I'm adding some last minute words here, written after the compatibility issue in stereo discs had been resolved by Columbia's decision to begin with a non-compatible disc, minus the full ASRA modification of the difference-signal as described below and in the article on page 26.

An expensive and futile propaganda war of words was thus avoided - but as I see it, the principle of some modification of the vertical cut difference-signal (below) in the 45/45 disc is surely bound to be applied to all coming stereo records.

Maybe it will be unheralded and probably it won't be drastic enough - as with ASRA - to make the records playable on most standard monaural pickups. But it'll be there, for good engineering reasons, if I understand the situation.

I suggest that a year from now you put meters, vertical versus lateral, on a batch of stereo records, any old brands, and see what you get in output. Dollars to doughnuts ....

Since writing the rest of this piece, I have heard ASRA personally on extensive AB comparisons. It works. In most of the comparisons I could not spot any difference - I guessed wrong again and again. In some places, loud transient die-away echos for instance, I could reliably hear a stereo difference; but a barely audible decrease in input level (ASRA is a dynamic device) removed it.

ASRA must be used carefully, of course, and adjusted to suit each recorded sound. But as a recording tool I'd call it at least as practical as a limiter - and less of a distortion.

This new CBS project has a number of intellectual booby traps in it for the unwary, both layman and engineer. A lot of people are already floundering (herumstolpern) around in considerable confusion, I've discovered.

The first thing to know about it, for instance, is that - in spite of the title "Columbia compatible stereophonic record," the development announced at the end of March isn't a record at all, but a compatible system for cutting a stereo record - more specifically, a system for preparing the final pair of signals that is fed to the stereo disc cutter.

The intent is compatibility - as between a stereo disc and a standard LP. The idea was to produce a record that would play with "full stereo effect," to use the CBS phrase, and yet also play safely via any standard monaural pickup, for full monaural effect.

This is not the case with the present 45/45 record, its two stereo signals unaltered. Thanks to the vertical modulation, many monaural pickups - most - will have trouble tracing the 45/45 grooves and many of them, lacking enough vertical compliance, will ride them roughshod, so to speak, doing serious damage. That's the incompatibility that CBS was out to lick.

As Dr. Peter Goldmark, who launched the LP itself ten years ago, said at the CBS press conference, the new CBS system produces a stereo disc that is even compatible with the incompatible disc - the now-standard 45/45.

The CBS disc, for that matter, is in practice a 45/45 record (though it could be a lateral-vertical if desired), with the crucial and tricky alterations introduced into the vertical component of the groove, reducing the vertical "bumps" to the point where they can be negotiated safely by monaural pickups.

The stereo record itself can conform physically to already accepted standards. Only the amplitude of the vertical component is involved.

Thus any stereo pickup will play a CBS-cut disc. All 45/45 pickups will do it, without any special circuit arrangements whatsoever. (Any lateral-vertical pickup, too, if the signals are shifted about by a phasing circuit to equal the 45/45. This is, of course, also true with any other 45/45 disc; it can be played by a lateral-vertical pickup with appropriate circuit changes.)

Similarly, any orthogonal stereo cutter (i.e. a cutter with two cutting motions 90 degree apart, whether 45/45 or lateral-vertical or otherwise), will cut a CBS stereo disc; you just insert a special CBS modifying amplifier between the cutter head and the two stereo signal sources to feed the L-V cutter; an additional "matrixing" circuit (more about which later) is necessary to feed a 45/45 cutter.

The alteration, again, is confined exclusively to the vertical component of the signals. Thus the central feature of the whole arrangement is that, according to CBS, though the vertical component is radically reduced in power - the stereo effectiveness of the whole is virtually unchanged.

With the groove mostly a lateral one, plus only a small vertical component, it is to all intents and purposes a standard LP - and yet it is also a stereo. Have your cake and eat it.

The purpose of all this is a canny one, you see. With an almost fully interchangeable disc, compatible every which way, there can be a much needed "smooth transition" from normal LP recording to full stereo LP, with practically no dislocation in the record business itself.

The changes can be concentrated in the equipment side of the market. This is usefully important if stereo disc is to move in smoothly. This is a cake that could well be worth having and eating, too.

There's no use trying to understand the CBS stereo system until the concept of sum and difference signals is digested; for the whole business, as we'll see, depends on the combining of the two stereo signals - call them L and R - into two compound signals, one the sum of the two and the other the difference. Sum-and-difference, of course, is nothing unusual, nor in any way new to a well trained engineer.

The principle applies to plenty else in stereo, and to any area where two signals are involved. (It's related, after all, to such matters as inter-modulation distortion, where two frequencies generate extra spurious frequencies by combination.)

But I'll confess that to many non-engineers, including a good many of our readers, I am sure, this is not an easy concept to grasp. An understanding of sum-and-difference is an absolute requirement if you are to fathom the CBS stereo system and, for that matter, the whole subject of stereo disc.

The algebra involved is ultra-simple; just addition and subtraction. But to "visualize" (or should I say, auralize) what actually happens when two stereo signals are added together, or subtracted one from the other, requires an exercise of the imagination that can bring sweat to the fevered brow.

It did to mine. I'd already torn out hair trying to understand the M-S stereo microphone, the one that has two mikes mounted together at a single point over the central portion of an orchestra and yet gives a spread-out stereo effect.

Just as I got that subject partway under my belt, including the sum-and-difference circuits that make it possible, along comes CBS. I took a couple of days off to do homework at that point before I could emerge and say that I'd figured CBS out, after a fashion. It was all about sum-and-difference.

Let me look at tins aspect of dual-signal sound before touching on the CBS application of it to stereo disc. A new word has sprung up - new, at least, in this connection - out of the CBS presentation.

It is matrixing, and it comes, I gather, from color TV. The old and familiar term for it was a phasing circuit; CBS calls this a matrixing circuit or a matrix. I'd say that the derivation is slightly doubtful but I'll admit that it's a handy term and likely to get into general use. So matrix it is, in place of phasing circuit.

If the trained engineers will skip forward at this point, I'll get down to business with sum-and-difference. If you are to have stereo sound, two sound-signals of the same music, slightly unlike (nur leicht untzerschiedlich), containing some elements that are identical and some that are different in phase and quality, (with all degrees and shades of difference between) you can quickly assume that these relationships can be treated in terms of like and unlike - or better in degrees of like and unlike.

That gives us an oppositeness (Gegensätzlichkeit) vs. a sameness (Gleichheit), two contrasting elements in the total relations between the two stereo signals.

Some elements are the same; some are literally opposite (180 deg. out of phase) and many are different by varying degrees. If you take these opposing factors as, so to speak, your criterion of comparison, you may analyze two given stereo track L and R into two factors; the degree of sameness between them and the degree of differentness.

Feed tracks L and R into the simplest of simple "analyzers" - the matrix - and you get two signals coming out, each combining L and R, but from opposite points of view; one represents the sameness-elements in the combined signal; the other represents the unlikeness-elements, each in the degree of sameness or unlikeness that exists at any moment, of any of the millions of sound patterns in a musical or other dual stereo sound signal.

Why make it so complicated? Only because, here, I have put a simple arithmetical or geometrical idea into semi-verbalized listening terms. As to circuitry, all you do is to feed tracks L and R into a "matrix" and extract (a) the sum of the two, and (b) the difference between the two. L + R and L-R.

It can be done via various circuit elements, as the editor points out to me, but the classic "matrix" involves transformers plus center taps. Feed in L and R, feed out a Sum signal and a Difference signal. (Nope, this is not merely reversing the phase of one track, but almost. Both outputs are combinations of L and R, but combined in opposing ways.)

Again, though electrically this idea is ultra-simple, the implications in terms of loudspeakers and stereo sound are not too easy to grasp, so to speak, in three dimen sions.

Next, comes the crucial element for stereo grooving, the relationship between these nice sum and difference signals and the combination of two simultaneous signals that goes into all stereo disc groovings.

For those who are still somewhat doubtful as to how a single stylus can produce two signals, take the simplest system - lateral-vertical. Lateral motions of the stylus account for one track, vertical motions for the other. One track feeds to each speaker; the stylus moves in both directions at once but its two internal receiving mechanisms are fixed to respond only to the vertical or horizontal elements in the motion.

Wiggle your needle sidewise and one signal (Anmerkung : besser Channel) goes squawk, the other is silent. Wiggle the needle vertically and the other element gives out with a Mat, the first is silent. Wiggle the needle both ways at once - i.e. diagonally - and both signals are generated.

Ha! Now we approach (herantasten) the 45/45. The Westrex and similar 45/45 systems merely tilt the driving mechanisms, or responding mechanisms, an eighth-turn around, so that each is diagonal to the record surface, making a V at 90 deg.

Feed track L to one side of such a groove cutter and track R to the other, and you have a 45/45 recording in which each track obviously is cut diagonally, one aiming to the right, the other to the left; and in each signal the groove motion is part lateral, part vertical. That's what we mean by diagonal, in layman's language.

Thus 45/45 cutters tend to look like sawed-off V-8 auto engines (ein aufgesägter V8 Motor), two "pistons" (Kolben oder Antriebe) driving a stylus between them. When both push equally, the stylus moves up and down, but if one pulls while the other pushes, the needle sways laterally.

Get it? All combinations of the two motions are possible - and hence all angles of cutting from lateral to vertical.

The stereo cartridge can reproduce the same motions and the same signals in its two diagonal responsive elements, opposed to each other in the same way, at 90 deg. or the square.

So, vertical-lateral is replaced by half-and-half, at one diagonal, and half-and-half, at the opposite diagonal. The great advantage is that a stereo disc cut this way has two tracks that are identical in physical characteristics, though opposite in their "right and left" diagonal direction.

Your two simultaneous recordings can thus be treated with the same equipment, curves, etc., and will wear, sound, and function in all respects like identical twins, right-handed and left-handed.

The 45/45 groove system is now officially accepted by EIAA, together with standards for dimensions, cutting, etc., and so all coming stereo discs will no doubt have this characteristic groove configuration.

Oddly enough, the 45-deg. groove cuts are, in terms of lateral and vertical, a sum and a difference. Each track, in the groove, is part vertical and part lateral, since again, this is what diagonal motion actually amounts to.

Equal but opposite combinations of vertical and lateral motion. Therefore we have a marvelously useful relationship. We can think of - and use - the sum and difference signals, as well as the plain track L and track R signals, in terms of the lateral and vertical elements in stereo recording.

A lovely rule is seen gracefully to emerge from all this. Each time you convert a pair of signals from track L and track R to a sum signal and a difference signal, or vice versa, you in effect move the two 90-deg.-apart groove elements an eighth-circle around. Each time you rotate your actual groove cuttings, you in effect change the played electrical signals from one of these forms to the other.
.

Thus you see how a Westrex 45/45 cutter can also cut lateral-vertical recordings. You don't have to turn the cutter around 45 deg. physically - tipping it drunkenly sidewise at a 45-degree slant so one "piston" pushes up and down and the other is canted all the way over to the pure lateral!

Instead, you simply run signals L and R through an electrical matrix. One of them comes out entirely vertical, the other entirely lateral. Your cutter still does its cutting at the two 45-deg. angles, but the resulting record groove has one track up and down and the other sidewise.
Tricky.

Suppose that you have a record in which tracks L and R are lateral-vertical, as in the excellent English Decca discs of last fall, demonstrated here in New York. You can play the lateral-vertical record with a 45/45 stereo pickup - but what will come out, you can see, is a combination of both tracks from each of the pair of responsive elements in the pickups. Each of them, operating at a physical 45-deg. angle, responds to both vertical and lateral motion - each produces half of each, track L and R, but oppositely, according to the opposed angle.
.

What can you do with such a mixed-up signal? Again, run it through an electrical matrix and it will come out, as nicely as you please, one resulting signal all-vertical and one all-horizontal.

Your lateral-vertical record will then play correctly via the 45/45 pickup, as though you actually had rotated things an eighth-turn. (The only trouble is that transformers and associated circuitry may run into cash; other simpler ways of doing it might be easier, if and when.)

So the conclusion is that 45/45 signals can be converted into lateral-vertical signals and the reverse; anything may be made to play, or cut, anything. Inter-changeability, via electrical matrixing.
.

Now for CBS again. The CBS laboratory study is exactly the sort of canny figuring that Dr. Goldmark's group is already famous for.

The CBS Labs have always been intensely interested in the information aspect of sound reproduction. What are the essential information elements in musical or other reproduction?

This has wonderfully interesting theoretical aspects, but also some hard-boiled commercial usefulness. Anything that simplifies sound reproduction, giving more effective sound for less cash, is highly practical in the commercial sense.

Readers, for example, will remember the first Columbia 360 phonograph, the original "hi-fi" table model, which used an ingeniously tricked up speaker enclosure to achieve a "big" sound in a very small box with small speakers.

(The entire inside of the machine, with the lid closed, was part of the enclosure, and the two speakers pointed out sidewise in opposite directions, for spatial effect.)

That was not literal high fidelity, but it was a typically ingenious economizing of the essential musical effect.

A more significant addition was the extra small speaker optionally available with the 360 phonograph; it was not merely another speaker on an extension cord, but had a special "curve" applied to its signal and to the speakers in the main phonograph, dividing the spectrum between the two to give a maximum pseudo-stereo effect.

That's the sort of thinking that CBS likes to do, and the sort that has gone into the new and immensely more significant development. First, the Goldmark researchers made a long attack upon the nature of those two combined signals we've already mentioned, the sum of the two tracks and the difference.

Now you'll remember that if two identical signals are fed into a pair of stereo speakers (correctly phased, so both push and pull together), you will hear the resulting sound exactly halfway between the two speakers, in the center.

That was the basis of my recent description of the three-track stereo technique, as combined into two tracks; the center track, equally divided between the two speakers, is heard in the middle between them.

In any two stereo tracks, the sum element, the parts of the sound that are the same, in all their varying degrees of sum-loudness, are heard in the middle and monaurally. This is the bulk of the music in all stereo.

On the other hand, the elements that make for the stereo effect - the differences between the two signals - are contained in the difference signal, which can be "extracted" by subtracting one track from the other.

If you play only the sum signal on your two speakers (having put the two tracks through the adding-and-subtracting process) you'll hear a monaural recording, without spatial side-to-side differences. The stereo message is contained in the difference signal that is derived from the two tracks.

You can guess what concerned the Gold-mark-CBS team. Can this difference signal be "boiled down" to its essential ingredients in terms of concentrated stereo effect - and so reduced in power?

If so, then you can have a more efficient stereo recording, with the normal monaural element just as in standard LP records and the stereo difference element added in minimum form.

If you assign the monaural or sum signal exclusively to the lateral part of the recording - the standard part - and concentrate your boiled-down difference signal into the vertical part of the stereo record, then the vertical element will be significantly reduced in power and the vertical "bumps" on the record will be much smaller.

Result: Compatibility.

With reduced vertical excursion of the groove, standard LP cartridges can play the stereo disc. What they "hear," the lateral signal, will be the full monaural effect of the music - exactly as in a standard LP right now.

To the monaural cartridge, the specially treated CBS stereo disc will seem like a standard record, with a slight vertical modulation that will not interfere with stylus tracking.

To the stereo pickup - any stereo pickup - the disc will play with full stereo effect, if CBS is right, though RCA claims this method will result in a reduction of stereo effect. This disc will be all things to all pickups, ultra compatible. (Only a very few monaural pickups with extremely low vertical compliance will run into any trouble.)

The crux of the whole thing, of course, is in first, the listening and, second, the nature of the CBS "Black box," the gadget that does the streamlining of the difference signal, for the vertical cut.

As to the listening - I heard a press demonstration and the stuff sounded just like any stereo, to me, granted poor listening conditions. The stereo difference was very definitely present, and no doubt about it. Though it may not be "perfect," I suspect that the system works, and works very well.

The "black box" itself is called ASRA, "Automatic Stereo Recording Amplifier", which is a fairly significant name.
.

It Is, I gather, available to anyone for a consideration; you insert it before the stereo recording head and make your records. (Any type.) The statistics on what ASRA actually does to the vertical signal are certainly impressive. CBS says that on AB testing, there is no audible stereo change as between a standard stereo reproduction and the same with ASRA inserted in the circuit to cut down the difference signal, leaving the monaural sum signal intact.

Yet at some sound levels (the crucial higher levels) the ASRA difference-signal output has no more than two per cent of the energy of the corresponding sum signal. That's fantastic!

In terms of lateral and vertical modulation, at peak levels the ASRA vertical cut does not exceed one eighth or so of the modulation in the lateral direction - the standard monaural part. The recording really is almost a standard disc.

Now the pay-off: WHAT IS ASRA?

Well, CBS isn't saying, but most of us are doing some fast deducing. My first deduction is that ASRA isn't very complicated, relatively speaking. It is an amplifier (says its name) and it must have a tube or tubes.

It is automatic, which implies quite clearly, along with other parts of the CBS account, that it is dynamic; it does more to the signal at high volume levels than at low ones. It adjusts itself to the incoming volume, cutting out more power at the higher levels, letting through virtually all the original difference signal at low levels.

(Might as well, since the vertical component won't be bothersome then.) This is my guess - without studying the specs.

In fact on the very face of it, this ASRA gadget just has to do certain things. If its peak difference-signal output is down to two per cent of the energy in the sum-signal, then clearly it is removing much of the bass, where the sound-signal energy is largely contained.

ASRA clearly lets through the upper end and suppresses the entire lower end of the difference signal. (It doesn't suppress the bass in the sum signal, the monaural element, and this bass still appears on both final channels.)

It just happens that stereo effectiveness is already known to be concentrated largely in the middle highs, from perhaps 250 cps up to 7000. Above that, we don't get much sense of spatial location out of sound, nor do we from lower tones: (Their overtones do the space-locating.)

So it seems to me that any good technician - following the Columbus-and-the-egg theory - could now sit down and work out a reasonably effective limiting circuit to cut down on the stereo difference signal, before recording, using simple filters, fixed or even dynamic (like the H. S. Scott, but, we trust, not infringing on Mr. Scott's rights) plus a compression circuit that would keep the difference signal from growing as loud as the sum-signal.

Details are unimportant - even a rudimentary application of this sort of thing would do something to improve a standard stereo signal in the desired respect. Maybe just a simple set of filters, removing bass and reducing the mid highs a bit.

That's all my own speculation - you take it from there. It remains finally to look a moment at the great commercial world. If the CBS modification of the stereo recording signal works (and I think it does), who else will buy it?

I point out, first, that though perhaps a good many recording companies would go along with CBS directly, others might not be willing to, notably CBS's arch-rival (der Erzrivale und das ist quasi nur RCA) of well known name.

Would they have to? Consider that:
(a) A straight 45/45 record is entirely compatible with a CBS-modified 45/45 on all stereo pickups. The stereo pickup will play both, with equal ease. Any 45/45 stereo pickup. Moreover, a good many monaural with high vertical compliance can cope with the regular 45/45 reasonably well without groove damage, for a sound exactly like the CBS sound.

But CBS has the advantage in compatibility. Some may argue that ASRA isn't that perfect and who wants to play stereo discs with monaural pickups anyhow ?

(b) It is even more probable that, if the CBS plan shows signs of catching on, other companies will develop their own modification of the vertical groove component (the difference signal) to match CBS.

The titles and trade marks could be anything under the sun and probably will be; but the fact is, as I understand it, that a similar result could be approximated at least, by other companies, in varying degrees, to taste.

So it doesn't look as though everybody would have to go along with CBS, except in the general principle - which can take on any old fancy name you wish in commercial practice.

(c) Does CBS really have an exclusive and protectable principle here? I wouldn't dare hazard a flat statement. But my own feeling, at this early stage, is that it would seem unlikely. There are too many tricky patents and what-not already involved. The exact ASRA circuit and the trade production model may well be exclusive in practice - but somebody else is likely to be able to develop an approach to the same without too much trouble. My guess, anyhow, and purely, strictly unoffical.

(d) But, on the other hand, and knowing Dr. Goldmark's type of work, I strongly suspect that CBS does in fact have an exceedingly good circuit - I suspect that ASRA is both simple and devilishly efficient in its chosen purpose, which is to squeeze the maximum of energy out of the difference signal, yet leave a maximum of intelligence-carrying signal intact.

It may be that by sheer superiority in a very simple circuit, superiority in the arrived-at values, the Columbia stereo reccord might actually be as good as any other in sound (both stereo and monaural) and the most compatible, for the changeover period. If so, the other companies might just as well resign themselves to a species of ASRA once and for all. They can always rename it for their own products.

(e) As a matter of fact, it would appear almost certain that a modification of the difference-signal will eventually be made on all stereo records, of any and every make. It's just plain necessary, in order to allow for full cutting level; a full-strength vertical modulation is almost bound to run into groove complications, even with variable groove depth control along with the usual "margin control." So what's the argument?

After all, CBS used a pair of RCA Olson speakers for the ASRA stereo press demonstration! Sign of compatibility on an intercompany level, we can hope. JE