from quadro to surroundsouna ' elektor may 1977 - 5-19

flOm quadlO,~O (glU)lflfound (giound.1:'.Fellgett and M. Gerzon

In previous artieles ('Quadrille'and 'musie oft¡hespheres', Elektor 17, September 1976,

. p. 910) it was noted that 'Amb isonies' was~,erhaps the most promising of the quadrophonie~vstems proposed so faro This led to a reaetion

~úom Prof. P. Fellgett, pointing out (amongstother things) that he preferred to use the phrase'.surround sound' sinee 'quadrophonies' ratherover-stresses the number four - quite apart from

",oeinga eondemnable mixture of classie tongues.Sinee we had very little information onAmbisonies at our disposal, we took theopportunity to ask him to write a short artiele

--explaining the basie theory and philosophy ofJ}mbisonies. His text reaehed us only threeweeks later (!), and it is presented here in fui!.

J~here some further eomments or explanationseemed neeessary, we have given this separatelyas an 'editorial note'. But first let us see whatProf. Fellgett has to say.

-./.1Livesound

Sound.lield

microphone

fTransmission bybroadcast or recording

Encoding

~. Imaginedsound

Synthesis

Comparisonby

listener

The aim of sound reproduction is togive pleasure in listening. From theearliest days of mono it was appreciatedthat pleasure and realism go together.Not only did the pioneers strive toextend frequency-range and reduce non-linear distortion, but they sought torefine microphone techniques to givethe sense of space and depth of livemusic. Yet realism was still marred bylack of the sense of direction of directand reverberant sound. Stereophonywas the first step in filling this gap,and in the hands particularly ofA.D. Blumlein achieved great refine-ment of technique, and an open trans-parent sense of space of great beauty.Even this was incomplete becausedirection could be reproduced only overa limited stage in front of the listener.The final step has been taken in thepresent decade by the development ofsurround-sound technology capable ofrepresenting direction all around thelistener.Speech, and many natural and everydaysounds, can give listening pleasure, butmusicis unique in its power to sustainrepeated listening. The first requirementof a surround-sound system suitable forrecording is therefore to sound musical.It must be able to fulfil the needs ofsynthesized avant-garde or 'pop' music,and equally to reproduce the sonoritiesand ambience of all the music writtenfor performance from earliest timesthrough baroque, c1assical and romanticto the present day. This requirement issymbolised in figure l. The test is notconformity with any artefact that is atechnical means to an end, such as amixdown or a 4-track master tape. Thetrue criterion is comparison of what thelistener hears with the original live orimagined music.

The patents rights in the circuits described,in this article are owned by the NationalResearch Development Corporation.

Loudspeaker

. leeds!Decodmp ~ loudspeakers

Sound in

listening room

9833.1

Figure 1. Block diagram to iIIustrateevaluation of a good surround-sound system.

, J

"-1rom quadro to surround sound5-20 - elektor may 1977

2

Conversionto studioformat

Directionencodingmicrophone

Live r\performance l{

, Mono I, Stereo 1

Steering andpolar of virtualmicrophones

T o conventional

processingRecording(optional)

Recording ortransmission(2,3eKchannel)

4-track

master tape Q(pairwiseblended)

.L Surround

... soundEncodinqMixing

9833,2

Mono sources r\(multitrack, l{multimicrophone,etd

3

Pressure WLQ

SHELF FILTERI

UJCJ:::Jf-:ixCl.-20:«f-,«~2«ICl.

Forward

velocity XRQ

SidewaysvelocityYTQ

Input channels(two or three)

SHELF FILTER

Al! shelves haveidentical phaseresponses

/

4

2UJ ~

U oZ c::UJ 0-0o: E '"UJx"i;u- - OJ-u-o:.e~i5 ~ ,~ ~o(S2 ~~"'" .e°"" - OJ:::J ,- ~(/) ~~

j¿¿ Sum

RIGHT Q-j¿ o:

Oxf--(/)0:-f-~«0:2

(/)o:

UJUJ(/)f-

o « ~oIIC!>Cl.(/)

lo

ilolo Diff

LEFT Q-jlo

i1ijI1I¡

J

UJ , ¡¡¡CJ -'":::J , '"'"f-..J , '"Cl.X o2-«O: Ef-f- ' o:::J«Cl.2 ,f- .8:::JO o-, f-

9833 ,3

Speaker layoutshape compensation

S-formatinputs

..Q-)00

.g:)

CJ '"UJ -'"UJCJ

'W'"'"f-

:::J;;j+ f-

:: \

"..J c::Cl.X '"Cl.UJ« 2-i;2l9 «1=O-Cl. f-« ;¡::uu-:::J 2

'RFCiUJu-UJ E(/)..J f- '"«UJ« :::J

'RSoiJ52 O f-

9833,4

srlayoutcontrol

from quadro to surround sound

Evidently no technology can pass thistest unless it can han die live perform-ance as we11 as synthetic pan-pot andecho-effects. It is found that liveambient material provides the mostsensitive touchstone, for musicality' isquickly lost if the loudspeakers areaudible as separate sources of sound, orif ambience pools in particular direc-tions instead of being smoothly distrib-med. The real problem is to leave thelistener undisturbed by obtrusion of thetechnical means of radiating sound intohis room. If this can be solved, itfollows almost automatica11y that awide range of synthetic effects, fromthe simplest to some of great subtlety,can be made available by artificialsimulation of elements from theresponse to natural ambient sound.The purely engineering requirements arealso exacting. To be acceptable today, atechnology for surround-sound mustcover a11essential needs from studio tolistener. It must provide for the varietyof microphone techniques and artificialprocessing, and be compatible with a11valid philosophies and aims of recording.lt must provide a compatible series ofl1}ft.ansof transmission beginning with abasic 2-channel implementation suitablefor dissemination by existing stereomedia (disc, tape, fm) and, extending toit, effective use of multi-channel media.Very desirably it should enable the basicversion to be enhanced by adding athird channel of reduced bandwidth oramplitude suited to carrier-disc orenhanced fm stereo broadcasting. Thebase-band signals must have good monoand stereo compatibility. The listenershould not be restricted to having hisloudspeakers in a square (a shape to

.~ch few domestic room s are suited),and he should have freedom to enhancehis system by using more than fourloudspeakers if he wishes.

Figure 2. Some facilities in ambisonic techo'-mology.

Figure 3. Block diagram of a generalisedambisonic decoder (courtesy MAG).

Figure 4. Block diagram of a domesticambisonic decoder of medium complexity for2-channel (45JB) material. This will accept2.channel or 3-channel B.format inputs anddeliver mono, stereo, superstereo or 45JBdecoded outputs. It can drive rectangular orregular hexagon loudspeaker layouts via 3 or4 power amplifiers. Loudspeaker distance andpsychoacoustic compensation are provided.

It goes without saying that any designwhich is free from compromise is over-engineered. A most important require-ment is a body of theory competent toshow how specific design aims can berealised in a minimal design, and toenable rational engineering choice s tobe made in accordance with statedpriorities among competing require-ments.The earlier so-called 'quadrophonic'proposals for surround-sound fe11shortof fulfi11ingmany of these demands, andindeed made no attempt to meet someof them. They were conceived beforesufficient systematic understanding wasavailable to avoid arbitrary choices, anda proliferation of mutua11y incompatiblesystems resulted. A new approach wasneeded.

Second generation surround-soundThe first of the new genera tion ofsurround-sound systems commercia11yavailable was the UMX system devel-oped by Professor D. Cooper (Universityof IlIinois) and Dr. Shiga of the NipponColumbia Company, who marketed thesystem.To this has been added extensivetheoretical work and experiment carriedout at the Universities of Oxford andReading under the auspices of the BritishNational Research DevelopmentCorporation, who hold extensivepatents in many countries on theinventions which have resulted. Thecomplete technology is called'Ambisonics', and within it a compatibleseries of signal formats known as 45J(a development running number) hasbeen internationa11y agreed. This engin-eering standard is the subject of currentsubmissions to appropriate standardsauthorities in a number of countries.There is no simplistic 'secref ofAmbisonics. lts capability of highperformance is due to clearly formu-lated engineering and aesthetic aims anda body of experimentally validatedtheory which places the realisation ofthese aims whithin normal methods ofengineering designo Some uniquefeatures may however be especiallynoted.Ambisonics is not an isolated 'system'but a comprehensive technology center-ing around engineering specifications ofthe signals for public distribution.Specification 45J includes a basic 2-channel format, designated 45JB, whichis of itself a carefully designed surround-sound implementation with well formu-lated and balanced mono and stereocompatibility. This is the lowestmember of a compatible series whichincludes members using 3 channels, orwith a 4th channel which may conveyheight information for periphony(spherical surround). Particular atten-tion has been given to fulfilling the needfor '2V2-channel' specifications in whichthe 45JB basebands are enhanced by athird channel of restricted bandwidthor amplitude suited to carrier disc orfm broadcasting.

elektor may 1977 - 5.21

In the studio .

Ambisonic studio technology includesthe soundfield microphone for naturalsound, and a wide range of pan-pot,reverberation and spread effects con-forming to an accurate specification.Correctly designed ambisonic pan-potsare actua11y simpler than pairwise'quadrophonic' pan-pots, or can bemade more versatile. Existing pairwisepan-pots can be converted by additionalcircuitry. There is provision for usingarchive material recorded in pairwiseformat, including a comprehensive rangeof choices for the producer to optimisethe compromises inseparable from thisinferior method of encoding direction.A studio format, known as B-format, isrecommended which facilitates pro-cessing and reduces sensitivity to smallerrors of amplitude or phase, especiallyin recording. An attraction of thisformat is that material held in it can beencoded at a later date into any reason-able surround, stereo or mono form fordistribution, so the risk of obsolescenceis minimised. Figure 2 iIlustrates someof the facilities available in ambisonictechnology.

. . . . . and at homeThe ear and brain do not locate sound

by only one mechanism, but by many.The goodness of localisation, and theimpression of naturalness and correct-ness of sonority and tone, are found toimprove when different mechanismsagree. Ambisonic decoders are designedto give correct localisation by as manyspectral (frequency domain) and bi-spectral (first order non-linear) mech-anisms of the ear as possible. Decodersfor 45JB include a 'front preference'option to cater for material in whichfrontal sources are predominant. Anynumber of loudspeakers in any layout(within reason) can be used. Decodersfor four loudspeakers include 'layout'and 'distance' controls permitting ad-justment for the ratio of length tobreadth of rectangular layouts, and forsize of room. Decoders can be providedfor trapezoidal layouts, or for five, sixor even more loudspeakers if wished.The inter-relations between loud-

speaker feeds demanded by psycho-acoustic criteria produce the usefulbonus that four loudspeakers can he fedfrom only three power amplifiers, orsix loudspeakers from four amplifiers.Since the mechanisms of the ear, andtherefore the psychoacoustic criteria,change with frequency, all decoders(except the most basic for the lower endof the market) employ special phase-compensated shelf filters. This fre-quency-dependence improves not onlylocalisation but also the subjectivetonal quality in comparison with naivedecoders lacking this feature.Loudspeaker emphasis and signal-dependent gain devices can be incorpor-ated in ambisonic decoders for those

occasions when a more gross effectmay be deliberately sought. Am bisonic

~~

- .,"O~ goen

"Oc:::;¡

~::;¡en

,~ f'

y x IV.o .o .o S-'olmo! Inputs PHASECOMPENSA TED

SHELF FILTERSIVITH

RESISTIVE INPUTIMPEDANCE & GAIN

OUTPUTMATRIX

>..Eo..

.:>t'"'iI

NNlb

~O

-i;..::;¡ITEe...

SUM & DIFFERENCEMA T R I X

8 POLE/ZERO PHASE SHIFTERS

L8

LF

."" "- eRes" OeoAC"AN" SW"C~O HO" 2 WAY

NORMAC

rr~

AF

'fORMA' ~2''"O" 3 WAY

NORMA' O IOR ,"O" 2 WAYI

SeA" OAS

NORMAC"- ~¿~~~;'~:~I>cAR' O

,...,...en...

0,"- 3eoc"WAYOo

Many 'pI""is;on' "s;stOIS a" achieved

by se,ies o, p"allel pailS

component values shown "e ideal values

O~3O o "'- ,ouoseeMee 'AYOU', eo" 3WAY

00

ID

....

speakec signal outpuls

.CRo,C8

." IT)--() IV

~~~b;)-~~~'d---th;?U~uc.t~'Sansui'ir19m;t1i~;'p,'lecoder, with simplei'aed circúitry. An ambisonic decoder¡tJl,signal-dependent gains has been

iesignéd having improved conformityWitlt" psychoacoustic 'requirements.ligtire"3 is a block diagram illustrating:generalised ambisonic decoder; ofourse not alIof the features need be;~sent in 'a particular consumer1:9quct.

~:'features of ambisonic decoders

.i¿"'02,,::;~!?or~ or less realisable (by suitable~"k~motlificationsof circuitry) with source

erial encoded in any reasonable way,/"b~~h?JI.~ best results require correct 45Ji:~(6re<lliivalent) encoding. BBC 'Matrix H'~~p.co<liÍigis close enou~ to 45JB to be

p1.ay.able directIy into ambisonic 45J..dé~oderswith psychoacoustic enhance-:~IIl~l1Í'of performance. There is effective"~cqinpatibility with Nippon Columbia

ienon, UMX, and a simple switchingaffles RM, QS and SQ material to ben,dl~'dwith a degree of approximationI?~l1djngon the source encoding. A~re({decpde'. mode is available. This

~

e,s'FilOt deliberately give a synthetic~jfdo-surround effect, but enablesGiilisatioILin the frontal stereo stage to.'Jmproved by the use of four (or

'e);Joudspeaker playback.!re ~l is a block diagram of aestic decoder of medium

)h1I?lexity for 2-channel (45JB)at~rial. The blocks are expanded into

L. circuit configurations in figure 5nhese are in turo shown in moreili1Í figure 6 (a), (b), (c) and (d).

í~!lre 7 shows various ways of usingJh~'~qÍ1tputs for four or six loudspeakers. sin.g three or four power amplifiers.

oder functionsdecoder circuit given here has a

-position mode switch, labelledo', 'stereo', 'superstereo' and'. It should be noted that the

o' and 'stereo' positions are not

~~n:edbY simplyswitchingoff,

'

,.

"

.

"

,,

'

,,

'

,

n

,.

'

.

'

.,.

t

',

ed chan nels or shorting outputsthef.O"reproduces mono material through

r,.,6'>loudspeakers so that the sound. appears to be straight ahead, but

~rger distance than the speaker

~1~~~~\',It' avoids the 'coming from a~."bO}(. ,,'"effect of single-speaker mono~';rei>ro~ru6tion, and the 'in-the-head' and;'>~¡"qk.~¡cQlourdistortions associa~ed with~?~2-sReaker mono reproductlOn. In~l~g!~cHcé, 'mono' decode helps unlock

le''Spacious quality contained in good" ono recordings. -~~stereo reproduces conventional stereo¡{material over the conventional front-W,41!(i!ladrantstereo stage, but uses the extra8"(~Q~!l.kersto reduce undesirable qualities~':'I.()f';2:~Pfaker stereo reproduction -

~f:(;no~ablyexcessive speaker prominence,uristable images and tone-colour

;~':/(íistortions. 'Stereo' decode does not;i~~eñ~!!.~cethe quality of badly-recorded

ereo material (but neither does it

RIGHTQ . L

LEFTQ ."+75dB~.4

~ pme' oc baek.panel ,witohNORMAL-o

Notes: Resistors marked * may be 5%, **may be 10%Others shown should preferably be 2% or better

~

,.

Sb"

J.:

L .r

o:¡

¡[.i]p:

~~.. "'!

.f*1 1"" "

-jL

9833.6h

Notes: Áesistors marked * may be 5%,** may be 10%

Figure 5. Overall circuit diagram of domesticambisonic decoder corresponding to the blockdiagram of figure 4.

Figure 6a. Detailed circuit of input stages(sum/difference matrix).6b. Detailed circuit of phase shifter stage. Twoof these stages are required. one for the ~ andone for the .6. signal. Nominal performance:phase shift 90° :!: 1'/3 ° from 30 Hz to16 kHz. A 2% component spread gives 90°:!:3° over this range. Stage gain is 1.33 dB.

worsen it). Good material will obtain afurther sense of spaciousness withoutany gimmickry of directional éffecHThe effect is almost completelyunobtrusive and yet capable ofconsiderable reduction of listener'fatigue through presenting the ears witha more naturalsound. 'Stereo' decodec"also tends to enhance the reproductionof SQ-encoded records.Superstereo gives a wider frontal stagefrom stereo material for those who wishfor a more spectacular effect, without}descending to mere gimmickry. This,.position also provides excellent decodmg ¡¡¡

i

óC,

"

5-24 - elektor may 1977 from quadro>to surround sound

BeNORMAL~

SPAREO

MONO q,.SISTEREO o ~SUPE RSTE REO o45JB o

b",k.b.'o".m,k.

B.FORMAT ~W S2A-o NORMALO

SPAREo

l:

-jt>

l:

it>

x-o

l:

jt>

l:

-jt>

y-o

t>

~.y

jl:

t>

-il:

t>

Component' marked * may be 5%, **may be 10%

.w

.x

of Regula~ Matrix encoded records, witha full 360 reproduced stage.45JB decodes system 45J recordingsand broadcasts. It is also suitable (butnot optimised) for decoding 'Matrix H'for good ambient reproduction, and it isreasonably compatible with BMXrecordings.B-format enables those with a 4-channelrecorder to get optimal studio-qualityreproduction fram master recordingsmade in the ambisonic B-format studiomode.

Editorial notes.

Let's be honest: we are impressed by thecapabilities of ambisonics. To sum it upbriefly:- an ambisonic system will accept practically

any input.- an ambisonic system makes the fullest use

of the available number of transmissiónchan nels.

- an ambisonic system can drive almost anyloudspeaker layout, using a minimumnumber of power amplifiers.

- the designers of ambisonics have evidentlydone their homework.

Great!Now let's be practica!. At the moment, noambisonics recordings are commerciallyavailable. There is very little point in buildingan ambisonics decoder until some record!company starts supplying these recordings, (}~until some broadcasting corporation starts'transmitting in ambisonics. From a technicalpoint of view (please note: not necessarily a.commercial or even a consumer point of view)we wish that this situation were different.In theory, and even - as we have heard onfirst-hand authority - in practice, a surround-sound demonstration using ambisonics isunbelievably convincing. Speaking only fromthe theoretical point of view, we are preparedto go along with this: we have favoured UD-4so far, since this system seemed to make fulluse of the available transmission capabilities.In our opinion, ambisonics is an extension of

UD-4: technology and mathematics have been ¡complemented with psycho-acoustic research.1We have tried to give all systems an equa'

f

chance in the pages of this magazine, and we l.

have noted the good points of each system. I 1,

.

.

previous articles we have even noted tl~" 1there may be room for more than one system'.,However, if only one system is to be chosenfor universal application, our money is onambisonics - on a purely tethnical andtheoretical basis. We can't wait to hear a

practical demonstration! . fFinally, some comments on this article. It is, Iof course, incomplete. It is sufficient to whet fthe appetite - without giving something <.:O

!

bite on. The author explains that they arestill. negotiating with various parties and

..cannot release more information. .The circuits are quite basic, but they are

tapproved by the inventors. There are noalignment points, to our relief, so we don-'f'

o"d ,o 9;" ,1;90m,"' ;OM'"";O".'. SO 1009":there are no commercial releases of ambisonic

recordings, we do not feel obliged to offer atprinted circuit board.For that matter, Mr. Gerzon (who designed

Ithis decoder within 48 hours!) has pointed'

out that there is room for improvement and ¡simplification. When the time seems ripe for aconstructional article with a p.c. board, we ¡

hope to take him up on this.- However, the jcircuit shown here does give an idea of what Ian 'ambisonics' decoder looks like, and that

.

;'

t

.

.

sufficient for the presento ti.~

from quadro to surround sound elektor may 1977 - 5-25

Sd

.

.I *~" CR oe CB

1 Q'w

**,," w

6"

Q 6800

x~

LB

LF

Q 6800

Y0"1

33'

m RF

<::>°BAse e LOUOSPEAKE R LA YOUT

RBm

Notes 11Res;stors macked * may be 5%, **may be 10%21680 nF capac;tors may be ;mplemented as

220 nF and 470 nF ;n parallel31200 k may be replaced by 180 k and 22k ;n ser;es4i1f potent;ometer;s 4k7, make up to 5 k

by two 150 S1 ser;es res;stors, one at each end

¡'.,

l. 7a,. 4 SPEAKERS D

:tAM'"RFQ

"Qo---{>-l &"M'~

~~~R

"Q~ ~"A'G","Q 1>19833.7,

~

~POWER AMPS

7b 7d

4 SPEAKERS D3 AMPS

LFQ

RFQ

CRQ

LBQ

LFQ

WQ

WQ

POWER AMPS

POWERAMPS

f-.

~~

Figure 6c. Detailed circuit of resistor matrix +shelf filter stage.

Figure 6d. Detailed circuit of output stages,including loudspeaker distance compensation,loudspeaker layout control and amplitudematrix.

Figure 7. Three or four of the outputs fromfigure 6 may be fed to power amplifiers todrive four- or six-Ioudspeaker layouts. The'+' sign indicates the positive loudspeakerterminal in each case.

~

6 SPEAKERS4 AMPS o

9833.70

6SPEAKERS O4 AMPS

+ +REGULARHEXAGON

9833.7d

7c

RFQ

LFQ

WQ

CBQVPOWERAMPS