Rob's web

House cleaning the Low-Frequency phone bands

Home - Techniek - Electronica - Radiotechniek - Radio amateur bladen - QST - House cleaning the Low-Frequency phone bands


Transmitting and receiving techniques for intelligible reduced-sideband operation.

Something can be done - as this article demonstrates - about the ever-increasing litter of spurious sidebands and heterodynes that daily are wasting valuable kilocycles in the lower phone bands. But it's going to take more than complaint over the air and at the club, or in letters to QST, to clear the mess. Required is adoption by all phone men of a live-and-let-live philosophy of operation, the exercise of a goodly chunk of gray matter, and finally and most important, pulling the switches on the rig and wielding a soldering iron for a few nights. Read on!

Interference in the low-frequency phone bands is something to marvel at or cuss about, depending on whether you're an onlooker or participant. But much of it can be avoided if we're willing to take a realistic viewpoint on amateur 'phone operation and adopt techniques that are designed for most effective communication. Broadcasting methods, either for reception or transmission, have no place in a communications picture.

The fact is that the phone bands are not being efficiently utilized. A lot of energy is being radiated, in the form of unnecessary sidebands, which serves no purpose other than adding to the general interference level. And the average operator seems to expect that he should receive signals, in a crowded 'phone band, with no more effort than a BCL puts into tuning in his favorite program.

Let's take the transmitting situation first. Although maximum efficiency may not be achieved until we have single-sideband carrierless transmission, that system is complicated and difficult, not readily adaptable to amateur needs, and there is some question as to whether it is practically usable at all under the conditions of amateur operation. It is beyond consideration at present. In the meantime, there is every reason for confining sidebands to those that are absolutely necessary for intelligible voice communication. Every reason, that is, if we really want to improve con ditions in the 'phone bands. If "quality" is more important than communication, and if showing the world what a swell job of splashing and splattering can be done is more important than show.. ing consideration for the other fellow, then there is no reason at all for keeping sidebands down. There is then also no reason for complaining about interference.

The necessary channel width

Whenever the question of limiting the audio bandwidth - and, consequently, the transmitted channel width - comes up, the argument is sure to be advanced that the higher-frequency side-bands are relatively small in amplitude and do not cause much interference. Whether or not there is justice in this contention depends upon where you draw the line between "high" and "low" audio frequencies. We propose drawing that line just as far down the scale as it is possible to go and still have a high percentage of intelligibility. What the maximum necessary audio frequency is is not open to much question, actually; the facts have been determined in long and intensive studies by people to whom every cycle means money. No one can argue with real sincerity that the land-wire telephone does not do a good job of providing intelligible communication - with a high-frequency cut-off below 2500 cycles. (For "high fidelity" it goes as high as 2700!) The average ham phone has plenty of sideband energy above 2500 cycles.

If we do not need frequencies above 2500 cycles for intelligible voice communication then there is no justification for transmitting them. Amateur phones, as a general rule, go far above this limit. because it is easy to make speech amplifiers "flat," and present-day microphones have good response well up in the audio range. The only practicable way to hold the bandwidth down to the minimum necessary is to use a filter,(1) preferably a good low-pass job, in the speech amplifier. Such a filter will have to be a part of every modulation system if the best use is to be made of the congested phone bands.

It may surprise those who have fixed ideas about audio-frequency range and "quality" to find that a system with a 2500-cycle cut-off sounds quite good. It is no surprise at all to those who know what the ordinary broadcast receiver gives out in the way of an audio-frequency range. Whether or not the sidebands are present in the transmission, they are usually cut badly by the receiver's selectivity - and through listener preference - by the tone control. More important than the actual upper-frequency limit is the relationship between the high- and low-frequency cut-offs. As a rule of thumb, the most natural-r sounding result is obtained when the product of the maximum and minimum frequencies is about 500,000; in other words, if the high-frequency cut-off is at 2500 cycles the low-frequency cut-off should be at 200 cycles.(2) Extreme low-frequency response with a relatively-low high-frequency cut-off results in boominess, while the reverse gives a tinny effect. The ideal characteristic for amateur work, therefore, would appear to be about as shown in Fig. 1. A satisfactory characteristic at the low end is easy to achieve by choice of interstage coupling constants.

Fig 1
Fig. 1. An audio-frequency characteristic of this type provides ample intelligibility, but confines side-bands to the minimum necessary for good communication. Contrary to what might be expected, it also gives satisfactory "quality" - if not too greatly distorted by fiver selectivity.

Cleaning up the transmitter

The installation of a low-pass filter is only a beginning; it does not guarantee that the actual sidebands will not have frequency components above those passing through the filter. Consider the conventional transmitter line-up shown in block form in Fig. 2. For many reasons the filter probably will be installed in the speech amplifier; since the operating voltages and currents are low, components are inexpensive and easier to obtain or make than if the filter were installed, say, between the modulator and Class C amplifier. Any amplitude distortion that occurs in the system after the filter will result in the generation of harmonies that nullify the effects of the filter in confining the bandwidth. There are several places where such distortion can occur:

  1. In any subsequent speech-amplifier stage because of overloading or inherent nonlinearity.
  2. In the driver for the Class B amplifier, because of insufficient power output, poor regulation, and so on.
  3. In the Class B modulator itself, because of improper loading, overdriving, insufficient power capability, inadequate transformers, and poor commutation.
  4. In the Class C stage because of nonlinearity (insufficient excitation is the most probable cause) and/or overmodulation.

Fig 2
Fig. 2. Every point in the system offers an opportunity for distortion - and unnecessary sideband frequencies. Proper design and adjustment of a phone transmitter to take minimum spectrum space requires know-how and care.

In fact, if a phone transmitter is to take up the minimum in channel width, there can be no skimping in design anywhere along the line, and the utmost care has to be used to see that the proper operating conditions are achieved and maintained. The speech equipment, exclusive of the low-pass filter, should be designed for low harmonic distortion and should have ample reserve capacity to handle larger-than-normal signals without folding up. It is only with such an amplifier that the filter can really go to work.

There is one bit of extra insurance that can and should be used. "Building out" the modulation transformer will help eliminate any high-frequency harmonics that may have been generated after the filter. In its simplest form, such building out requires only one or two high-voltage mica condensers, which fortunately are quite inexpensive in surplus these days. It is possible, too, to build a high-level low-pass filter for installation between the modulator and modulated amplifier and thus catch the unneeded frequencies at the end of the audio chain.(3)

Despite the conscientious use of all precautions in the audio system, the good work can be undone completely if the Class C amplifier is not linear. And even a perfectly linear Class C stage will raise a rumpus throughout the band if it is over-modulated. Here is where the oscilloscope is practically indispensable. It is almost impossible to adjust a 'phone transmitter for proper operation without a 'scope, and there is no better modulation monitor. For checking either linearity or overmodulation the wedge pattern is best, which is fortunate because it is also the simplest type of pattern to get, requiring little more than the scope tube and its power supply.(4)

Confining the audio bandwidth to the minimum necessary, it should be observed, does not imply a "cheap" 'phone. Quite the reverse. It takes generous design, quality materials, and a thorough knowledge of how to set up and adjust a 'phone transmitter. Cutting the corners in design may do things to the audio characteristic, but they are usually not the right things. Combined with improper operating conditions, a poor audio system not only achieves a poor voice quality but also a channel much wider than is occupied by a good 'phone that actually handles much higher audio frequencies.

Under present conditions, spurious sidebands - by which we mean those resulting from harmonic distortion in the audio system, nonlinearity and overmodulation in the modulated r.f. amplifier - account for very much more of the existing unnecessary interference than do those legitimate sidebands above 2500 cycles that result from the use of flat audio systems. Until the spurious stuff is eliminated the improvement that can result from reducing the audio band is not going to be very marked. Cleaning up the transmitter and, once that is done, seeing that it is not overmodulated, is an obvious step that every 'phone operator is supposed to take anyway. Once it is actually done on a large scale, there is still something in the order of a 2-to-1 improvement in interference conditions to be realized by keeping the audio band to a minimum. The improvement is not automatic, however; in fact, it may be difficult to realize that it exists unless adequate receiving methods are used.

The receiver

It is not generally realized, we believe, how important receiver limitations are in establishing the interference level. Imagine, if you will, using a receiver having an effective bandwidth of 100 kc. for the reception of amateur 'phone on the lower frequencies. With a receiver like that it would hardly matter whether or not transmitters were confined to the minimum channel width necessary for good intelligibility; in fact, all the malpractices we've been talking about would be completely unnoticeable. Of course no one would think of using a receiver with such a bandwidth in the congested 'phone bands. The thought-is introduced simply to emphasize the truth that once transmitters have been brought to the point where they occupy no wider channel than is absolutely necessary for intelligible voice communication, the realization of the improvement in interference conditions that thereby becomes possible is entirely up to the receiver.

Except during those times of the day when there is relatively little congestion, existing receiving practices are entirely inadequate. Note that the word used is "practices," not "equipment." Modern communications receivers are capable of doing a great deal more in the way of separating stations than most owners ever suspect. The receiver question deserves examination from the beginning.

Let us assume that transmitters are all properly modulated and that no audio frequencies above 2500 cycles appear in the modulation. The radio channel width is therefore 5000 cycles, and there is no energy transmitted outside that channel. Theoretically, then, two transmitters spaced exactly 5 kc. apart. will create no mutual interference. The natural inference is that the receiver should have a passband of 5 kc., preferably in the shape of a selectivity curve with straight sides as indicated in Fig. 3. If such a selectivity curve could be attained, either station could be tuned in without the slightest interference from the other, no matter what the relative signal strengths. However, practical receivers do not have such straight-sided selectivity curves; the solid curve is typical of present-day performance. With such a curve there will always be some interference even though the actual transmissions do not overlap in the spectrum; whether or not the interference is serious depends upon the ratio cf the strength of the undesired signal to that of the desired station. It is apparent from the curve that if the undesired signal is around 15 db. stronger than the one actually tuned in (with 5-kc. separation between carriers) the desired and interfering signals will be about equal in strength.

Fig 3
Fig. 3. The extent to which a practical selectivity curve departs from the straight-sided ideal is a measure of the interference that exists because of receiver shortcomings. The two signals blocked-out along the baseline are entirely independent in the frequency spectrum, but interfere with each other in a practical receiver.

However, there is nothing in the book that says we have to make the receiver passband wide enough to accommodate both sets of sidebands on the received signal. One set - whether the upper or lower does not matter - together with the carrier is sufficient for distortion-free reception. The fact that both have to be transmitted, at least at the present time, is unfortunate but does not alter the fact that only one set is actually needed by the receiver.(5)

Despite the fact that accepting only the carrier and one sideband does not in the least affect what is transmitted, this type of reception contains the possibility of increasing by 33 per cent the number of stations that can operate simultaneously without interference in a given band. Continuing the assumption that each transmitter occupies a channel of 5 kc. and that the receiver has an ideal selectivity curve of the type shown in Fig. 3, also 5 kc. wide, a carrier-frequency separation of 5 kc. as shown in A of Fig. 4 will permit interference-free reception of any one of the three carriers shown. But if the receiver passband can be narrowed to only 2.5 kc. so that only the carrier and one sideband are accepted, alternate carrier-frequency spacing of 2.5 and 5 kc. permits getting one more carrier into the same total space. The interference-free channels are numbered 1 to 4, the regions where sidebands from adjacent carriers overlap, and where interference does occur, being shown cross-hatched. (This wasted space is necessitated by the fact that the transmissions are double-sideband; if the carrier and only one sideband were transmitted it would obviously be possible to have channels only 2.5 kc. wide and thereby actually double the effective width of the band.) Of course, in actual amateur 'phone work the idealized spacing between carriers does not occur, but on the average the ratio should still hold.

Fig 4
Fig. 4. Showing how the effective spectrum space is increased by single-sideband reception of double-sideband transmissions.

There are other possibilities that do not appear on the surface. Referring again to Fig. 3, imagine that the undesired signal is moved a bit to the left so that part of its lower sideband, but not the carrier, falls within the ideal 5-kc. passband. Those sidebands that do fall within the passband will cause some interference to the desired signal, of course, but the carrier with which those side-bands are associated does not get through the receiver. Essentially, modulation is a heterodyne process, and for intelligible speech to result, when the signal is detected, it is necessary that both the carrier and the sideband be present. (This is one of the complicating factors in single-sideband carrierless 'phone.) In our ideal receiver the interfering sidebands produce not intelligible speech but monkey chatter - bursts of noise that are detected by beating with the desired, not the undesired carrier. This is still interference, it is true, but it is considerably less troublesome than the voice interference that would result under the same conditions in a receiver having the typical selectivity curve shown in Fig. 3. Because in that case, since the undesired carrier gets through, there would be two voices going at once, and it is much harder to concentrate on one voice when two are talking than it is to pull a voice through random noise of equal intensity.

Interference suppression

Besides selection of the carrier and one side-band only, there is still another way in which the effects of interfering signals can be reduced. Fig. 5-A shows a state of affairs that is more common than not in amateur 'phone bands - two carriers so close in frequency that they both lie within the passband of the receiver. Using our idealized receiver, we have arbitrarily placed the desired and undesired carriers 2 kc. apart, the desired carrier being tuned in at the edge of the passband so that the whole of one set of sidebands can be accepted. The undesired carrier, being of the same amplitude as the desired one, will be detected in the usual way and the modulation on both signals will be heard, along with the heterodyne beat between the two carriers.

Fig 5
Fig. 5. "Exalted-carrier" reception, in which the desired carrier is amplified so much more than its own sidebands and interfering signals that certain forms of interference are suppressed in detection.

In the simultaneous linear detection of two modulated signals on different carrier frequencies, it is well known that the modulation on one of the signals will disappear if that signal is considerably weaker than the other. This effect, which arises in the mechanism of linear detection, is very helpful in producing an apparent increase in selectivity, in that, in amateur phone, it eliminates one of the two voices that otherwise would be heard. Although neither the amplitude of the heterodyne beat nor the amplitudes of the monkey-chatter beats between the undesired signal and the desired carrier are affected, the elimination of the normal modulation on the undesired signal is nevertheless a worth-while gain. But to obtain it, the desired carrier always must be considerably stronger than the undesired carrier.(6)

The obvious way to ensure that that condition always will be met is to shape the receiver's selectivity curve so that the desired carrier is amplified a great deal more than anything else coming into the receiver. A selectivity curve something like that indicated in Fig. 5-B is called for, a curve having a sharp "spike" extending perhaps 20 or 30 db. beyond the flat plateau that takes care of the sideband. In single-sideband reception the spike should be at one edge, as shown, but for double-sideband reception it should be placed in the center of the selectivity curve. The relatively-great amplification at the carrier frequency keeps the desired-carrier amplitude well above the amplitude that any other carriers not exactly on the peak can attain, at least in the majority of cases. Thus the modulation will be stripped off any unwanted signals that may come within the normal passband.

A consequence of this type of selectivity is that the desired sidebands are considerably attenuated with respect to the carrier amplitude, although it should be noted that if the curve otherwise has a flat top as shown in Fig. 5-B this sideband attenuation is the same throughout the audio range and hence does not affect the way the signal sounds. The only effect is that the audio output of the detector is reduced, a thing that easily can be overcome by incorporating additional gain in the following audio amplifier. It should be noted, too, that to cope with the ever-changing interference in a ham phone band it is quite necessary, in using single-sideband reception on a double-sideband signal, to be able to shift the "spike" on the selectivity curve from one edge to the other of the passband at will.

These receiving techniques offer a great deal in the way of reduction of otherwise unavoidable interference. When adopted, they also begin g make the reduction in transmitted sidebands, 1s-cussed earlier, pay off in better conditions in the 'phone bands. For when every means at hand is used to increase receiver selectivity, the unnecessary interfering sidebands that fall within the passband of a receiver set to a desired signal are causing just that much interference - monkey chatter - that might have been avoided.

The crystal filter

Although these reception methods have been discussed in terms of ideal receivers, they aren't something for the future. They can be used right now - by any amateur who owns a reasonably-good communications receiver having a crystal filter. The crystal filter provides almost exactly the type of performance we've been talking about - if used correctly. Like any other practical device, it falls short of the ideal in some respects, and the receiver that backs it up has its shortcomings too, but with existing equipment it is possible to go a long way toward taking advantage of everything on the horizon that promises to reduce interference.

Take the spike on the selectivity curve in Fig. 5-B, for example. The filter can provide that - if it's a sharp filter. The tendency to make crystal filters broader and broader in recent years is a concession to those who want their selectivity in tepid doses; we're not talking about the kind of selectivity that cuts the desired sidebands along with everything else, nor about "medium" or "half-sharp" selectivity. The typical sharp-filter characteristic shown in Fig. 6 shows how the crystal meets the requirements already outlined. It has a highly-peaked resonance curve, and if the phasing condenser is set off neutralization there will be a flat-response portion on one side (which is where the desired sidebands can be placed in single-sideband reception) and a deep notch on the other. The notch eliminates most of the other sideband and the normal selectivity curve of the receiver's i.f. section takes care of the remainder.

Fig 6
Fig. 6. Crystal-filter selectivity curves (without other i.f. selectivity). This figure, reproduced from page 31, March, 1947, QST, shows how the single-sideband characteristics of the filter can be varied by shifting the phasing notch. The single-sideband "effect can be en. hanced by use of an i.f. amplifier having the proper shape of selectivity curve, and with its center frequency placed in the middle of the flat portion of the crystal curve, thus putting the crystal peak at one edge of the system as a whole.

The notch can be shifted to the other side of the resonance peak, reversing the performance so that either sideband can be selected at will. Furthermore, the notch is movable so that a particular undesired carrier on the rejected-sideband side can be eliminated with practical completeness.

Tuning with a sharp crystal filter puts a heavy premium on frequency stability. We have yet to see a receiver (although admitting that we haven't had a chance to play with a few of the latest models) that will not "chirp" or "flutter" when the tuning is done with a really sharp crystal and the a.v.c. is on. The higher the signal frequency the more the instability shows up. Without a.v.c. things are considerably better, because there is no continually-varying gain to react on the frequency of the receiver's h.f. oscillator. Another disadvantage of a.v.c. is that between-signal noise is terrific, inasmuch as the audio level, when set for the modulation on a carrier, is much too high when there is no carrier present because then the receiver returns to something approximating normal operation.

Even though the elements of high-selectivity reception are available in the better present-day receivers, there is still ample room for improvement. The one most needed is a really high order of frequency stability. There is also need for steeper-sided i.f. selectivity curves and narrower i.f. bandwidths independent of the crystal filter, further to cut down adjacent-channel interference. Effective use of the filter demands plenty of audio gain. And an operating convenience, if not a real essential, would be a means for holding the audio output at a more or less constant level - a volume compressor, in other words. Oddly enough, there appears to be no need - if it's communication we have in mind - for bigger and better r.f. gain, bigger and better audio quality, bigger and better S-meters. Yet those seem to be the features on which most of us judge receivers.

As we said at the start, it's time we took a realistic viewpoint on amateur 'phone - if we want better communication.

Notes

  1. Formulas for the design of simple filters are given in the article, "Let's not overmodulate," by J. W. Smith and N. H. Hale, QST for November, 1946, page 25. Further details can be found in any textbook or handbook on communication engineering. The purpose of the present article is to outline broad considerations rather than exact methods.
  2. From a report, "Down to earth on 'High Fidelity,' " prepared by G. M. Nixon, C. A. Rackey and O. B. Hanson. The frequency range 200-2500 cycles is about the range reproduced by the small table-model b.c. receivers that probably represent the great majority of sets in use. Whatever the complaints about "quality" - and most people seem not to mind - on these sets, it is certainly not on account of lack of highs.
  3. W. W. Smith, "More on speech clipping," QST, March, 1947. To cut off at 2500 cycles, different filter constants are required (see Footnote 1) but similar construction can be used. Information on building out the modulation transformer also is to be found in The Radio Amateur's Handbook, 1947 edition, Chapter 14.
  4. See The Radio Amateur's Handbook for details of Class C amplifier adjustment and use of the oscilloscope for modulation checking.
  5. A receiver operating on this principle was described in the June, 1941, issue of QST, by J. L. A. McLaughlin, under the title, "The Selectable Single-Sideband Receiving System."
  6. This is a form of "exalted-carrier" reception (M. G. Crosby, "Exalted-Carrier Amplitude- and Phase-Modulation Reception." Proc. I.R.E., September, 1945). Exalting the carrier also reduces the distortion caused by selective fading of a modulated signal.

George Grammer, W1DF.