Kees A. S, S. thattookplaceover15 yearsagobe- _E LP six years before tweenengineersat PhilipsandSony. : Schellackdisc H Laservision was

Transcription

1 Kees A. S, S The INTRODUCTION We threw around I shouldsayat theoutsetthati amnot DVD the deaof a spral a hstoran but an engneer, and that CD track wth bts whati wllsaycomesfrommy per- or pts, and from sonalrecollectonsof the meetngs that tme t took thattookplaceover15 yearsagobe- _E LP sx years before tweenengneersat PhlpsandSony. : Schellackdsc H Laservson was Even so, lke the best hstorans, I 'ylnder RJt born. In 1973 I shallgveyou to 100yearsof the launch dskrecordngthatledup ofathe brefsummaryof CD. the _ II II [---'_ It... '] electroncs,and consdered startedworkngon servo systemsand that we f A BIT OF HISTORY one could store vdeo nformaton As you wll see from Fg. 1, the equvalentstoragedenstyof recordngme- Fg. 1. Growth of storage densty of recordng meda. optcallythen one da n bts/mm 2 grew gradually from could also manage that of the orgnaledsoncylnderto audo. The management at the tme the recent DVD. We are now dealng _ ' dsmssed ths dea sayng that audo :;;, wth storage denstes some one ml- I _187'7Phonog?aPh was far too smple and not worth the effort,so weleftt alonefor the tme lon tmes greater than those avalable I 1887Fat Dsc beng. And then came the launch of at the end of the last century. Fg. 2 ' /3 rpm LP Laservson n 1975, whch was an unshows some key ponts n the evoluton _ : of audo dsks. One must remember 1958 Stereo LP expected flop. It flopped n a bg way, that t was at least 10years aftered- _ and of some 400 players that were son's cylnderbeforeberlnerand oth- ' sold some 200 were returned because ers came along wth a method of dsk Fg. 2. Key ponts n audo dsk evoluton, buyers had been under the msapprerecordng. In fact, I vsted Edson hensonthat t wouldalso recordpro- Labs and was told that all record!ng the 19505, the stereo LP record delv- grams--of course t could not. After devces up untl about 1920 were ef- ered a major step n sound qualty, but two years the Phlps management defectvely prototypes and that there was as far as optcal recordng s concerned cded to wthdraw t. no way of replcatng meda, so to have everythng before 1969 must be called a recordng ht at that tme must have prehstory. EARLY DIGITAL AUDIO been qute an achevementi thnk!in It s my vew that dgtal audo was OPTICS AND LASERVISION made possble at that pont, and be- *Presented at the 103rd Conventon of the In 1969 a colleague of mne suggested tween 1970 and 1980 numerous meet- Audo Engneerng Socety, New York, that t mght be worthwhle developng ngs were held between Phlps and NY, USA, 1997 September the use of optcs for storng pctures. Sony. In order to understand the tech- 458 J. Audo En 9. Soc., Vol. 46, No. 5, 1998 May

2 Kees presents a re engameer's vew of the ycars 1 up to the launch of the CD, and the crucal decsons that were made durng that tme would determne the techncal success or falure of thc medum. From a personal pers pectve he comments on the hstory of the frst mass-market dgtal audo medum and delvers some predctons for the future of optcal meda n data storage., only wayof star- Theproblemwththswasthatt was ng perhaps one not really much more mmune to drt Manufacturer Type ts Resoluton hour of sound at and scratches than an analog LP, althat tme. though there was a certan mprove- BBC Mcrowave Toward the end ment n soundqualty, so they decded of the 1970s PCM to look for a dgtal soluton. In Soundstream tape 37.5/ adapters were de Phlps and Sony both veloped, whch demonstrated the frst prototypes of a Many PCM adaptor /16 were dgtalrecord- dgtal sound system usng a laser ng processors that dsk, and n 1979 a hgh-level decson Denon tape used ordnary ana- was made to jon forces n the devel-... log vdeo recorders opment of a world standard. Phlps as a means of star- had lost the market for Laservson, ng dgtal audo but had consderable optcal expertse, Fg. 3. Early samplng-frequency/resoluton choces, data, snce vdeo as well as expertse n serve systems recorders were the and dgtal and analog modulaton sysncal choces concernng sample reso- only devces wdely avalable wth terns. Sony's huge expertse n error luton and such at that tme, one must suffcent bandwdth. Ths helps to ex- correcton, PCM adapters, and n look to what was happenng wth sm- plan the choce of samplng frequency channel codng would complement de equpment and wth dgtal audo for the CD, because the number of ths deally. A reasonable summary n broadcastng (Fg. 3). The BBC n vdeo lnes, frame rate, and bts per would be that most of the "physcs" the UK was one of the frst to adopt lne end up dctatng the samplng fre- was provded by Phlps and the dgdgtal sound, and they were workng quency one can acheve f wantng to tal audo experence by Sony. wth a 13-bt dstrbuton system for store two channels of audo. The samthe transmtter network, sampled at 32 plng frequences of 44.1 and THE SONY-PHILIPS LIAISON khz. Stockham's experments n khz were thus the result of a need for In a number of meetngs wth the Soundstream system compatblty wth the NTSC and PAL were held n both Tokyo and Endnvolved sample rates averagng vdeo formats used for audo storage at haven. The frst, n August 1979 n around 40 khz. He used computer the tme. Endhoven, and the second, n actatape as the means of storage, whch ber '79 n Tokyo,provdedan oppormade t practcal to store data at 16-bt AN OPTICAL AUDIO DISK tunty for engneers to get to know resoluton because of the byte-orented From 1973 to 1976 two Phlps eng- about each other and to dscover the nature of computer data storage. Eght neers were gven a mandate to develop man strengths of each team. There bts would have been too few, and 14 an audo dsk based on optcal tech- was a great deal to learn from each bts would have meant neffcent use nology, and they started by exper- other. Both teams had workng prateof the storage space, so 16 bts seemed mentng wth an analog approach us- types and decsons had to be made logcal. Computer tape was really the ng wdeband frequency modulaton, concernng modulaton and error- _, J. Audo Eng. Soc., Vol. 46, No. 5, 1998 May 459

3 correctonsystems.parallelexper- m lent choce of 16-bt mentswereconductedn both loca- resoluton, although tons,andnaturallytherewerenumer- Phlpshad developeda ous occasons on whch each team felt 14-bt D/A converter at t hadthe bestor mostpractcalsolu- thetme,leadngphlps ton--ethertherscouldcorrect the to arguentallythatt longestbursterrorsor gvethelongest was mpossble to replayngtme,and so on. Thedelcate desgntsconvertersfor prototyp electroncequpmenthad to 16btsna shortenough be transported wth the engneers tme.butmy colleague acrossthe world, and consequently Carel Djkmans sad, traveledfrstclassn a separateseat "Noproblem.I knowa bookedespecallyfor t. The arlne smalltrckto turna 14- KLMreally loved us for that, be- bt converternto a 16- Cause,as you knbw_boxes of elec- bt converter--t's troncsdo not drnkchampagneor ask called oversamplng," formorefood! Sowemanaged tosolve ByMay1980almosteverythngwas thatproblemquteeasly n order. The modulaton system was Fg, 4. Choce of CD dameter was based on sze of as t happened. As a restll a pont of contenton, wth each CompactCassette. sult, all the frst Phlps partystllclamngonewasbetterthan players had oversamthe other, and then there was a phone traded off to optmze playng tme png converters n them. call from the current charman of and relablty. The decson was made The cross nterleaved Reed- Sony, Mr. Ohga, who told us that f we by the top brass of Phlps. "Compact Solomon code (CIRC) chosen was could not make a decson wthn a Cassette was a great success," they much better than that proposed by week, management would make t for sad, "we don't thnk CD should be Phlps, although extremely complcatus. I was tempted to say that f they much larger" (Fg. 4). As t was, we ed at the tme. Sony was proposng uswere able to make a decson so easly, made t 0.5 cm larger. (There were all ng 16-kbyte RAMs for nterleavng, why dd they not do so sx months sorts of stores about t havng some- whch would have cost around $50 to ago? but refraned n the nterests of thng to do wth the length of us, and added sgnfcantly to the comdplomacy. So we moved quckly and Beethoven's 9th Symphony and so on, mercal cost of the players. "Are they made all the decsons concernng the but you should not beleve them.) crazy, those Sony guys?" asked many mechancal specfcaton of the dsk The samplng frequency of 44.1 khz at Phlps, but n fact the prce of RAM and soforth, was decded. It wasthe only chocewe and ts capacty have changed so could make after long delberaton, and quckly that t s mpossble to buy less DECIDINGTHE PARAMETERS the selecton of 44.1 khz as opposed to than about 1-Mbyte RAMs today, and The dsk dameter s a very basc pa was smply for the reason that for less money than 16 k then--so fast rameter, because t relates to playng t was easer to remember--there was have we moved. So that turned out to tme. All parameters then have to be no other mason. Sony made the excel- be a good decson. The 8-14 bt channel code was m agreedandallthespecfcatons between them led to a playng tme of 75 mnutes. Geometry and the other physcal parameters were decded, and a plastc layer coverng the pts was added (a basc paradgm of all optcal dsks) to protect the data surface from damage and to nsure that drt and scratches on the surface were well out of focus for the laser pckup (Fg. 5). (I should note that the recent DVD has lost a small part of ths paradgm, as t s only 0.6 mm thck, and so the optcal pckup wll be more vulnerable to drt and scratches. Modern error-control codes can easly cope wth the result- Fg. 5. CD layers, ncludng plastc layer to protect data surface and keep drt and ng errors.) Just to show how :mportant scatches out of focus of laser pckup, the AES was, and I thnk always 460 J. Audo Eng. Soc., Vol. 46, No. 6, 1998 May

4 _:... _?v_thc Dsc!... :: z ::: wll be, the CD system ncludng modulaton and error correcton was descrbed by some of the engneers n- analoguesgnal 24bytes volved n a conventon =1frame EFMsync pattern preprnt publshed n 1980 n left =6 samples 27 bts Cross-Interleaved New York [1]. Fg. 6 shows rght Reed-Solomon code the block dagram of the encodngmodel used for CD, PCM,44.1kHz 24-btsync+ 588bts resultng n the well-known 16btspersample 3-btmergng 4.3-Mbt/s data rate. per channel Idata T.E A.D OD //.8 framesof 32bytes So CompactDsc was born and, as you can see from Fg. CD-ROM 124 byteseach 1byte -- '-_-_ --_ 561bts 7, the audo dsk market sectors of 2352bytes control & dsplaynfo 8-to-14 code +3-bt mergng shows a gradually declnng sale of LP and an exponental... subcodechannels,groupedn 98-byteframes ncrease n sales of CD. The sale of players also clmbed sharply (Fg. 8). Personally, I Fg. 6. Block dagram of CD encodng model resultng n 4.3-Mbt/s data rate. _: CD '" LP O ". O '""... '-.., "., 20.. I r/', _,, 1982 '83 '84 '85 '86 '87 '88 ' '83 '84 '85 '86 '87 '88 ' Fg. 7. Sales comparson of LP and CD between 1982and1990. Fg. 8. Sales comparsonof turntable and CD players between1983and1990. was not at all sure that CD would the problems succeed, that as Iarose had wth seen CompactDsc Famly I Laservson, buti amhappy to seethat t s expected some CD-Audo CD-ROM CD-Recordable 5 bllon CDs wll be sold n RedBook YellowBook OrangeBook I the world n 1997, whch I I I L... L... makes t a very successful CD-ROMXA CD-MO CD2W0 CD:E market. (I saw also that 100 CD- YellowBook OrangeBook orange Book 0rangeBook: bllon hamburgers would be GreenBook Mode 2 Part I Part II part II I sold n the U.S. ths year as I : I - -- : well,so perhapst s not so Enhanced bg after all...) Now we have CD-Brdge Musc CD : numerous other "books" n / BlUeBook. Multsesson CD the CD standard,wth CD- r ROM players outsellng CD VdeoCD : Phot0CD :. WhteBook [. t Audobyabout70%to30%, - : whch s somethng! never expected (Fg. 9). _,, Fg. 9. Expanson of CD-based systems. 462 J. Audo Eng. Soc., Vol. 46, No. 5, 1998 May

5 7OO% 600% 500% 4OO% 300% 200% 100% xl "_... reference wavelength NA margns EFM- CIRC- othergans (CD) EFMPlus RS nosubcode no 3rd layer CRC I Optcs Desgn _ Drve and Precson _:_Decodng Electroncs Fg.10. Dfferent characterstcs contrbutng to ncrease n DVD capacty compared wth CD. LOOKING FORWARD Lookng at Fg. 10 t can be seen that a of the lens s only possble f the dsk As a researcher I lke to look forward number of dfferent characterstcs s made thnner, so future mproveas well as back, and I should lke to have led to the ncrease n capacty of ments here wll only be possble f the make some predctons for the future the DVD compared wth the CD. The dsk s made thnner and thnner. Blue of optcal dsk meda, based on DVD. wavelength of the laser and the nu- and green lasers are already possble. Ths new dsk has a capacty some mercal aperture (NA) of the objectve It wll be apprecated that a large part seven tmes larger than CD, and we lens have both been altered, and new of the potental ncrease n capacty s need to look at ths to see how the op- research s makng further advances purely dependent on physcs, not on tcal dsk growth path wll be defned, possble n ths area. The greater NA recordng specalsts. Other advances n capacty have already been made for Table 1. Comparson of DVD and CD-ROM man parameters. DVD. Some thngs can only be done once--for example, removng the thrd layer of CIRC CD-ROM and removng subcode. Reducng the track ptch and other read-outwavelength(nm) 780 physcalmargnshasbeenpossreferencena 0.45 blebecausewe nowknowhow dscdameter(cra) 12 to manufactureand readoptcal dscthckness (mm) 1.2 dskswell.themanparameters layers snglelayer: of thedvdcomparedwththe datacapacty(gbyte) modei:0.68 CD canbe seenn Table 1. Ths modei1:0.78 leads me to make some predctons to the year 2010, concernreference scannng speed (m/s) 1.2 ng the physcal densty of varreference channel bt rate (Mbt/s) 4.32 ous storage meda, as shown n mn.pt(orland)length(gm) 0.85 Fg. 1t. Therecordngdenstys track ptch (gm) 1.6 compared wth CD, whch has a recordng code EFM densty of 1 bt/pm 2 compared sector sze (bytes) 2048 wth the 6-7 bts/pm 2 of DVD. error correcton CIRC Hgh-densty forms of DVD wll max.user bt refspeed (Mbt/s) 1.41 extend ths further, so that n 2002 (probably n May!) we... shall see a 20-Gbyte dsk, and n 2006 (almost certanly n 464 J.AudoEng.Soc.,Vol.46,No._,1998May

6 _F: DVD-ROM4.7GB '_ '_ CD-RMD ,-_ Year... Fg. 11. Predctons of future physcal densty of storage meda. June...) we shall see a 40-Obyte dsk. So that s the end of my story, and of REFERENCES Meantme the growth n DRAM and course t has been based. I am an eng- [1] L. B. Vres, K. A. Immnk, J. G. magnetc dsk capacty contnues to neer and I have told the story from an Njboer, H. Hoeve, T. T. Do, K. Odagrow at an exponental rate, although engneer's pont of vew. Many of the ka, and H. Ogawa, "The Compact the prce of DRAM s currently very decsons concernng the CD were Dsc Dgtal Audo System: Modulamuch hgher than that of optcs, made by very clever marketng people ton and Error-Correcton," presented Whether companes wll market these --for example, the dea of usng a jew- at the 67th AES Conventon of the products s another matter entrely-- el case for storng the dsk. Those dec- Audo Engneerng Socety, J. Audo ths s just a predcton made by an en- sons are not the responsbltes of en- Eng. Soc. (Abstracts), vol. 28, p. 931 gneer, gneers,for verygoodreasons. (1980Dec.),preprnt1674. The author presentsthe CD Story to the 103rdConventon of the Audo EngneerngSocety, New York, THE AUTHOR Kees A. SchouhamerImmnk was born n Rotterdam, The Netherlands,n 1946.He recevedan M.S. degreen electrcal engneerngn 1974and a Ph.D. n 1985from the Endhoven Unversty of Technology, The Netherlands. Dr. Immnk has worked hs entre professonallfe at the Phlps Research Laboratores n Endhoven, where hes currentlya researchfellow. He s alsoadjunct professor at the Insttute of Expermental Mathematcs, Essen Unversty, Germany. He has contrbuted to the desgn and development of a varety of dgtal audo recorders such as the Compact Dsc, Compact Dsc Vdeo, DAT, DCC, and, very recently,the MultmedaCD. Dr. Immnk holds 35 U.S. patents, s coauthorof three books, and has wrtten numerous papers. He was a member of the techncal advsory board of the MagnetcsTechnology Center, Sngapore, s cofounder and trustee of the Shannon Foundaton,and s a dstngushed lecturerof the IEEE CommuncatonsSocety. He served as program and conference charmanof varous nternatonalconferences, and receved many awardsfor hs part n the dgtal audo revoluton. A fellow of the lee and IEEE,he was awarded the lee Sr J. J. Thomson Medal n 1993, and the SMPTE Ponatoff Gold Medal for Techncal Excellence n 1994.Dr. Immnk joned the AES n 1980, and has been a member of the JournalsRevewBoard snce He was a commttee memberof the Netherlands Sectonfrom 1986to 1992, and ts charman from 1988 to He was papers charman of the AES 96th Conventon and papers vce char of the 104th Conventon held n Amsterdam n March 1994 and May 1998, respectvely. He became an AES fellow n 1985, was awarded the Socety's Slver Medal n 1992, andwas electedto ts Board of Governors n J. Audo Eng. Soc., Vol. 46, No. 5, 1998 May 465