narrowband" FM, the sort we use in ham radio and old-fashioned land mobile radio (think those "police radio scanners") before the Digital Age. In any event, SSB is the ultimate RF "platform" (to avoid calling it a carrier) for baseband modulations.**

Demodulation is similarly all DSP today. It's filter, sample and off you go. Analog downconvert from RF is optional (but helpful). You never see an explicit BFO, as it's always implicit in SSB. The difference from AM is the detector. A product detector (with that BFO tone built in) rather than an envelope detector. (And here, I am ignoring coherent AM detection, which you can do with up to a few dB improvement over envelope-only, but only if the radio band's coherence time supports it!) Direct sampling receivers are the cheap path. But now we are now off our subject!***

 

* I've met otherwise reasonable guys who have more than 30 rigs laying around the shack. (OK, half of those are cheap handhelds. But the other half!) A Collins S-Line (and I do think the "S" was for "Sideband") box weighs 16 lbs. It takes two or even three of these boxes (RX+TX+PS/Speaker) to make a QSO. But it's the guys with the earlier, heavier boxes, some going back to the '40s, a typical Collins 75A-3 receiver weighing in at 50 lbs. (Hey, that's what I used shipboard in 1969! A great receiver.) So what has this to do with SSB? SSB is much lighter, even the valve stuff. In transmitters, this is at least because of the 7dB power advantage of SSB over AM for voice. (See SSB101.)

 

** A "non-centered" signal describes any modulation energy which resides to one side or the other, but not both symmetrically, of what could be called a "center," unmodulated RF carrier frequency. (This center frequency was denoted as "the carrier" in ancient "AM days"). That central carrier simply disappears in the pure math of SSB, as the modulation's total information (which "rides the SSB,") goes to zero at 0 Hz offset relative to a "reference frequency," which would be at the same location as for the AM unmodulated carrier component. The vestigial sideband, used in ancient analog TV, is a weird case that fits neither a symmetric spectrum or a SSB spectrum, and is fortunately obsolete worldwide.

A "centered" modulation is a totally different creature. Non-SSB-implemented CW (direct key the carrier) is a centered modulation. Centered spectra are common in data-only signaling, where it's unnecessary for a carrier's amplitude and phase to "track" any variation in a continuously-varying baseband spectrum, e.g. that of an analog voice. Centered modulations are standard in virtually all terrestrial LOS, satellite and space digital communication links, where analog information, like that of voice, are mere streams, or more likely IP packets, of bytes in the data.

 

*** Direct receivers ditch the mixer, but have other needs, mostly being enough dynamic range. Even with super RF band filters, an A/D converter needs protection from really big signals nearby in-band. The higher the ENOB, the more the protection needed. To get that in direct sampling means costly converter chips, so they compromise and offer less DR than the best hybrid conversion receivers. I'll take hybrid any time, even though that mixer and its "roofing" filter cost $$$ too. They just perform better in a clutch situation. In any event, either a hybrid or a direct sampler detects SSB the same way. In the end, you have to get the other (missing) sideband into play in order to hear it in a speaker! Real audio is always a real, symmetric signal spectrum. Try to tell your ears otherwise!