Floppyshop 2/Floppyshop - 2.zip/Floppyshop - 2.iso/diskmags/3565-4.665/dmg-3609/news.txt/cd_hist.asc

Text File | 1989-04-05 | 21KB | 399 lines

                        ##########################
                        #                        #
                        #      COMPACT DISCS     #
                        #      ¯¯¯¯¯¯¯¯¯¯¯¯¯     #
                        #        A HISTORY       #
                        #        ¯¯¯¯¯¯¯¯¯       #
                        ##########################


I  recently  read  somewhere that Atari were holding  off  from  releasing
their  CD ROM drive until a standard could be established.  Hasn't  anyone
told  them  there  is  already  a  standard  available,   and  being  used
vigourously in the PC world?.

This  alone should convince them that there is no point in waiting for  an
alternative now.  IBM has already set it.  Remember the battle between  3"
and  3 1/2" discs?  What happened to 3" discs/drives when Big Blue  choose
the 3 1/2" version?

Anyway, I went trawling the BBSs  and found this article on the history of
CD  development.  Don't  know who wrote it but thanks if you  are  reading
this.

                                                            Dave Mooney

                              ~~~~ sof ~~~~


Note of Success comes from CD
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Philips has made more than its fair share of mistakes during the past  ten
years.  Laservision  video  disc  came too  late.  Metal  audio  tape  was
announced  too early and withdrawn because of technical problems.  By  the
time it was ready for relaunch, the trade and public had lost interst.

The  N1500  video  recorder came too early and could only  record  for  an
hour.  Its successor,  the long play N1700 VCR,  came too late to  compete
with VHS and Beta.

V2000 video came even later and died a spectacular death.  The relaunch of
Laservision  as  CD  Video  has failed,  too.  The  early  promise  of  CD
Interactive.  a mixed media disc,  has never been honoured.  That, too may
now arrive too late to succeed.

These  disasters tend to obscure the fact that Philips changed  the  world
in  the 1960s with the compact audio cassette and has changed it again  in
the 1980s with compact disc audio.

Although  Philips  granted licences on the compact  cassette  and  policed
them  to maintain a world standard,  the Dutch company never made a  penny
out  of  royalties.  The Japanese only adopted the system  in  return  for
royalty free licences.

But  the  situation  on cd is different.  Philips took Sony  on  board  as
codedeveloper,  screwed the hardware and software industries to the ground
and  now  earns a royalty on every disc and player made  anywhere  in  the
world.  Instead of sitting back and counting the free money that rolls in,
Philips continues to develop the system.

The latest development,  by far the most radical since cd was launched  in
1982/83,  is  called bitstream.  This is a grassroots rethink on  the  way
digital  signals  read by laser from a cd are  decoded.  Several  Japanese
companies,  for  example Technics,  are working along similar  lines.  But
Philips appears to be in the lead.

Music  is  recorded on a cd in digital code.  This code is  made  up  from
words,  each  containing 16bits of information.  Each word  describes  the
original  analogue  signal  which has been measured  or  'sampled'  44,100
times a second.

So  far  cd players have done the logical thing and  converted  the  16bit
digital words back into an analogue waveform.  Even though the digital  to
analogue converter is working on 44,100 measurements a second,  the result
can  never  be  a smooth waveform:  it is stepped  like  a  staircase.  An
analogue filter must be used to round off the sharp edges.

Initially it was thought that this filter must have a very step curve,  to
slice  off  spurious noise signals above 20kHz,  while leaving  the  music
below 20kHz untouched.  But steep filters tend to distort the phase of the
music signal.  So Philips developed a technique called oversampling.  Each
digital  sample  is processed several times,  to spread the noise  over  a
wider frequency range and allow the use of a more lazy filter curve.

The  Japanese followed the Dutch lead and oversampling is now  a  standard
trick in cd players.

But a problem still remains.  A conventional 16bit dac has to process some
bits  with an accuracy 65,536 times greater than for other  bits.  However
carefully the dacs are made and adjusted at the factory,  and however many
times  they  oversample the digital signal,  they are still  working  with
16bit binary code words.

In any digital system that relies on binary code words,  some bits in each
word  will  be much more 'significant' than others.  Any errors  in  these
most  significant  bits  (msbs) can make the stepped  output  signal  jump
wildly up or down, introducing nasty random noises into the music.

Each  bit  has equal significance and together they describe the  way  the
signal is changing.  Up a little,  down a little, down a little more, up a
little and so on -rather like instructions to a car driver backing into  a
tight corner.

Because  all  bits  have  the  same  'significance',   errors  are   never
catastrophic  -  they just steer the signal slightly off  course  and  the
mistake is easily corrected.

Obviously,  the faster the bits stream,  the more accurately they describe
the signal.  BitStream does 256 calculations on each of the 44,100 digital
words available each second.  This creates a stream of single bits running
at 11.2MHz, more than 11 million a second.

Taken  together,  the density of the bits paint a highly accurate  picture
of the waveform,  because each bit is a snapshot of the signal taken at  a
very brief moment in time.

Philips will start using BitStream chips in the most expensive cd  players
this winter,  and by 1991 all Philips digital audio equipment will rely on
them.  By  the  mid  1990s all the electronics for a  cd  player  will  be
integrated into a single BitStream chip.

                                ~~ o0o ~~~

One bit wins the decoding race
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

The 'bit race' is officially over.  The four and eight times  oversampling
systems  used  by  CD players are now 'out',  and so are  18  and  20  bit
decoding.  One bit decoding is the 'in thing'.  The break with 6 years  of
CD  tradition  has arisen because the Japanese are offering a  single  bit
decoder  that  rivals  the  Bitstream  system  designed  by  Philips.  The
Japanese  system  is  called multistage  noise  shaping  (mash),  and  was
developed by Nippon Telephone and Telegraph.

Bitstream  and  mash do not directly convert the stream of  16bit  digital
words read from the disc into an analogue waveform by digital to  analogue
converters  (DACs) usually made by US company,  Burr Brown.  Instead  they
both   convert  the  words  into  a  more  rapid  stream  of  single   bit
instructions.

Whereas  Bitstream fashions the bits into bunches which vary  in  density,
mash bits vary in width.  While conventional DACs are referred to as pulse
code  modulation devices (PCMs),  Bitstream uses pulse density  modulation
(PDM), and mash pulse width modulation (PWM).

Supplier  Technics admits that mash has been ready for more than  a  year,
but the company has sat on it for fear of creating confusion.  Now Philips
has  pre-empted Technics with the Bitstream launch.  Technics must  follow
suit with mash.

All   the  major  Japanese  companies  are  following  with   single   bit
processing.  Most will buy mash chips from Technics,  although Sony  looks
likely to favour Bitstream - not good news for Burr Brown.

According  to  Tadashi  Abe,  the  man in  charge  of  CD  development  at
Technics:  "Mash  is very difficult to explain - especially after so  much
talk about oversampling and 20bit decoding.  But it is now well understood
in  Japan and suddenly popular.  We will now try to explain it in  Europe.
We  have  to because mash will eventually take over from  'old  type'  DAC
decoding.

"We  started using mash in some CD players last October but did  not  tell
anyone. Now we are using it in all but two of our new CD players.

Actually  we  wanted to introduce mash one and a half years  ago,  but  we
were fighting the 18 and 20bit battles - the idea was opposed.  The  sales
force did not understand the system.  Thanks to Philips, which has started
to promote the idea of one bit processing, we can now join in."

Seiji Kaji,  director of Technic's audio division,  added: "We now want to
use mash technology for all digital equipment, amplifiers and so on."

Firing  the  first  shot in the  single  bit  battle,  Technics'  Japanese
engineers have produced a simplified explanation of the system,  which  is
far clearer than the muddled Bitstream documentation from Philips.

Technics'  simple  explanation  stems  from  an  analogy  with  water:  it
compares  conventional  pcm  16bit decoding with filling  a  bucket  using
water from 16 different sized cups.  In theory each cup's size is an exact
multiple of the others:  in practice the sizes are never quite  right,  so
each decoded sample shows an error.

This  means  the system is non linear as different  steps  have  different
errors,  resulting  in  audio distortion.  Technics prefers  to  fill  the
bucket  many times using one cup only.  That way all cupfuls  are  exactly
the same size and any error is linear:  the bucket either fills too slowly
or too fast, which is equivalent to an overall gain error in audio.

The  clearest demonstration of non linearity can be given with  a  digital
recording of music,  recorded at a very low level so that background noise
becomes  audible.  With PCM decoding the music distorts as quite  passages
fall into the noise floor. With pdm and pwm there is less distortion.

In a mash decoder a digital filter looks at each 16bit word many times  to
oversample,  or  more accurately resample it.  Mash currently works at  32
times rate,  but will later work at 64 times and eventually 256 times when
chips are available which can work fast enough to do all the  calculations
involved.

At 32 times oversampling,  each of the 44 100 words per second is examined
32 times.  The filter also looks at the words which came before and  after
the word under examination and splits the difference between them into  32
slots.  In this way the filter can construct a smooth curve between words,
like  a graph with many lightly registered points between the main  points
on the curve.

This  has  the  beneficial  effect of  spreading  the  random  noise  that
pollutes  any digital signal.  It is caused by chopping the waveform  into
discrete  segments over a very wide range.  The noise in the audible  band
is reduced to one thirty second of its original value.

Next  a  noise shaper (the part of the system which is actually  the  mash
circuit)  pushes  the noise towards the high frequency end  of  the  range
reducing the amount of noise left in the lower frequency audible band.

At  the  same time all the many samples are circulated in  a  mathematical
calculator  loop,  which  works  like the negative  feedback  loop  of  an
analogue  amplifier.  It  calculates  the  value  of  each  sample  in  an
interactive process.

A  rapid stream of values describing the signal very accurately comes  out
of  the  loop.  With  more values there is improved  accuracy  -  but  the
circuit must work faster,  bringing a cost penalty because consumer  chips
have a speed ceiling.

Each  sample  is represented by a pulse of fixed level but the  width  can
vary between 11 values,  which is equivalent to a 3.5bit word. This sounds
a  course  measurement,  because at 32 times the normal  44.1kHz  sampling
speed  the pulses are streaming so fast they work together  to  accurately
describe the value of the signal.

Because  all the pulses have the same level,  and their speed  is  rigidly
locked by a quartz clock, there is no risk of non linear distortion.

As  a final step the pulses are easily converted into an  analogue  signal
by passing them through a filter which rolls off high frequencies.

                               ~~~ o0o ~~~


'No good reason why we can't commercialise this'
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Industry experts were stunned.  Tandy, a company best known for a chain of
cut  price electrical stores,  was claiming to have developed an audio  cd
player that could record and re-record blank discs. It cost just $500.

Competitors  were  sceptical  and noted Tandy said it would be  18  to  24
months  before  it  launched the product.  But the  company  was  defiant.
"There is no good reason why we can't commercialise the technology,"  John
Roach, Tandy's chief executive, told the press.

Four years later, and Tandy is still finding reasons   why    it    cannot
commercialise the  technology,  at least not within the time it  promised.
So the task of bringing cd recorders to the market has been left to  other
manufacturers.  And although the several products launched in the last few
months are priced for the professional market,  cd recorders could one day
become affordable to consumers,  if only those in the top end of the hi-fi
market.

Unlike  Tandy's  dye polymer technology,  which promises cds that  can  be
written and erased hundreds of times,  the recently launched cd  recorders
use  a write-once medium.  And while Sony is promoting  its  magneto-optic
mini  disc as a rewritable cd,  due to be launched this  year,  the  discs
cannot be played on a conventional cd system.

The current generation of cd recorders began to reach the UK in the  first
half  of  last year,  with Yamaha selling a handful of  its  12,000  pound
machines to the broadcast industry.

Although  that price has dropped to 8000 pounds,  it has taken Philips  to
bring the cost down to a more realistic level.  This year, the company has
begun  manufacturing  recorders under its own trade names and  badged  for
other companies.

This  led to a flurry of recorders reaching the market including  a  3,000
pound  model badged for Micromega of Paris,  a 3,500 system from  Marantz,
the  trade name Philips uses for its top end audio equipment and  machines
from Mission, Studer and Meridian.

Blank cds cost 29 pounds and the recorders are being marketed for  storing
radio jingles,  sound effects and other applications that require  digital
sound  and  random  access.   Matsushita,   Kenwood  and  Ricoh  are  also
developing recorders or have launched them in the US and Japan.

Write-once cds have,  ofcourse,  been available to computer users for some
years.  But  the  failure of manufacturers to agree on  a  single  optical
technology and a set of specifications means a disc written on one  system
can  rarely  be  read by another.  With audio  cds  so  well  established,
manufacturers  have ensured their recordings are compatible with  existing
players.

There is,  nevertheless,  a move to standardise the recording process  and
ensure  blank  and  partly  written  discs  can  be  interchanged  between
recorders.

Last October,  Philips and Sony,  joint inventors of audio cd, linked with
Meridian Data,  Ricoh,  Teac,  Sun Microsystems,  Kodak and other computer
companies  to form the Frankfurt Consortium,  with the aim of promoting  a
standard called cd-recordable (cd-r).  Its specifications are enshrined in
the Orange Book,  a superset of previous cd standards. This means cd-r can
read and write not only audio cd, but also cd-rom, cd rom-xa and cd-i.

"Our objective is to gain international acceptance for cd-r standard  that
ensures  complete compatibility for the end user," said  Frederick  Meyer,
the consortium's spokesman.  "Such a standard will enable the end user  to
place  a  disc  in a cd-r drive knowing it will  work  regardless  of  who
manufactured the drive or who manufactured the disc."

Just  as Philips makes most of the cd-r machines,  so are the blank  discs
made by a single manufacturer,  the Japanese giant Taiyo Yuden.  It,  too,
allows  companies  such as Philips and Yamaha to resell them  under  their
own names.  Kodak is also putting its name on the gold coloured discs  for
the  Photo CD system it has launched.  TDK has recently developed its  own
discs, but these are still being evaluated for cd-r specifications.

Also  developing  discs  are  Ciba-Geigy  and  the  Cambridge  based  form
Plasmon.  When  Plasmon approached the Swiss chemicals group for  improved
coatings for the optical discs it manufactures,  it discovered  Ciba-Geigy
was  already  working  on a disc that was far  more  stable  than  current
technology.  So the companies formed a joint venture to produce recordable
cds guaranteed to retain data for up to 50 years.

Bob Longman,  Plasmon's technical director,  said tests show uv light  can
cause  some  discs to degrade after as little as three  years,  which  can
cause a loss of data.  The venture aims to have mass production  beginning
at  Plasmon's UK factory in December.  "It is a major breakthrough,"  said
Longman.

The  companies  are  saying little about the composition  of  the  surface
chemicals they are developing,  but physically the discs are identical  to
Taiyo-Yuden's, ensuring they are compatible with the cd-r specification.

Blank  cd-r  discs have a transparent plastic substrate  engraved  with  a
spiralling  grove to guide the laser.  This is covered with a special  dye
that  forms  the recording layer.  On to this is laid  a  gold  reflective
layer with a plastic cover forming a protective layer.

To  write  data,  the  laser  focuses through  the  substrate  on  to  the
recording  layer,  heating  a small spot to  about  250  centigrade.  This
causes the dye to melt,  and the substrate swells to fill the space in the
recording  layer.  In this way the laser effectively burns a hole  in  the
dye and builds up a series of pits corresponding to a prerecorded cd.

The  way  the  data is arranged differs from a  conventional  cd  in  that
additional   information  is  written  in  the  innermost   tracks.   This
information  consists  of  a  program memory  area  (pma)  and  a  program
calibration area (pca).

As a recording is made,  the track numbers and their start and stop points
are written to the pma.  Parts of the recording,  such as faults,  can  be
marked invalid in the pma, which will be skipped when the disc is played.

The pca is used to test the optimum power the laser needs to write,  which
for  any one disc can be between 4 and 8mw.  This value is then stored  in
the pca and read  before every recording.  When the user decides that  the
recording  is  complete,  the machine writes a table of contents  into  an
area  corresponding  to the lead-in of a conventional cd.  A  lead-out  is
also  written  at this time to designate the end of  the  recording.  Only
then will the disc work on a conventional cd player,  so a music  producer
cannot,  for example,  take a partially recorded disc home for the weekend
to play on his hi-fi.

One  of  the reasons the 8,000 pound Yamaha player is more than  twice  as
expensive  as its competitors is that it effectively updates the table  of
contents as the disc is filled,  rather than waiting until the the end  to
write it. This ensures the cd can be played on a normal system any time.

Cd-r  is a complicated process,  which is why the machines cost  so  much.
And  as  some of the most sophisticated parts are mechanical  rather  than
electronic,  the  price is unlikely to fall as fast as early  cd  players,
said Chris Barron,  technical service manager of HHB,  the UK  distributor
of the Marantz recorder.

"Technologically,  it's  not possible to achieve the level of  pricing  of
consumer  units,  because of the amount of processing and  servo  assisted
laser  control  that  are required to achieve the  correct  recording  pit
size", he said.

"Compatibility  is  very  important.  If the pit size  is  not  absolutely
correct the disc will not play back on many machines".

Furthermore,  Barron  does not see the price of blank discs falling  below
14  pounds  in the next two years,  because of  the  inherently  expensive
manufacturing process.

So perhaps the greatest hope of a mass market cd recorder still lies  with
Tandy.  The latest word from the company is that it will introduce a write
once machine this year "at an affordable price".

As  for the $500 erasable system,  don't hold your breath.  "It will be  a
good product for the 1990s," said Ed Junge,  marketing director.  "We hope
it will be the early 1990s".


                              ~~~~ eof ~~~~