Amateur radio began with damped-waveform transmissions: Spark gap-produced radio waves traveling through the aether. As the art developed, we moved towards continuous wave, pure DC transmissions: CW transmissions that continue to this day. They are marrowband, spectrally efficient, and offer one of the best modes to use during poor band conditions or when low power levels are utilized.
The art continued to develope as amateurs began to modulate the CW waveforms to generate amplitude modulation - the era of voice transmission began. In the process of AM, the CW waveform is superimposed with the audio intelligence - a miltiplicative process. The resultant transmitted signal consists of a carrier term, and a pair of sidebands (for a sinewave modulation term); each sideband is removed from the carrier by the modulating frequency. So, if you generate a carrier at 7290kHz, you will also generate a pair of sidebands at 7289kHz and 7291kHz if the modulation consists of a 1kHz sinewave. Furthermore, if the envelope is modulated to +/- 100%, the audio sidebands wille ach be 6dB below the carrier level indicating that each carries 25% of the power conveyed by the carrier itself. The power id additive; the carrier term is constant and all the power comes from the modulator (the modulation process). Many may remember that the required modulation power needed to fully modulate a carrier is 50% of the transmitter power. A 100 watt transmitter will require 50 watts of modulator power and half of that power will end up in each sideband: 25 watts. It is interesting to note that design of an AM transmitter is not quite that easy: This is average power but the envlelope of an AM transmitter will double in voltage at the positive peak. The peak power required is, therefore, 4 times the carrier power: 400 watts PEP. (Power is calculated by taking the squre of the voltage term times the resistance of the load - the output impendance is constant, so double the envelope voltage is 4 times the envelope power.)
During the late 1940's experimentation began with single sideband transmission of voice. This grew rapidly in the 1950's when Wes Shumm, at Central Electronics, proved that 10 or 20 watts was more efficient than 100 watts of AM; something not lost on Art Collins. This started the SSB shift that has been retained as the primary voice mode in ham radio even today. The reasoning is simple: Remove the carrier - it is simply a bias term for the detector and can be reinserted in the receiver anyway - as well as one of the sidebands of modulation since they are symetric and identical, thereby saving 75% of the power which is, in typical use, wasted. To put this in simple math terms, at 100% modulation of AM, the carrier power is 100 watts and the total sideband energy is 50 watts; a total of 150 watts (for a 100 watt transmitter). But, only 25 watts is needed to convey the information, so, if all the power could be converted into intelligence, the gain would be -10log(25/100); 7.8dB! That is also why linear amplifiers are not very efficient for AM use; so much of the power is actually wasted. Add to that the fact that only 1/2 the bandwidth is required to gather all the intelligence from a SSB transmission, which is another 3dB gain (10log(6kHz/3kHz) and you have an 11dB advantage for using the same total transmitted power fr SSB vs AM.
In commercial broadcasting, there was considerable dicussion and controversy over the use of SSB, DSB, and various detector styles. Two individuals, Costas and Kahn, argued in IRE proceedings over the advantages of each. I will not go into the depths of their debate but Kahn suggested that AM broadcasters could essentially reduce their bandwidth to 1/2 the normal bandwidth for AM (for 10kHz audio transmissions, 20kHz RF bandwidth is needed) by producing a compatible single sideband, with full carrier, signal. Many of us are familiar with AME - Amplitude Modulation Equivilant - that Collins included as a mode in the KWS-1 and other transmitters. Motorola did the same with their HF SSB radios. In this mode, a carrier is simpy reinserted into the transmitted signal to produce a bias term for diode detectors. But, the signal is ctually quite distorted on an envleope detector due to "quadrature" distortion. SSB is really a unique case where inphase and quadrature terms are equal, but sperated in phase of the audio term by 90 degrees, thus producing a single term rather than 2 sideband terms. Take a look at an old ARRL handbook and you will see examples of the "phasing method" of SSB generation. The quadrature term in SSB, with or without a carrier added, causes distortion to an envelope detector. Kahn developed a method by which the envelope remained compatible; additional spectral terms were added to force this compatibility. Kahn has developed several methods to do this over the years. The first was a simple AM/ linear PM system with audio phase shifters that would supress the primary sideband on one side of the carrier. He then modified the system to include a generated second harminic term that would cancel the second harmonic term on the undesired sideband and augment it on the desired side. Later, to the best of my analysis, he modified hiws system further to produce a modified arcsine, rather than linear, phase modulation term. I will go into the mathematical details of this elsewhere. Suffice it to say that PM, arcsine and arctangent modulation terms are very close to each other at small angles but have significant differences as modulation levels increase. The Kahn CSSB (Compatible Single Sideband) system was very successful and found use interntationally. Later, he modified the system to convey 2 channels of information to produce AM stereo. He currently offers a system called "Powerside" that produces assymetrical sidebands with reported advantages for AM broadcasters.
I started to consider similiar work for amateur radio. First, if the receiver bandwidth can be halved, there is a 3dB improvement in signal to noise ratio righ off the bat. In AM broadcasting, most radios are electronically tuned on 10kHz channel increments; we can't take advantage of this benifit. But in ham radio, we have continuous coverage, so we can use this improvement possibility. Next, if a portion of the energy from one sideband is moved to augment, or exalt, the other, that power is also recovered making the overal signal to noise ratio better. In the system I describe, about 2.1dB of power is removed from the carrier and redistributed to desired sideband. This energy cancels one sideband and causes a shift to the desired sideband. In told, about 5dB of energy is recovered. Added to a 3dB reduction in demodulation bandwidth, an 8dB improvement can be enjoyed over regular AM for the same transmitter power level. This is only 3dB off from SSB and we still get to listen to the wonderful audio quality of AM rather than the "communication-like" nature of SSB. An adapter is required to generate the signal; I will be showing these soon. All one needs to do to get the gains on the receiver side is switch from a 6kHz receiver bandwidth (or even higher) to a 3kHz filterbandwidth. Smaller filter bandwidths can also be used. Next, you simply tune the radio to favor the sideband that has been exalted. For example, if the tranmsission is UXSB (Upper Exalted Sideband - a term I have coined for this system) with a carrier at 7290kHz, tune the radio to 7292kHz. This will leave 500Hz below the carrier and 2.5kHz above the carrier. One can actually shift even higher by a bit, but the tuning is donme by ear - you will find that it will be quite natural to just tune for the loudest signal with teh best audio bandwidth and quality. The resaon to leave a small amount below the carrier in teh IF bandwidth is two-fold: It makes sure that the carrier is unattenuated and it also leaves room for the lowest freqeuncy modulation which is sent as a vestigiale sideband. In voiced energy, about 1/2 the power is below between 200 and 300Hz. In my system, all energy below 340Hz is sent as double sideband (conventional) AM. It is the higher freqeuncy energy that is sent as compatible single sideband.
My system differs from the Kahn system in several ways. First, Kahn developed his system to be properly demodulated by a special recevier that provides high performance detection of the envelope and phase information to provide high quality AM stereo. In my system, the modulation is modified to favor modulation depth at the expense of dual channel possibilties. I only care to convey one channel, simple to transmit and add to a normal AM transmitter, and retain compatibility. Furthermore, Kahn reduced high frequency information to teh sideband cancellation circuits to reduce occupied bandwidth. this was done sine music is oftern sent, using high amoutns of preemphasis, on the AM broadcast band. In voiced information liek hams use, the energy, even with pre-emphasis, is much lower at high frequencies. Therefore, the SSB effect can be retained at hgih frequencies; particularly if low frequency energy is conveyed as a double sideband term. In this way, the system I propose is much like a VSB system. Again, I term it Exalted Sideband transmission: UXSB and LXSB for Upper and Lower Exalted Sideband transmissions. It is also interesting to note that the Russians proposed a CSSB system in the 1950s that took the concept to the limit. They proposed a log-based PM term that would generate only a second order compatilbitily term. In fact, at 50% modulation, the carrier and first order modulation terms were identical in amplitude and a second order term, down 12dB, was produced. At 100% modulation of teh envelope, the carrier was down 6dB from the first order sideband and the second order compatibility sideband was also down 6dB: It looked like a 100% modulation AM signal that simply shifted over to one side or the other by the moduation frequency! As cool as that was, it is a very difficult signal to generate and has a hard limit at 100% modulation, that, if exceeded, will produce a lot of splatter much like huge amounts of overmodulation and clipping. Kah had the right idea for CSSB for broadcating and I think my system is the mix for amateur radio.
Some spectral images of a converted Collins KW-1 are shown. In each photo, the spectrum is shown at 7290kHz. The first phot shows the normal doble sideband AM output of the tranmsitter. Ideally, there would be a carrier one 2 sidebands. The transmitter was modulated with a 1kHz sineave to 95% (symmetric). The additional terms are caused by the approximatly 2.4% harmonic envelope distortion the transmitter produces as well as some incidental phase modulation, IPM, that the transmitter also produces. The blue line shows the -26dB level; below which the energy represents less than 5% of the total energy. The scales are 1kHz/div horizontal and 5dB/div vertical. No attempt was made to reduce the harmonic distortion but the adapter does have a special circuit that can be adjusted to cancel some of the IPM.
The fillowing images then indicate the spectrum produced by the Exalted Sideband transmission. Since energy falls off at higher frequencies, amateur radio occupied bandwidth rules are never even close to being exceeded by the use of the signal on the air. In fact, about 1/2 the energy falls below the 340Hz point at which normal AM is produced. Energy below 1kH represents about 35% of the total energy and above 1kHz represents only 15% of the modulation. It can easily be seen that at these levels, the compatibility sideband energy is actually below that produced by a typical AM transmitter.
More will be added shortly and I expect to demo the system at Dayton 2009.