.

von NORMAN H.CROWHURST

Here is a fair and impartial analysis of some of the methods proposed for stereo broadcasting. The author describes each, and comes up with some conclusions as to the probabilities of acceptance of any one of the systems within a reasonable time.

Maybe this is rushing in where "angels fear to tread", but it seems to me as if some angels have been stepping a little out of bounds. As a man whose background is almost entirely in audio, I thought problems in matrixing and FM were best handled by the radio men. But I have encountered so many contradictions about the audio end of this that I decided to take an interest. This is what I found.
.

First the technical aspects need clarifying. There are really two separate parts to the problem: the transmission and the audio. From the transmission viewpoint, the question is how we standardize subcarriers so we have two channels for stereo, and then, can we use one or more additional subcarriers on the same transmission for some other purpose?

Subcarriers can vary in allocation of the total modulation "percentage" and in location of the subcarrier's frequency, above the audio. All this is related to the possible separation between associated channels on the same FM carrier, service range, signal-to-noise ratio, and possibly the ease in obtaining satisfactorily low distortion.

On the audio end is the question of how stereo program is distributed between the two available channels allocated to it, the main modulation and one sub carrier. As far as the transmission problems are concerned, there is no basic connection.

The transmission channels do not care whether they are transmitting left and right, or L+R and L-R, or some other combination. There may be some argument about whether full fidelity (whatever that may mean) is required of the subcarrier channel, or whether a narrow band will do, but there is no essential connection with what the two channels are used for.

So statements that "matrixing robs the main channel more than straight left and right," or that one or the other way of impressing the audio has inherently higher noise level, are just bunk. What does affect the system noise level, frequency range, distortion, and so on, is the distribution of available transmission energy, whichever way the audio utilizes it.

Those are the only purely technical considerations. Then there are some economic ones - or maybe these should be called political! At present FM-multi-plexed subcarriers are being used for background music private services. As an economic fact, these services have enabled several FM stations to continue in business. These are not likely to be prepared to relinquish this income, unless stereo - or whatever else shows up - offers a more lucrative possibility.

Naturally, some high fidelity enthusiasts are appalled at the absorption of some of the "high fidelity" FM medium for such mundane (if not outrightly immoral) purposes. But we have to be practical. People who run FM stations must eat too. So there is an issue here as to whether it is possible to have a high-fidelity subcarrier as well as full-fidelity main carrier modulation and a low-fidelity subcarrier.

On this question one can get differing answers ranging form some who do not believe even a single subcarrier of any kind can be used without "robbing" the main channel in some way, to people who affirm that as many as three or four sub-carriers are quite possible, without any diminution of quality.

As far as I have been able to ascertain, no figures exist on performance with two subcarriers, although some experimental tranmissions have been made, and even tape recordings taken of them. More than two subcarriers have yet to be tried. The possibility is postulated solely on figuring. Performance details with only one subcarrier are far from technically complete in most instances.

To complicate the "political" aspect, arguments appear about what the second subcarrier is to be used for, and about the question of privacy or piracy.

Multiplex Services assured me they are not worried about the home user picking up a private service for non-commercial listening in his own living room. Their objection is that availability of multiplex adapters for home use means restaurants can easily buy and use the same sets to "steal" the program.

The answer to this is that you can also buy devices for pickup of telephone conversations, but this does not imply legalizing of wire tapping, infringement of privacy, or an official "nod" to blackmail by means of it. Criminal infringement, using legally marketed equipment, must still be cared for by proper police action and legal process.

Apparently realizing this is a weak case, some are arguing that there are more "legitimate" uses for subcarriers, such as for expanding educational facilities. Now, we are all in favor of anything that will help out the appalling shortage of classroom space and faculty time in the nation's colleges and schools. Probably nothing in this country is more vital at this time.

What surprises - and disgusts - me, is to hear a proponent of one system accusing the proponent of another of being opposed to such usage, when each of the systems involved in the argument is equally applicable (or inapplicable) to it.

In view of these attitudes, it takes quite an effort with a mud-scraper to get at the underlying facts. Let's take a quick look at the current proposals in the field.

.

This system has been explained adequately before. The essential feature of the system is the use of matrixing in the audio. The main modulation carries left plus right, the subcarrier left minus right. Matrixing is necessary on reception to get true stereo, which means an FM transmission using this system will not be strictly compatible with existing AM/FM (Anmerkung: Mono-) transmissions.

The original Crosby recommendation was for a subcarrier of 50kc/s modulated between 25 and 75kc/s (25kc/s deviation) and occupying half of the available main carrier modulation, which would be 37.5kc/s. For the unmodulated subcarrier, this is a modulation index of 0.75, varying between 1.5 and 0.5 for the maximum swing of the subcarrier frequency.

This can in turn be broken up into sidebands of sidebands, most of which is hypothesized on the basis of "100%" modulation over the range of audio frequencies, modulation index, or deviation ratio being referred to 100% modulation at the highest accepted audio frequency. But if the modulation index is 5 (the figure for "straight", unmulti-plexed FM) at 15,000 cps, 100 per cent modulation at 1000 cps will represent a modulation index of 15 times this, or 75. This requires about 100 significant sidebands to transmit 100 per cent modulation at 1000 cps, extending about 100 kc/s on each side of the carrier.

But practical program modulation is not as simple as this. One does not get 100 per cent modulation of any one frequency, but a peak excursion corresponding to 100 per cent when the instantaneous peaks of the component audio happen to add up in one direction.

All this has to be done twice over in multiplex: once in modulating the sub-carrier, and again in applying the composite audio modulation to the main carrier. Then it all has to be transferred through the various stages from the audio modulator to the antenna without further distortion, which imposes more rigorous requirements than previously necessary for FM transmission.

Crosby reports his system capable of quite low distortion figures, while others reporting on WBAI (New York City) experimental transmissions using the Crosby system find much higher distortion figures. There is the reason for the difference - system capability, vs. actual equipment performance. And this is not unique to the Crosby system. It has to be faced, in precisely similar fashion, by all of them.

While the Crosby system is intentionally a wide deviation system, there are no absolute figures in it. It is quite as possible (or impossible) to put in another subcarrier, regardless of whether or not sum-and-difference matrixing, which is Crosby's basic feature and claim, is used. Distortion, frequency response and dynamic range on the subcarrier are not affected by whether the signal it carries is a matrixed one or not.

By this designation I refer to the earlier proposal put forward by William S. Halstead. This was to use "straight" multiplexing without any matrixing: left channel on the main modulation with right on the subcarrier. To squeeze in two subcarriers (in reasonable "comfort"), the proposal was to restrict the right channel audio to a maximum frequency of 8.000 cps, on the theory (which was supposed to be backed by experimental evidence) that frequencies above this do not contribute to stereo, so it's enough if they come from the left only. (Fig. 2).

The argument for the system was that it did not interfere with the private service operation and that it is compatible with the present AM/FM stereo, by transmitting the right channel on both the subcarrier and AM.

The argument against it was, that it is not compatible, in the sense that FM receivers without adapters only receive the left, rather than a mixed or mono-phonic program. Also very few high fidelity people will accept the hypothesis about not needing anything above 8.000 cps on the right.

Everybody in the business has done some changing of position, and now Halstead is going along with what we will call, for the sake of distinction.

Developed by Richard W. Burden, this originally aimed at overcoming the objections to the Halstead system. To do this, it employs an alternative to matrixing. One might call it double matrixing. To the left audio is added L-R, to produce 2L-R for the main channel. The same component is subtracted from (or R-L added to) the right audio, to produce 2R-L for the subcarrier. (Fig. 3)

This means both channels (with the intent of covering the AM/FM situation) carry left and right in different proportions and phase. When reproduced over stereo speakers, the L-R from one speaker is expected to cancel acoustically with the R-L from the other, leaving just L and R to be heard.

A further suggestion for the system is to apply a further matrixing network to the receiver adapter to retrieve the true L and R electrically, which certainly makes better sense.
.

By this designation I refer to the system that has been demonstrated by RCA/NBC with the Perry Como show. It uses a different means to achieve an objective similar to that of the Burden system, allowing both channels to carry full program content, yet produce stereo when both are used together.

It is not proposed for multiplex, but since its compatibility problems are similar, but a method different from Burden's is used to overcome them, it is included here.

In this case the "Haas" precedence effect is used. The right audio is time delayed by a fraction of a second, attenuated slightly, and added to the left channel and vice versa. Thus both channels carry left and right in equal strength, but with time differences (Fig. 4). Because of these time differences, which are not enough to be noticeable on the individual channels, the correct sounds are associated with the loudspeakers that emit them first.

The four systems so far described are each basically different ways of handling the audio fed into the transmitter. Originally Crosby favored a wide-band sub-carrier and Halstead a narrow-band one, but these were not really esssential to either's method.

Actually, it would seem logical that Halstead's original system would benefit more from use of wide band, while a matrixed arrangement could better be adapted to a narrow channel for the difference signal.

But the Calbest system is predicated on use of a narrow band subcarrier. They argue (again on the basis of "tests") that frequencies above 3.500 cps are not necessary to stereo. So they propose to carry all frequencies above this on the main modulation and to restrict the subcarrier to frequencies below this.

To achieve this without making one of the stereo channels highly deficient in reproduction, they reinsert the highs at the receiver from the main channel. To explain this let's designate frequencies below 3.500 cps by the subscript m and those above by the subscript t. On the main channel they transmit the complete program Lm+Rm+Lt+Rt.

This is the full mixed program, and in this respect is the same as the Crosby. It is the sub-carrier content that is different. On the subcarrier they put just Rm. (Fig. 5)

The receiver juggles and re juggles this material so the left speaker presents Lm+1/2(Lt Rt) while the right one presents Rm + 1/2(Lt + Rt).

Disregarding, for the moment, the questionable validity of the 3.500 cps assumption, this system is compatible with the present AM/FM transmissions, and does do the other things claimed. The question is, whether Calbest's "findings" about the need for stereo difference above 3.500 cps are final.

We do not dispute their tests, but has everything been taken into account ? If we grant, for argument's sake, that present stereo does not benefit from maintaining separate channels above 3.500 cps, is it safe to presume that improved techniques will not later reverse this decision? I believe they probably will do so.

There is no need to assume, as some seem to, that one man tells the truth, while his opponent must necessarily be lying. The unfortunate facts are that so little work has been done to coordinate tests, and that there are so many variables in both multiplex and stereo.

Programs can be recorded by a wide variety of microphone techniques. Many of the tests claim to have tried a wide variety of recorded material, that should have proved their proponent's contentions applicable to any kind of program technique. But there is at least an equal variety of loudspeaker types, methods of arrangement, and types of listening room. How many have also conducted extensive tests in this area ?

Probably millions of people, on single channel hifi, could not tell the difference when frequencies above 8.000 cps (or even 6.000 cps!) are removed or left in. But there are also many discerning enthusiasts who would immediately notice this difference. It is largely a matter of listening education. But because of these differences, various test groups would lead to varying results and conclusions.

There are many contributing reasons why all the tests quoted may be quite truthful against their own, often unstated, terms of reference, and yet apparently contradict other equally truthful results, because neither presents the whole story.

There is a sense of urgency about the whole thing. The high fidelity industry wants to get stereo "on the road" and right now radio is a "hole" or weak spot. Availability of radio as a stereo source would help sell stereo as an entity to the public. The FM operators and (from page 30) multiplex manufacturers want to see a decision because meanwhile everyone is waiting to see "which way the cat jumps".

However, the fact is that there is as yet insufficient evidence on which to base a decision, in either area. The FCC needs answers as to the practicability of putting however many channels on one carrier they may ultimately decide to allow. Radio is still a public medium, and all interests must be properly protected. As yet there is insufficient data.

Just what stress will go to the importance of maintaining potential fidelity of future stereo transmission remains to be seen.
.

But I feel that stereo is only just beginning, and that efforts so far will be judged as crude by tomorrow's standards. Not only will it be possible to transcribe better fidelity on two channels, but new techniques will emerge, enabling the two channels to contribute to an overall result with much better precision. In consequence, differences which are today observed as indistinguishable may become important tomorrow.

So I cannot advocate the built-in quality limitation of Calbest, which may not be discernable today, but may become very important tomorrow. The intentional quality difference of the original Halstead system has already been abandoned, so I need say no more about it. Both the Burden and Bell systems (if the latter should be so applied) are methods of maintaining compatibility with a system (AM/FM) where quality was inherently different anyhow. This may aid the transition, but I also feel there is a danger that commitment to such a system as standard will restrict the possible future of stereo.

Work in other areas, both with recording media and speaker development, suggests the stereo of the future may utilize channels that are basically "mono" and "stereo," rather than "left" and "right." From this viewpoint, I cannot help feeling that the Crosby system is compatible with the "stereo of the future."

But, much as many would like to see an early decision, I do not think we can expect one, nor would it be good to have on, in the long run. The analogy with the decision on color TV is often mentioned, from various angles. Maybe it's a good one too, if we don't try to apply it in detail: at the time, the decision seemed long in coming; but how far has color TV progressed since? It can hardly claim widespread acceptance. The FCC should not be pressured into an early decision, but will, we hope, wait until sufficient facts are available for a clear-cut decision.

.

Bilder zu "STEREO 1959"

Fig. 1. Basic arrangement of the Crosby system.

Fig. 2. Basic arrangement of the system according to the original Halstead proposal

Fig. 3. The revised Halstead arrangement represented in the Burden system.

Fig. 4. The Bell system, as it would be applied to "straight" multiplex; actually the experimental transmission used TV and AM-FM channels, separately, so it has not been applied to multiplex as yet.

Fig. 5. Basic arrangement of the Calbest system, which is the only one postulated on a narrow band subcarrier.