Independent Sideband

Check out: 4-Way Quadrature Modulation Decoder using the MC13020 Chip which deocdes ISB, C-ISB, QuAM, C-QuAM.

Check out: Discrete QuAM & ISB Document with circuits.

Independent sideband offers some definite advantages over any of the other systems. Since the channels are separated by frequency and not phase there is no loss of separation under adverse conditions. In a QuAM, C-QuAM®, FM or PM system information for both channels are contained in both sidebands and are separated by phase. One sideband contains the Left signal at +45° and the Right signal at -45° the other sideband has Left at -45° and Right +45°. During certain conditions that cause the sidebands to be received asymetrically whether it be phase or amplitude will cause loss of separarion. A wierd phasing effect called platform motion caused by multipath reception that is more pronounced in stereo mode is not much of an issue in an ISB systen. The phasing effects will play out acousticly in the listening enviornment but this acoustical matrixing is less pronounced than if the sideband information is matrixed during detection. Selective fading, or multipath, where one of the sidebands fades more than the other will only effect that channel that is associated with that sideband and will appear as reduced levels and equilization effects. Adjacent channel interference will appear only in the channel that is associated with the sideband next to the adjacent signal. During severe interference of one the sidebands and not the other the receiver can have a switch to select either the upper or lower sideband to reject that sideband which has the most interference. The stereo coverage is essentially the same as mono because of the frequency separation of the channels and nothing can cause the mixing of the two that would reduce channel separation. Two monophonic radios tuned to upper and lower sidebands can also be used to receive a stereo transmittion.

In broadcasting ISB can be used to tailor sidebands if interference to an adjacent station is an issue. One of the audio channels can be equalized to reduce the higher frequencies while the other channel can pass its full frequency response. During mono broadcasts transmitting in only one channel suppressing the sideband that is interfering with an adjacent channel allows more power to be shifted to the other sideband increasing coverage area without increasing total power output. Kahn developed this techique and markets a product called PowerSide which is a variation of his ISB AM stereo system. Using one of his AM stereo exciters little to no modification is necessary only that you send the mono signal to one of the channels and not the other.

In non ISB systems the phase is approaching its peak when the envlope is approaching its peaks whether it be positive or negative. This creates problems during reception because interference can shift the phase of the signal much closer to 90° than when the transmitted signal is at 0°. In a linear QuAM or ISB signal this is not an issue beacuse the carrier is not necessary to detect the signal with synchronous detection and is only there so the PLL will have something to lock to. A linear QuAM or ISB system can handle negative modulations greater than 100% with no problem. In fact they can handle a full 360° phase rotation and have no carrier at all much like the chroma signal in a color TV but would require a sync signal for the PLL to lock to i.e. the colorburst signal on the back porch of the horizontal sync pulse. When the envelope reaches -100% modulation the phase information is lost, as is the case in the Magnavox and Belar systems that require hard limiting to to recover L-R. This can create problems when the phase is at a deviation away from 0° and the envelope detector output goes to 0 in zero time causing a pop or click. An interfering signal can cause this to happen when the noise vector equals the signal vector and can cause a 180° phase rotation creating a transient spike that is detected with any form of angular detector. In C-QuAM® even though it is a modified form of QuAM it relies on the Env Det & I Det to obtain phase information to remove distortion. The particular nature of the decoding process makes it vulnerable to decode errors caused by this type of intrference. The signal is divided by the cosine of the anguar modulaion and as the phase approaches 90° 1/Cosθ approaches infinity and an interfering signal can push the already deviated phase towards 90° or even past it resulting in the maximum gain that the circuit can produce putting huge spikes in the L-R component.

In an ISB system the audio signals that are sent to the in phase and quadrature carrier signals are 90° out of phase. This naturally results in carrier phase approaching 0° when the carrier amplitude approaches ±100% modulation. In a non-linear ISB system like Kahn/Hazeltine it is of no consequence if phase information is lost during negative peaks of envelope modulation because the phase is approaching 0° thus greatly reducing if not eliminating the effects noise has on a 180° phase inversion. When a C-QuAM® decoder is used to decode a ISB signal the cosine corrector circuit does not behave as irradically like it does on a C-QuAM® broadcast during interference. If the pilot detector circuit on the MC13020 as enough gain it will trigger to the Kahn pilot signal and switch to stereo. Upon listening you will notice that even during moderate interference conditions an ISB signal will not have a lot of pops and clicks as a C-QuAM® signal will as the cosine corrector circuit is not pushed to the extreme limits on negative modulation peaks because the phase is near 0° on the broadcasted signal and not at or greater than 45° for a C-QuAM® broadcast during negative modulation peaks.

The Kahn system uses a synchronous detector for L-R just like QuAM or C-QuAM® but the method for removing distortion only involves information from the envelope and not phase as in C-QuAM®. Their inverse modulator only partially demodulates the envelope by increasing it by +6db on negative peaks and decreasing it by -3.5db on positive peaks and no phase term enters into the equation other than that an interfering signal will add to the envelope but will have little impact on the inverse modulator. The Kahn system does tend to preserve the synchronously detected quadrature signal much closer to what a regular QuAM receiver would do than a C-QuAM® receiver during negative modulation peaks.

When generating a linear ISB signal using the phasing method it's easy to demonstrate through vector addition certain advantages over regular QuAM when it comes to regular stereo vs mono peak envelope power. During a regular mono transmittion the L+R envelope modulation can go from -100% to +125% peak envelope power but when you add L-R into the equation with a phase deviation of ±45° the PEP will have to go to +218% modulation to maintain the same monophonic loudness. This is because the L-R channel peaks at the same time that the L+R does. This puts the phase peaks at both the +125% and -100% envelope peaks thus requiring extra headroom for stereo transmittion. A regular QuAM signal would have to reduce its mono PEP by 50% in order to prevent over modulation. An over easy compressor could be used to reduce both channel's gain by -3db so the envelope would not exceed the +125% modulation and this approach has pluses and minuses depending on how you look at it. The 3db compression may add more compression to an already compressed signal but would benefit program material that has a wide dynamic range like symphony. If it operated on phase deviation alone there would be times when the amplitude of the signal would be reduced and the envelope would not reach its PEP. This is one of the pluses of C-QuAM® modulation is that the L-R modulation does not increase the PEP because the signal is multiplied by the cosine of the angular modulation. Essentially the cosine of the phase deviation is the exact amount of gain reduction necessary to compensate for the increased enveolpe power during stereo transmittion. During single channel modulation with a 75% envelope modulation that channel is modulated to ±150% with a peak phase deviation of +23.2° and -71.6° thus giving a +6.5db modulation advantage for each channel over regular QuAM. The drawback to C-QuAM® is that the 71.6° phase deviation occurs on the negative modulation peaks just where the cosine corrector circuit is most vulnerable to interference. I doesn't take much of a vector of noise with the right phase to push the cosine corrector over the edge into a positive feedback latchup condition.

Since the L-R component of an ISB system is 90° out of phase with the L+R component the L-R phase peaks are at the 0% envelope modulation marks and the 0° phase marks occur at ~100% modulation of envelope modulation. This mostly eliminates the problem of having to sacrifice the PEP during mono recption for the extra headroom that a stereo QuAM broadcast would need. The phase deviation for single channel 75% modulation is symetrical with a ±48.6° deviation with a sawtooth like waveform centred over the 0° phase axis unlike the phase deviation of a QuAM signal that is asymetrical. The fact that the phase peaks are at the 0% modulation marks mean that both positive and negative phase peaks will add equally to the envelope and results in almost entirely in 2nd harmonic distortion for envelope detectors and probably is a factor in why Kahn only corrects for a 2nd harmonic term in L-R. The asymetrical characteristic of the phase of a QuAM signal also adds harmonic distortion to the envelope but are dispersed over a range of even harmonics 2nd, 4th, 6th, 8th...

This asymetry of QuAM modulation is so extreme that a PLL must lock onto the unlimited linear output of the L-R detector. If the signal was limited it would create a large phase phase error for a single channel that was 75% modulated. The area underneath the curve using 0° phase deviation as the dividing line differs by a ratio of 2.6:1 and would require a phase adjustment by the PLL so the area underneath the curve for both ± phase deviations would be equal. Under ISB the use of a limited signal for PLL locking may have an increased advantage over noise and would not have adverse effects since signal limiting would not effect the capability of the PLL to track the 0° phase mark of the carrier.

ISB is also insensitive to phase locking inaccuaties when it comes to separation during stereo reception. The channels will become phase shifted when the PLL is not locked at 0°. A phase shifting will occur for both channels and will play out acousticaly in the listening enviornment in which the L+R component of each channel will be 180° out of phase between the channels when the phase detector is off by ±90°. There will only be complete nulling of L+R information during a full 90° phase detector offset and when the listener is sitting in the exact center of the room when using speakers. Since there is no acoustical mixing during headphone listening there will be no L+R nulling under these conditions but the phase shifting of the individual signals may or may not be appearant to the listener since the ear is not necessarily sensitive to phase distortion. If the carrier rotates 90° in relation to the sidebands separation is not affected but the envelope of the signal would be L-R for mono reception. This makes stereo reception superior to mono when these types of distortion occurs.

The Kahn/Hazeltine ISB systen is rather complex in design at the transmitting end but is not really much more complex than C-QuAM® except for the added audio phase shift networks necessary to recover the individual sidebands. In studying the Kahn system it appears that several factors are involved to meet certain design criteria. An ISB model was chosen for all the obvious reasons. A Compatible envelope for monophonic receivers, ability to receive an ISB signal using two monophonic radios tuned to upper and lower sidebands, the simplicity in receiver design for stereo reception, minimum distortion in reception for both an AM stereo receiver & the two radio approach and good immunity from interference when decoding in an AM stereo receiver. There are probably other criteria that came into play but these are probably the most signifigent ones. It appears that the Kahn ISB system is a best fit approach to satisfy all these needs and then some. The generation and reception of the Kahn system is not mathematically the same so complete separation between the channels and complete distortion cancelation is not possible on an AM stereo receiver and definitly not when using two mono radios either but is a balanced system that gives the listener good stereo enjoyment for a variety of reception modes.

The Kahn system starts out by audio phase shifting L+R by -45° and L-R by +45°. The L+R portion is sent to the envelope modulator while L-R is processed through the a complex circuit that adds a second harmonic term for distortion cancellation. Both the Left and Right signals are processed separately at 0° phase shift through a constant gain frequency doubler to obtain the second harmonic component. They are then subtracted from each other to obtain the second harmonic term of L-R. This signal is then gain modulated with a rectified phase shifted L-R signal. It is then added to the phase shifted L-R signal and then sent to a linear phase modulator. Given that a linear ISB signal is just a QuAM signal that has its L-R audio phase shifted by 90° in relation to L+R you would think that this would be a better approach to generate the phase modulated term without having to resort to this complex means of 2nd harmonic distortion cancellation when using linear phase modulation. Since the Kahn stereo receiver uses a quadrature synchronous detector and partially demodulates the envelope to reduce distortion why not generate the signal so that the generation is the mathematical inverse of this reception method. You would get complete separation and complete distortion cancellation. It might be possible that using linear phase modulation may offer better compatibility with the two radio method for stereo reception though this and maybe other reasons is why Kahn uses linear phase modulation. The following paragraph desrcibes a Motorola patent where in phase and quadrature balanced modulators are used in such a way that would produce a phase modulation term that is more characteristic of linear phase modulation when L+R modulation is present maybe supporting the argument for more of a linear phase modulation scheme.

Motorola had a patent issued on Jan. 22, 1980, No. 4,185,171 for a compatible single sideband system that uses a modified quadrature signal to generate it. The drawing details show that the 90° phase shifted L-R audio is multiplied by 1+(L+R)/2 before it is sent to the quadrature modulator. 1+L+R is sent to the in phase modulator. By modulating L-R with a partial L+R component this tends to linearize the phase deviation of the non-linear phase modulation that characterizes QuAM modulation. The output of these two modulators is summed with the output of a third in phase L+R modulator that is used in a negative feedback loop with the inputs of the comparator being the envelope of the QuAM signal and 1+L+R. The net effect of the negative feedback is to remodulate the in phase channel to force the envelope to be equal to 1+L+R so the envelope will be compatible with envelope detectors. The patent also shows the method for decoding it which is pretty much the same to what Kahn uses. The envelope of the incoming signal is 50% demodulated with a inverse modulator and then detected by a quadrature detector for L-R and 1+L+R is detected with an envelope detector. The recovered L+R and L-R are sent to audio phase shift networks before they are matrixed into Left and Right. Why Motorola filed this patent when they had been promoting C-QuAM® for several years before that I wouldn't know ;-).

Although the Kahn system is a well balanced system for a variety of reasons it would not be my best approach for a non-linear envelope compatible ISB AM stereo system and I don't completely rule out the C-QuAM® approach either. I think C-QuAM® is a good approach for a non-linear AM stereo system, minus its one major shortfall of decoding errors during negative modulation peaks, and cosine correction is a natural approach to envelope compatibility. The best AM stereo system would be a linear ISB system using audio phasing and quadrature modulation. If the FCC hadn't got stuck on the idea that the envelope had to be equal to L+R because of the 25% or so quadrature distortion when modulated with test tones, a linear system would offer stereo separation for the full frequency range of L+R without the excess sideband splatter. The Kahn system has to curtail its L-R frequency response to 1/3 that of L+R and C-QuAM® to 1/2 in order to control excess splatter and meet FCC regulations. For the Kahn ISB system L-R frequency response, distortion and separation issues are the main deterants even though the overall performance of the system is well balanced. Kahn's latest exciter supposedly is mathematicaly the inverse of his decoding scheme so the distortion and separation issues are probably resolved.

A linear QuAM or a linear ISB system is the best model for a two channel system since they are both an AM signal that occupy the standard monophonic bandwidth of plus and minus of the maximum modulating frequency. Both are QuAM signals and the only difference betheen the two is that in the ISB system the L-R component is audio phase shifted by 90° in relation to the L+R component resulting each channel being single sideband. Since they are both natural AM waves any departure from this to obtain a compatible envelope should produce the minimal amount of spectrum spread as compared to the Magnavox and Belar systems which the angle modulation term contains no information pertaining to the in phase channel i.e. the phase of a QuAM signal has both L-R as well as 1+L+R information and the resulting envelope of a QuAM signal is close to 1+L+R. To a lesser extent the Kahn ISB system suffers from the same but since that system adds a second harmonic distortion correction term that originally appears in the composite modulation term and the signal characteristics of ISB i.e. during maximum and minimum points of envelope modulation ±100%, carrier phase is at 0° and during peak phase deviation the envelope modulation is at 0%, helps to tame some of this. It almost it appears like Kahn is trying to reproduce a linear ISB signal in that the envelope has been remodulated with 1+L+R in sort of a backwards kind of way. The most direct approach to this would be to be to generate a linear ISB signal by either the filter of phasing method and remodulate the envelope with 1+L+R. Of course the phasing method of this is just a ISB version of C-QuAM®. To test this on a current C-QuAM® transmitter where the audio preprocessing does not introduce any differental phase and/or amplitude variables preceding modulation, audio phase shift the program material by matrixing Left & Right to obtain L+R and L-R, audio phase shift L-R by +90° and then rematrix them in the usual way you would to obtain Left & Right. Left channel will contain Left@+45° & Right@-45° and the Right channel will contain Left@-45° & Right@+45°. If the phase shift is done digitally and stored back into a digital format like CD or Wav and the transmitter is tweaked properly you should get ~40db of sideband separation for the QuAM signal and less for the final C-QuAM® signal. You should have the fundamental modulating tone suppressed in one of the sidebands during single channel modulation. To receive it distortion free you would need a radio with a C-QuAM® decoder and an audio phase shift network that would reprocess the two channels into Left & Right signals.

The ISB version of C-QuAM® helps to overcome the one major flaw that C-QuAM® has during peaks of negative modulation. Since the phase is approaching 0° on negative modulation this does a great deal to help eliminate the pops and clicks associated with regular C-QuAM reception. The state of the cosine corrector circuit is much more predictable during negative envelope modulation peaks and can be controlled during interference conditions that would cause it to malfunction. With a variable limiting level on the control signal of the output of the error amp, controlled by the negative modulation of the envelope, would do a great deal to tame the irradic behavior of the cosine correction circuit. This protection circuit would have no ill effects on cosine correction if the level of limiting was always above the level of the cosine signal during normal operation with little or no interference present.

Phase deviations are signifigantly less in ISB mode as compared to QuAM mode. For a single channel that is equal to 75% envelope modulation the phase deviation for regular QuAM is +23.2° and -71.6° and for ISB is ±48.6° and is symetrical. At 87% modulation QuAM is +25° and -81.2° and ISB is ±60°. Since ISB's overall plus to minus peak phase deviation is pretty much equal to QuAM's but is centered around the 0° axis the cosine correction gain is much less generating less harmonics in which 1/Cos(71.6°)=~3 for QuAM and 1/cos(48.6°)=~1.5 for ISB having half the gain modulation of QuAM. Having the phase of ISB centered around the 0° axis resulting in the cosine correction term being equal for both positive and negative swings will result in predominantly 2nd harmonics on the correct side of the carrier more so than QuAM will.

The following table shows sideband levels for both C-QuAM® and C-ISB© at various modulation levels incremented by 3db. The following spectral analysis information was obtained by using the Fast Fourier Transform function in Excel. It may or may not match other approaches of spectral analysis for AM given limitations in Excel and the approach used. The spectral analysis for C-QuAM does match other published dada i.e Dec.'78 issue of PE and the same method was used to obtain the C-ISB© spectrum.

Sideband Levels In Relation to Carrier

Modulation Level C-QuAM®
Harmonic 25% 35%50%70%
Fundamental -15 -12 -9 -6
2nd -42 -36 -29 -23
4th -71 -60 -48 -38
5th -88 -74 -59 -47

Modulation Level C-ISB©
Harmonic25% 35% 50% 70%

In C-ISB© the fundamental sideband, there only being one as in SSB, is 3db greater than in C-QuAM® which has two. Having one sideband that is 3db greater than a DSB system which has two has a 2:1 power advantage over noise which is why SSB is very popular. This 3db gain in S/N helps to overcome 3db loss that is usually accompanied when most receivers switch into stereo. The total advantage of SSB is in suppressed carrier mode which has a 9db advantage over DSB with carrier resulting in an 8:1 power advantage over noise.

If you notice for C-ISB© that opposite sideband suppression of the fundamental decreases as modulation increases. This is because cosine correction has more of an amplitude reduction for the quadrature channel that it does for the in phase channel. The quadrature channel waveform gets squashed on both positive and negative peaks while the in phase channel gets squashed on positive peaks and elongated on negative peaks. This tends to add more odd than even harmonics to the quadrature channel and more even than odd harmonics to the in phase channel. These harmonics generated from the cosine modulation of the carrier and the funtamental also add to the fundamental on the opposite sideband reducing separation. The same thing is happening in C-QuAM® but is not apprearent since the fundamental exists on both sides of the carrier.

This asymetrial sideband effect that is a byproduct of cosine modulation of an ISB signal may initially appear as a disadvantage over regular C-QuAM® modulation but the overall spectral energy is less because cosine modulation is balanced for the positive and negative phase swings resulting in a peak gain modulation of 1.5 @ ±48.6° at 75% modulation versus C-QuAM® with a gain of 1.1 @ +23.2° & 3 @ -71.6° at the same modulation level. For C-ISB© only the second harmonic on the fundamental side is only slightly greater than it is for C-QuAM® and although the harmonics on the oppposite side of the fundamental decay at a slower rate than C-QuAM® they start at a lower level with the second harmonic being more than 12db less than that of C-QuAM®. Not until the 5th harmonic is the C-ISB© harmonic greater than C-QuAM® but only by 3db. At this point the harmonic level is 40db below the fundamental representing only 1% the amplitude of the fundamental which is not much of an issue.

Regardless of whether you use Kahn's method to develop the phase modulation term or C-QuAM® for an ISB signal the predominance of the second harmonic term seems to be the byproduct of the non-linear multiplicative action of forcing the phase of an ISB signal to carry an envelope of 1+L+R. All other harmonics are much less and is why Kahn only corrects for a second harmonic term to increase separation and reduce distortion. Since the C-ISB© approach is much less of a brute force action because the envelope of a linear ISB signal is close to 1+L+R, not much correction is necessary as compared to Kahn's approach where his method of using linear phase modulation that initially has no amplitude component and then amplitude modulates it with 1+L+R is definitely going to produce more harmonics. The cosine correction approach when applied to to a linear AM wave that contains phase modulation be it QuAM or ISB is going to produce the least amount of harmonics possible as conpaired to the Belar, Magnavox and Kahn systems. Cosine correction is one of the most natural ways of modifing a linear AM wave with phase modulation to obtain a compatible envelope for envelope detectors. Applying cosine correction to an ISB signal whose phase modulation is equally balanced for both postive and negative swings is going to reduce the amount of gain that cosine correction has to apply and better compatibility with synchronous detectors that do not have distortion removed through cosine correction. Likewise a linear ISB signal should have better compatibility with an envelope detector only introducing almost an entirely 2nd harmonic term which is already present in most program material and would help to overcome the poor high frequency response most monophonic AM radios have because of narrow bandwidth.

©2001 - J. S. Gilstrap

This document may be shared with anyone provided that the copyright notice stays with it. In fair use, quoting from this article for published documentation I require that my name referenced. Publishing in full in printed media will fall under different terms and must have my permission. Reference not required for casual quoting in internet discussion groups but would be appreciated. All trademarks i.e Belar®, Magnavox®, Motorola®, C-QuAM®, Kahn®, Harris®, Popular Electronics® and others are property of their respective owners.