.

As indicated in last month's article, the two walls of a disc groove carry the two stereo signals so that the left channel modulates the inner wall and the right channel the outer wall.

The directions of these modulations are such that the inner wall moving to the right and up is in phase with the outer wall if it is moving the stylus to the right and down. This maintains compatibility with mono discs such that a mono record will replay with equal sounds comming from each of the two stereo speakers when tracked by a stereo cartridge.

The apparent widening of the mono source of sound is usually deemed preferable to the reproduction through a single speaker. To achieve full compatibility a stereo disc must be replayable on a mono set up and though all stereo positions from left to right will be reproduced in mono, centre front sounds recorded at the same level as side sounds in stereo will replay relatively louder when reproduced through a mono system.

Finally, what has been recorded on a disc marked 'Quadraphonic'?
.

The answer to this is a long story, but suffice it to say for now that the intention is that it be played back through four speakers placed roughly at the four corners of a square; the details of the quadraphonic systems are dealt with later.

The stereo disc will normally have been mixed down from up to 16 tape tracks for replay through loudspeakers, as this is the most common of replay situations. If a listener believes that he is getting truer fidelity or more realism by wearing headphones then this is a personal preference which has to be accepted. There are some who believe that a stereo recording is poor unless it is quite clear which speaker a particular performer appears to be coming from.

This type of recording, mostly in the popular field, accentuates positions as being either left or right and increases separation at the expense of the continuous sound stage across the line of the stereo speakers. I believe there is room for both types of recordings and with at least 20dB of separation available on a disc it is sufficient to indicate the discreteness of the 45/45 disc system. However there is little to compare, in stereo reproduction, with a full orchestra continuously spanning the stereo stage giving apparent width and depth to the recorded music.

Realism is a word which has been used greatly in the description of recordings. It describes the attempt by the producer and engineer to create at the listener's ears the sensation of being in the location where the sound was recorded.

This idea has long since been abandoned and folk who think they are listening to a faithful reproduction of a piece of music as heard from the best seat in the hall are deceiving themselves.

The overall impression may be the same, but with the introduction of microphones to pick out soloists true fidelity is destroyed. The resulting illusion, which is determined by either the recording engineer, producer, conductor or composer or some combination of these, will either be a compromise in an attempt to produce realism, or a production to give a suitably satisfying sound, albeit devoid of any realism.

Musical works recorded on location will contain the reverberance of the hall inherent in the recording and I know it is possible to locate the origin of a recording by the recognition of the ambience on the disc. Studio recordings can have added reverberation and this is, of course, part of the overall illusion.

A disc is a commercial product and if the consumer is satisfied with what the manufacturer is producing then, no matter what tricks the latter gets up to, he is achieving one of his objectives.
.

I know of one steel producer who for some of his product uses the consumer as his quality check, so that he does not make intensive study of the finished steel as it leaves him until customer complaints reach a certain level.

This, to some extent, has to be the attitude of the record producers who have to batch sample discs and be prepared to exchange a faulty product when it is detected. Since the most critical listeners are those who buy classical recordings and tend to be the most vociferous about the quality of the article, one can hardly blame the manufacturer who takes more care over the production of a classical disc than over an LP pop record.

The measure of the quality of a disc which satisfies the listener is a value which varies with experience. It will also depend on the type of equipment used to play it.

I believe that many of the criticisms in the ??? field of audio reproduction are related to a lack of, or misguided, education stemming from preconceived ideas.

Some cannot ??ecipher the messages of stereo as reproduced but of two loudspeakers and some of those who can do not accept any reproduction which does not relate to reality.

Audio Recording is a medium through which an Artist can transcribe his ideas to the listener, and if they do not relate to reality as we know it from our past experience, then why not accept it for what it is?

This has been one of the main criticisms of quadraphonics, but if we accept it as a different form of the audio medium we should be prepared to be receptive to anything it has to offer.

From the point of view of quality of production, someone coming face to face with 'high fidelity' for the first time, having listened to lower quality equipment, is often dismayed at the thinness of some of the Sounds and transparency which is there.

(ab hier viele Fehler - es fehlen links am Rand 4 oder mehr Buchstaben )
, of course, is probably conditioned to toning to sound from a small speaker dosed in a small box and has a strong uern of the frequency response of that tern so formulated that anything else ins somehow wrong. The process of (version and acceptance of clear nsparent reproduction is part of the ncational process that everyone has to go ough when making the transfer. It is not nothing which is immediately acceptable ii I feel that some are put off by the fact a they have to do some work, by nding time, in assimilating the extent of at is available to be heard. ii is being said increasingly, now that record industry should not be attempting produce more and more rotten quadratic discs when they are making enough ion stereo discs already. However, :uifacturers will only produce what the lie will buy - with a few exceptions. I some sections of the public are crying for quadraphonic discs, hence the ply. I agree wholeheartedly with hael Gerzon1 when he points out that basics of stereo recording techniques c known long before stereo became mercial, while little serious ligation into quadraphonic recording uiques had been made prior to its 'duction a few years ago.

There is no deal to learn and progress is very rarely at the moment.
.

Let us look at the presently available 4-channel systems (Feb. 1975) in some detail and decline what they have achieved both technically and commercially.

Then it should be possible to see the areas for development in the whole field of multi-channel sound reproduction on disc.

Truly, the two commercially available matrix systems SQ and QS followed by JVC's CD-4 system. UD-4 and the Ambisonic system of NRDC are recent contenders and will be left to the end.

Before we detail the matrix systems, it is necessary to distinguish these from the others. A disc, as has been said many times, has only two walls and can thus carry only two channels of information. Thus to store four sound signals, these have to be combined in some way so as to form two.

The combination must be in such a way that there is a possibility of decoding the stored signals to reconstitute the original four. The mathematics associated with the encoding and decoding processes is called matrix algebra and thus the discs have been called matrix encoded. The discrete forms of recording are FM multiplex types using a wider bandwidth.

"SQ is a further dimension in recorded sound playback, evolved by joint cooperation between CBS and Sony to achieve a nearer step towards the ultimate in sound reproduction." This I quote from literature promulgated by CBS laboratories in Stamford, USA.

The SQ philosophy *2), *3) and *4), is one of full stereo compatibility of the quadraphonic disc, so that left to right separation is maintained at infinity, theoretically. Also claimed is the ability to record sounds throughout the 360° sound field including interior sounds.
.

The result on the stylus motions of the system is clearly seen from equations (1) where if Lf is the only signal input then only the inner wall will be modulated. Similarly for Rf.

If only Lb is present then LT and RT carry signals which are equal in amplitude but 90° out of phase, LT leading RT. For Rb only this phase shift is reversed.

To say that the back signals result in helical motions is an oversimplification, as we rarely get pure sine-waves recorded on disc. What many fail to realise is that whatever the motion of the stylus it can be resolved into two orthogonal movements; in our case the two groove walls are convenient, and this results in equations (1). (See also fig. 1).

So although the modulus (die Berechnung / Division mit den Rresten der Formel) of all four output signals is 1, the rear channels carry out-of-phase signals which do not lead to clear image placement.
.

For example the "centre front" to "centre back" separation can be increased from 0 dB to 6 dB by the introduction of blend resistors in the front and rear channels.

10% blend at the front and 40% blend at the rear results in the 6 dB separation as above but with a reduction to 20 dB Lf to Rf and 8 dB Lb to Rb.

This 10-40 Blend, as it is known, is regarded as a subjective optimum and provides a good satisfactory listening setup.

Further additions of logic to the decoder can result in infinite separation left to right and 20 dB separation front to back across the sound stage.

The effect of the logic circuitry in trying to decide where the original sound was placed is one of indecision if many sounds are mixed.

The introduction of the disc "Chase" by the group of the same name illustrates the 20dB separation which can be achieved with single sources of sound as the trumpet is heard to circumnavigate (umrunden) the listener.

With a complexity of sound sources on some discs ambiguity occurs and the aural effect is one of pumping. If using, say, the Sony SQD-2020 decoder, three channels are turned down then the decoder's indecision (Wankelmütigkeit / Schwäche) can be clearly heard.

This applies to all channels, but unless one listens very critically to the system with all four speakers up, the 'pumping' effect is not objectionable (unangenehm).

Whatever the theorists say about the asymmetry of this matrix, the public are buying decoders and discs and are enjoying playing them.
.

• Anmerkung : "and are enjoying playing them" ? Und das stimmte am Anfang eben nicht. Die Käufer waren allesamt enttäuscht, daß nach den sicherlich beeindruckenden Vorführ- und Demoplatten die späteren Quadro-Platten mit klassischen Musik-Inhalten extrem unvollkommen abgehört wurden. Und damit hatte die (analoge) Quadro-Technik ihren Stempel als teures unbrauchbares oder halbfertiges und viel zu kompliziertes Spielzeug drauf und wurde relativ schnell ein Flop.

.

CBS have done more work than anyone on the development of SQ and have, of course, tried to deal with the shortcomings of the system.

Aerial images cannot be satisfactorily recorded with a basic encoder and so an "interior switch" has been fitted, which brings in delay lines so that the rear signals are recorded 10-20 ms later than the corresponding front signals.

This and several other encoding techniques enable the system to be used at the expense of the producer. This is fine for the consumer, although complexities in recording and mixing are bound to lift the price of the quadraphonic disc-which is high enough already.

The disc characteristics are identical to a conventional stereo disc in composition, bandwidth, recording level, etc.

Stereo compatibility is maintained in that all positions around the 360° field will reproduce in stereo and the side sounds in quadraphonic replay appear near the speakers in stereo replay though not very clearly defined.

Mono compatibility is another problem in that centre back sounds disappear when the disc is played monophonically, centre left sounds are attenuated and centre right sounds boosted.

The shape of the graphs indicating the range of playing times show a very wide spread for the RCA discs, from one short section of Rachmaninov on the B side of a disc to Swan Lake which runs for a total playing time of over 55 mins.

This latter disc may have been pressed to indicate the long playing time that was possible on the CD-4 disc, but it is admitted that a few dB drop in level is needed in some cases.

EMI have produced a number of SQ encoded discs as well as stereo versions of the same material. A separate remix has been done for four channel reproduction as one does not get optimised results with an SQ disc played stereophonically.

I forsee that when quadraphonics really catches on, the expense of a four channel, two channel and possibly mono remix will result in quad/stereo/mono compatible discs being pressed, albeit optimised for quad.

The rear outputs are again shifted in phase to give
.

Examination of these equations reveals no cross-talk in the diagonally opposite output channels, for example, Lf' does not contain a Rf term or vice versa.
.

The vectors of the stylus around the edge of the sound held for QS are straight lines and if the sound moves towards the centre of the field they become elliptical and finally circular in the dead centre.

The reduction in stereo separation can be deduced from the diagram of QS stylus motions where the Lf and Rf directions are not orthogonal (as for SQ) but are only 45° apart.
.
Projecting, say, Lf onto the stereo axes gives a separation of 7dB compared to SQ's theoretically infinite value. Like SQ, where centre back sounds disappear when the disc is played monophonically, QS also has this limitation as seen in the diagram where vertical modulations result from sounds placed in this position.

Lower level rear sounds away from centre also result from mono playing of the disc. This means that to maintain full compatibility the engineer needs to keep soloists away from the centre and attempt to maintain 20dB of separation in rear signals to avoid the loss of information in mono playback.

Diagonal separation is characteristic of the QS matrix, together with crosstalk in adjacent channels. Sansui have also attempted, by using detection and nonlinear circuitry, to increase the apparent separation exhibited by the basic decoder.

They have put forward a variable matrix solution which, like the SQ logic system, has to detect and compare phase and amplitude values in the matrix outputs.

The SQ logic system varies the gain control of the amplified outputs, whereas the "Variomatrix" alters the matrix angle for unwanted signals, thus effectively increasing directionality. Separation figures of greater than 20 dB are claimed.

Again, like SQ, the disc has normal bandwidth and cutting restrictions as stereo. Pye have issued a limited number of QS encoded discs with a fairly high average playing time with little spread. What has happened to the output from this company is that their QS encoded discs which were initially marked 'Quad 4D Series' are now marked 'Stereo 4D series' with no indication on most of the discs that they are encoded at all. This has been brought on by the lack of customer understanding of what the quadraphonic disc is all about.

The artists on the Pye label are not making the sort of LP that hits the charts regularly and, unless Pye's ability for encoding is used by the smaller labels that can produce hits, it is unlikely that QS will make any headway through this channel.

"The CD-4 system is the only discrete disc system in the world."

This claim really needs to be modified slightly by the introduction of the words "commercially available".

CD-4 *7 und 8) stands for Compatible, Discrete, Four channel system and was developed by JVC, the "Victor Company of Japan", and as well as using the normal audio bandwidth normally up to about 18 kHz, this system uses a carrier which has been set at 30 kHz.

The location of the bands is shown in fig. 4, which indicates a reduced direct (audio) band which extends only to 15 kHz.

• Anmerkung : Auch das stimmte bei den Consumer Demodulatoren vor 1973 anfänglich nicht. Wegen der viel zu flachen Filterkurve (der Transistorschaltungen mit 6 db/oct.) mußte bereits bei 13 kHz das Tiefpaß-FIlter einsetzen, damit bei 15 bis 18 kHz nichts mehr durch kam und das FM Signal nicht stören konnte.

This enables adequate filtering to take place to exclude any effects from the high frequency band starting at 20 kHz.
.

The Japanese originators held listening tests in Japan with audio experts and concluded that 15 kHz was an adequate bandwidth for hi-fi reproduction.

There are those who would disagree and see this as a lowering of standards. The band from 20 kHz to 45 kHz is frequency or phase modulated about a 30 kHz carrier and is recorded 19dB down on the direct band to enable the stylus to follow the modulations.
.

.

1. Lf+Lb -> 30 Hz to 15 kHz direct Inner Wall
2. Lf-Lb -> 20 kHz to 45 kHz FM/PM Inner Wall
3. Rf+Rb -> 30Hz to 15 kHz direct Outer Wall
4. Rf-Rb -> 20kHz to 45 kHz FM/PM Outer Wall.

As for stereo, and to maintain compatibility, the walls are modulated away from the centre of the disc when the sum signals are positive. Also, when the difference signals are positive the convention is that the carrier is deviated upward in frequency. This FM band carries a double sideband up to 10 kHz and a single sideband from 10 to 15 kHz.
.

Again for compatibility the direct band is to RIAA standard whilst the FM band has JVC's own noise reduction system applied to it.

This ANRS (Automatic Noise Reduction System) is an expander / compander type somewhat similar to Dolby but is not fully compatible. However, since it is only on the FM band the demodulating circuits take this into account in 4-channel playback.

Since the success of the CD-4 system relies on the production of a satisfactory disc, it is probably worthwhile having a look at one or two further parameters.

The RCA Quadradisc, it is interesting to note, has a specification on cutting level which shows a drop of 2dB at 22 minutes playing time. No further drop is stated. Separation figures given are Right to Left 25dB and Front to Back 20dB, both quoted at 1 kHz with a signal to noise ratio of 56 to 59 dB. Comparative demodulator figures are >45dB and >30dB for separation and >60 dB signal to noise.

This disc system was deliberately designed to give discrete outputs and therefore had to go to a frequency multiplex system. The achievement of tracking a disc up to 45 kHz is acknowledged to be great but the technology is still advancing to allay the fears of those who are worried by higher distortion and lower cutting levels.

JVC have tried the philosophy of building as much technology as possible into the stage up to and including the disc. This then keeps the complexity of the playback equipment down and reduces the costs to the consumer.
.

The system has not only to be technically compatible but economically compatible with other systems too. This latter awareness has caused the originators to try to quell the fears of excessive wear by quoting a figure of 100 stereo playings as a minimum.

The full statement says that after tests there was still sufficient carrier level on the disc to enable decoding to take place even though the disc had been played 100 times using a conical stylus and tracking at 5g.

What this means in fact is that although the conical stylus wears the disc carrier away there are areas of the groove walls which are not touched by the stylus.

JVC developed a new stylus (Shibata) which more closely approximates to the groove shape and contacts more area of the walls. Thus when the worn disc is replayed using this latter stylus it is able to track unworn areas of the groove and detect sufficient carrier for the demodulation circuits to lock onto. The wearing process leaves its mark though, and noise becomes a problem. This is more noticeable in four-channel playback than in stereo.
.

• Anmerkung : Es gab weitere Restprobleme und Tricks beim Schneiden der Folien.

.
I believe technology will sort out the remaining problems of the high frequency disc which may include the "new universal vinyl compound formula" announced by RCA in May last year 1974.

Other manufacturers have developed other styli, including the Pramanik from Bang and Olufsen who are now convinced of the commercial superiority of SQ and CD-4.

Tracking distortion has been reduced on the CD-4 disc by introducing a Neutrex tracing waveform compensator, again at the cutting stage, though one factor must be worrying JVC and RCA, and that is the fact that cutting can only be achieved at half speed.

Originally the discs were cut on the lathe at 1:2,7 speed reduction but this has now been changed to 1:2. It seems unlikely that these discs will be cut in real time in the near future.

No commercial cutting of CD-4 discs has taken place in the UK and discs have had to be imported from Japan by JVC (UK) Ltd. and from USA by RCA and WEA (Warner, Electra, Atlantic).

The American companies are not keen for wholesale importation of CD-4 discs until there is sufficient hardware in the consumers' hands to justify it. Thus it is difficult for the dealer to sell CD-4 playing equipment knowing that there are only limited discs available.

Theoretically the CD-4 disc is fully compatible for stereo and mono playback but ignorance by the general public and wariness about the wear problem has led many to avoid CD-4 discs which are mixed in the records bjns with stereo discs.

Playing times are very variable and contain a higher proportion of short sides than the other systems, though I must admit that they mainly originate from Japan.

This problem of wear and distortion is going to be a headache for Nippon Columbia too, who have also developed an FM multiplex system originally named QMX *9) but now commercially called UD-4.

The bandwidth diagram for UD-4 is similar to CD-4 but the band limits are different. The base band is extended to 18 kHz and the FM band, again centred around a 30 kHz carrier, is reduced to a double 10 kHz sideband.

This system is a hybrid matrix/discrete combination in that a matrix called BMX is recorded on the base bands and additional FM signals allow two further channels to be combined with the BMX matric to produce discreteness.

Four signals as for the four speaker outputs are derived from the recorded signals but are not recorded as such on the disc. The mix is such as to maintain mono and stereo compatibility.

The four signals recorded are designated T(sum), T(diff), T(T) and T(Q) the subscripts standing for sum, difference, tertiary and quaternary.

T sum is merely the sum of all source signals without regard for azimuth angles. So T sum is an omnidirectional channel corresponding to mono and is represented on the disc by lateral modulation.

T diff is again the sum of all signals but each signal is shifted in phase to lag their counterparts in T diff by a phase angle equal to the azimuth angle for each source, relative to centre right, for example.

T diff is represented by vertical modulation on the disc. The third signal T(T) is the conjugate phase version of T diff and T(Q) is a double azimuth angle form of T diff.

The decoding of the BMX matrix enables sounds to be located in the four speakers with crosstalk in adjacent speakers at -3dB relative to 0dB in the desired speaker. Using the T(T) signal which is on the FM band it is possible to increase the directional location of sounds, as shown in fig. 5, and the addition of T(Q), the other FM signal, allows discreteness to be achieved.

It was found subjectively that it was not necessary to transmit the full bandwidths of T(T) and T(Q), as not all frequencies were needed to determine directionality.

Thus a reduced band can be used which it was found possible to reduce even to 1-5 kHz. This leads to some problems at higher frequencies but is generally adequate. The addition of T(T) and T(Q) affects direction location more for off-centre listeners than those central to the field, but has been found to approach the discrete sound of a tape by listeners in any position.

The total result of all this is to provide four-channel playback cheaply by decoding the BMX matrix or by using a more expensive set-up to achieve discreteness by utilising the FM bands.

There are now available UD-4 demodulators from Denon which include BMX (and CD-4) decoding facilities. Discs are also being brought in, and from my limited experience with this system there appears to be very good directionality and clarity together with low distortion using UD-4. Since the carrier is set at 30 kHz it is also possible to play the disc through a CD-4 demodulator with quite acceptable results.

NRDC, who have been sponsoring phase encoding methods at the University of Reading since 1970, announced last year, through IMF, their Ambisonic system. *10)

This is basically a recording technique to create total ambience, i.e. reality in four-channel playback. It utilises the BMX matrix of the UD-4 system and in fact NRDC have a prior patent on BMX in this country, thus they can expect some returns even if Ambisonics never take off.

The recording of the matrix signals which contain all the information necessary to provide the outputs as required for the NRDC system requires a special microphone or microphone arrangement.

This has been created by Calrec and it is hoped to do many recordings by this method to prove the system, which can either be recorded on disc or tape, the latter being demonstrated through the medium of cassette at last year's Sonex. No commercial record company is involved yet.

The hardware market is increasingly becoming aware of the need for four-channel amplifiers and built-in decoders and demodulators. SQ and CD-4 seem to be the favourites, though companies like Marantz are providing plug-in units for the customer to buy as and when he wants them.

These contain decoders of the complexity required at that time. Software is slowly moving on to the UK market and I am sure that some matrix encoded discs are creeping in without the designation of Quadraphonic in excess of the Pye discs, broadcasting is having a long look and is considering whether a discrete system is exceded.

There are, I believe, about five discrete systems being examined in the USA, and the BBC have their own ideas. An experimental broadcast was made at midnight on 5th July 1974 last year when Radio 1 carried the front channels and Radio 2 those at the rear. The material was a mixture of classics, pop, light music and drama and was generally well received.

The disc situation is the most interesting I think and there are developments going on now which could lead to new features of disc recording in the not-too-distant future.

Quadraphonics is here to stay, but my main hope is that the decision as to which system or systems finally exist will be made on comprehensive artistic and economic grounds and not from commercial pressure.
.

.

• 1 Hi-Fi News & Record Review Annual 1973.
• 2 'A compatible Stereo-Quadraphonic (SO) Record system', B. B. Bauer, D. W. Gravereaux, A. J. Gust, JAES, 19, No. 3, Sept. 1971.
• 3 'Recording techniques for SO Matrix Quadraphonic Disc', B. B. Bauer, G. A. Budelman, D. W. Gravereaux, JAES, 21, No. 1, Jan./Feb. 1973.
• 4 'Quadraphonic Matrix Perspective - Advances in SQ encoding and Decoding Technology', B. B. Bauer, R. G. Allen, G. A. Budelman, D. W. Gravereaux, JAES, 21, No. 5, June 1973.
• 5 The Japanese Regular Matrix, P. B. Fellgett, HFNIRR, Dec. 1972.
• 6 The Sansui QS Coding System, R. Ho, S. Takahasi, 42nd AES Convention Preprint No. F-6.
• 7 'A discrete Four-channel Disc and its Reproducing System (CD-4)', T. Inoue, N. Takahashi, I. Owaki, JAES, 19, No. 7, July/Aug. 1971.
• 8 'Further improvements in the Discrete Four-channel Disc system (CD-4)', I. Owaki, T. Muraoka, T. Inoue, JAES, 20, No. 5, June 1972.
• 9 'Discrete-Matrix Multichannel Stereo', D. M. Cooper, T. Shiqa, JAES, 20, No. 5, June 1972.
• 10 The NRDC ambisonic system of recording and reproduction, published by IMF, High Wycombe.
• 11 4-channel Stereo from source to sound by K. W. Sessions. Foulsham-Tab Ltd. pp. 176.

.

.

Carrier channels are operational, there is litle restraint. For the baseband BMX Matrix, it may be expected that considerations are similar to those for QS.

The difference between SQ, QS and D-4, UD-4 is in the point of compromise.

The matrix systems, whatever their respective inherits, compromise location and rely on a simple production system. The others, as a result of the insistence on four independent channels, compromise on noise and distortion and, slightly, frequency response - in other words, channel bandwidth.

In discussing the technical production, there is a clear distinction between processes up to the four or two (matrixed) channel masters and the subsequent cutting and duplicating.

In America, Columbia (CBS, no relation to EMI) use the best logic decoder, whereas London EMI tend to employ a simple decode stage, sometimes with passive blend.

The encode stage is standard. Sometimes a four channel master is made, for later encoding, but with frequent switching between modes to check that the two are compatible; at other times an SQ two channel tape may be recorded directly.

Alternatively, encoding may be at the cutting stage so that there is no tape intermediary. The latter is largely in order to minimise azimuth differences, phase distortions at the highest frequencies due to the impossibility of ensuring that the head block is perfectly in line.

Matrix systems, which rely on phase encoding, are naturally more demanding of tight phase tolerances. However, it is EMI's experience that a properly aligned two-channel tape can act as a reliable intermediary, and that is now their standard procedure.

The only position which is avoided is centre back, for this disappears in mono and is 180° out of phase when the SQ disc is replayed in stereo.

Such signals are very disturbing, although they have been used in stereo productions precisely for that reason. The remaining positioning is up to the producer, although corner sites are obviously more stable.

If the signal is fairly independent in the first place, as it would be if it came from a multitrack master, it will benefit from the playback logic improvements.

The most recent development is the introduction of full corner logic, with variable blend throughout. Another aspect of the very important domestic tolerances was EMI's choice of sets of KEF Codas and Concertos for their evaluations; they may be called "typical domestic speakers" and their image stability is less precise than, say, Quad electrostatics, Tannoy Golds or any studio monitor speaker worthy of the name. This favours a matrix system in direct comparison of position stabilities.

The QS matrix has a similar problem with centre back, but it is suggested that most four track tapes may be processed satisfactorily. Again, azimuth tolerances are tight, perhaps ±20° at 15 kHz, but a two channel tape intermediary is standard.

Generally, a narrow cone going backwards from mid-centre is avoided on remix; since everything behind the centre is, to some extent, out of phase on disc, heavy sound such as bass guitar tends to be placed forward.

This coincides with what seem to be subjective preferences in location. However, in common with SQ, apart from the centre back any position is fair game, depending on the producer.

Sansui also suggest use of QS Variomatrix for stereo enhancement, and it works quite startlingly on some program material. For example, the opening of Money is pulled to the four corners just as if it had been encoded, by virtue of the very distinct separation in stereo.

Stereo matrix replaying also, it is suggested, provides a subjective increase of width. This is a result of the 90° discrepancy between components for some quadraphonic image positions, giving an illusion of being slightly outside the speaker range when replayed through two.

There are no limitations in the control room when mixing for CD-4 disc that are not already familiar from stereo processing. In common with the difference signal between the two front channels being restricted, the difference between front and back, which is the FM signal, must be controlled. Meters are available which incorporate logic comparing each channel with the other three and then display the largest reading found.

Cutting a matrix record is exactly the same job as for stereo. However, due to the presence of out of phase signals in the encoding, care must be taken with the difference channel, which is the vertical component on the disc.

In stereo cutting, limiting the difference channel simply reduces the separation - the alternative to such limiting is reducing the cutting level.

Such a process on an encoded record confuses separation as well as reducing it, so care is taken with relative phasing at the remix stage.

CD-4 and UD-4 records must be cut at slower than actual speed since the cutter is not able to supply the power at high frequencies to cut the groove, due to the need to supply a large acceleration to the cutting stylus.

Present practice is to cut at precisely half speed; next month John Eargle considers briefly prospects for real time cutting which are still fairly remote.

Present cutting difficulties are considerable, and largely stem from the impossibility of aural monitoring. The lathe is set and proceeds automatically; after an hour or so the cutting engineer returns and listens.

If it is wrong, the process is repeated. Although in principle getting the disc equalisation correct is simple, practice is not so. Normal practice is adjustment with three-band limiting and equalisation. Low frequency noise must be reduced, since on replay in real time (i.e. full speed) the frequency doubles and may become obtrusive.

Cutter power can limit the available level. The first CD-4 discs were around 6 dB down on typical stereo levels, although now the difference is around 1-2 dB.

The maximum side time is around 30 minutes for typical material, and around 25 minutes for heavily modulated programs. Signal to noise is steadily improving, latest figures for CD-4 -being better than 55 dBA.

Present cutter powers can be 2 x 500 W (into the Ortofon cutting head). With such levels, heat dissipation is even more of a problem.

The pressing of extended frequency discs is more critical, and RCA have published details of their improved compound.

The only pressing plant outside Japan and the US capable of such work is Sonopress, at Gutersloh, West Germany.

UD-4 records have yet to be pressed outside Japan. Unlike CD-4, UD-4 does not use noise suppression in the high frequency band; details of their own pressing compound usage are not published.