This was published by 6Moons as part of a description of the PS Audio PerfectWave components. An illuminating description on analog vs. digital I had never read before. You can read the full article in the link at the end of this post.
CD isn’t really digital: Since 1982 when Philips and Sony introduced the audio CD, much water went down the bridge, not the least of it over the last few years. The CD was originally marketed as “Perfect Sound Forever”. As we know now, that was rather a bit premature. Admittedly, the medium was handy, shiny and compact from the onset and soon became outright ubiquitous. Still, it was handicapped by a number of drawbacks. Mind you, these drawbacks really only bother the audio version of the medium. The data version fares far better as we’ll shortly appreciate.
The disadvantages for the audio CD are for the most part mechanical. Contrary to popular perception, a CD is not really digital. Under a microscope, a CD shows many wells - lands and pits as they are called. If it were a true digital storage medium, it would show a far more regular pattern. Each pit or land would then represent a true 1 or 0. With an audio CD however, we see something altogether different.
Mind you, the CD developers did an incredible job making the CD platform what it is today. Rumors have it that the error-correction mechanism alone took ten years of development. Now to basics. The audio version of the CD has no room for a data re-read. The design relies on real-time read-once data capture. In case of a serious read error where data is completely missing, the error correction mechanism inserts a "guess" into the data stream to fill in the blanks. If the error is too severe, the correction mechanism capitulates and your speakers produce a ticking sound. Your CD is too badly damaged or filthy. Next.
So if what’s on a CD aren’t zeros and ones, what exactly does represent the analog waveform? For starters, the physical pattern includes only a part of the signal. Next, even that part is a mere representation of the original, not the actual thing. To wit, a CD contains only the amplitude information of the recorded signal. To recreate the signal in its entirety, we also need the time-domain data. We need to know when a signal rises or falls. These are questions the timing or clock information must answer to which forms the X-axis of a music signal’s graphical representation. The audio CD standard assumes a time constant which needn’t be encoded. Incidentally, this also saves disc space. The time component is provided by a fixed clock. By using the same clock frequency during recording and playback, everything is in perfect sync. At each ‘atomic’ tick of the clock, the laser reads synchronized data from the CD. In theory. The idea is beautifully simple but sadly only fit for a perfect world.
And we clearly don’t live in such a world. At Philips and Sony, the joint team of CD developers soon realized that coding the real zeros and ones of a music signal posed real problems. The rapidly changing pattern soon proved to not be recoverable in any reliable way, at least not without extremely expensive methods. One problem was the difficulty in reading correct data when a series of 1 and 0 called a word ended with a 1 and the next word began with a 1. How to tell the transition from one word to the next? The team came up with a modulation technique called Eight to Fourteen Modulation or EFM. In the so-called Red Book standard for CD, it is noted that a word may be 8 bits long and so can represent 0000 0000 to 1111 1111. With EFM, that 8-bit word becomes an equivalent 14-bit word. A 14-bit word is written such that it always contains at least two but never more than 10 consecutive zeros. This modulation results in the fact that information on an audio CD is never shorter than 3 bits and never longer than 11.
It is these bit packages which determine the lengths of the pits on a CD. They are designated T3 to T11 according to their lengths. A CD player reads these longer pits and lands far easier than single bits. The pits on a CD are arrayed in a long spiral sequence resembling an LP’s groove. Besides the T3 to T11 pits, there is room for a sub code to provide laser guidance along this ‘groove’. Another sub code contains motor instructions for speed adaptation After all, a CD does not run at constant speed. When reading near the center hole, the CD spins at 495RMP. When reaching the outer edge, it slows down to 212RPM. This is necessary to obtain the exact constant read speed of 1.2 meters per second.
While reading a CD, a raging stream of pits of variable lengths whirs past the laser reader whose response currents in the light diodes registering the reflections from the CD are completely analog in nature. Each time a pit transitions to a land or vice versa, it gets interpreted as a digital 1. Each time—here’s that clock again—a clock tick registers no transition, a 0 is read. The CD thus contains only symbolic zeros and not singular 1s per se (nor true 0s for that matter).
It must be clear by now that the timely detection of a pit transition resulting in a digital 1 is a matter of critical scrutiny. Even the smallest deviation in detecting the precise transition results in timing errors called jitter. Causes of jitter are many. It could be bad pressings where the pit walls aren’t perpendicular to the surface but crooked. Other optical problems prevent the laser beam to properly hit the reflective layer with the pit pattern through the polycarbonate. This could arise when a CD is still greasy from the stamping process. Other optical problems are the result of light scattering in the plastic layer rather than getting a clear lock. Wobbly or eccentric discs are also detrimental to correct data capture as the laser needs to physically adjust hundreds of times a minute to track the erratically spiraling pits. Poor power supply design compromises matters further as the laser’s motor power is diverted by other circuits. The large currents of the laser servo (which assists in tracking the pits) can interfere with the tiny response current in the optical read diodes. And so forth.
The upshot simply is utter confidence that calling the audio CD a mostly analog medium is correct. By the way, a data CD for computer use is truly digital by design. It operates with different error correction schemes and sports no spiral groove but is addressable directly to any given word and capable of re-reading certain sectors as needed. Is it any surprise then that the audio market is overrun by ‘after-market’ tools meant to adjust or correct the quality of audio CD data read-in?
Analog CD tweaks, a necessity in an imperfect world: We’ll list merely a few examples to drive home the point that CD is mostly analog by nature. Green markers color the edges of a disc to prevent stray light exiting the disc through its side where it can interfere with the laser’s focus on the actual data. Circumcision/beveling of the outer edge again counters stray light issues. Then there are mats made from plastic, carbon, graphite or even Corian aimed at stabilizing a CD and/or to increase its reflective properties. Various cleaning products champion clean surfaces. Then there’s Nespa "light therapy" where high-intensity photon bombardment alters the polycarbonate’s molecular structure (or eliminates gas pockets) to improve the sound. Then there are the various demagnetizing and anti-static treatments. All have audible effects. The issue is not whether those effects are positive or negative. The issue is their audibility. If 1s and 0s were just 1s and 0s, none of these addresses should make any difference whatsoever. Yet as we well know, they do.
Even playing a CD twice in a row changes things since the second time static charges have built up. We’ve often ripped a CD with EAC to make a copy and burn that at the slowest speed on an excellent MAME Gold CDR blank. While time consuming, the improvements can’t be argued with. In short, a CD has so many analog aspects to it that calling the medium digital is a profound error in our opinion. Perpetrating it invites grave misunderstandings which are epitomized in the throwaway line “it’s just ones and zeros”. Far from it.
https://audiophilestyle.com/forums/...ng-up-xxhighend/?tab=comments#comment-1131638
CD isn’t really digital: Since 1982 when Philips and Sony introduced the audio CD, much water went down the bridge, not the least of it over the last few years. The CD was originally marketed as “Perfect Sound Forever”. As we know now, that was rather a bit premature. Admittedly, the medium was handy, shiny and compact from the onset and soon became outright ubiquitous. Still, it was handicapped by a number of drawbacks. Mind you, these drawbacks really only bother the audio version of the medium. The data version fares far better as we’ll shortly appreciate.
The disadvantages for the audio CD are for the most part mechanical. Contrary to popular perception, a CD is not really digital. Under a microscope, a CD shows many wells - lands and pits as they are called. If it were a true digital storage medium, it would show a far more regular pattern. Each pit or land would then represent a true 1 or 0. With an audio CD however, we see something altogether different.
Mind you, the CD developers did an incredible job making the CD platform what it is today. Rumors have it that the error-correction mechanism alone took ten years of development. Now to basics. The audio version of the CD has no room for a data re-read. The design relies on real-time read-once data capture. In case of a serious read error where data is completely missing, the error correction mechanism inserts a "guess" into the data stream to fill in the blanks. If the error is too severe, the correction mechanism capitulates and your speakers produce a ticking sound. Your CD is too badly damaged or filthy. Next.
So if what’s on a CD aren’t zeros and ones, what exactly does represent the analog waveform? For starters, the physical pattern includes only a part of the signal. Next, even that part is a mere representation of the original, not the actual thing. To wit, a CD contains only the amplitude information of the recorded signal. To recreate the signal in its entirety, we also need the time-domain data. We need to know when a signal rises or falls. These are questions the timing or clock information must answer to which forms the X-axis of a music signal’s graphical representation. The audio CD standard assumes a time constant which needn’t be encoded. Incidentally, this also saves disc space. The time component is provided by a fixed clock. By using the same clock frequency during recording and playback, everything is in perfect sync. At each ‘atomic’ tick of the clock, the laser reads synchronized data from the CD. In theory. The idea is beautifully simple but sadly only fit for a perfect world.
And we clearly don’t live in such a world. At Philips and Sony, the joint team of CD developers soon realized that coding the real zeros and ones of a music signal posed real problems. The rapidly changing pattern soon proved to not be recoverable in any reliable way, at least not without extremely expensive methods. One problem was the difficulty in reading correct data when a series of 1 and 0 called a word ended with a 1 and the next word began with a 1. How to tell the transition from one word to the next? The team came up with a modulation technique called Eight to Fourteen Modulation or EFM. In the so-called Red Book standard for CD, it is noted that a word may be 8 bits long and so can represent 0000 0000 to 1111 1111. With EFM, that 8-bit word becomes an equivalent 14-bit word. A 14-bit word is written such that it always contains at least two but never more than 10 consecutive zeros. This modulation results in the fact that information on an audio CD is never shorter than 3 bits and never longer than 11.
It is these bit packages which determine the lengths of the pits on a CD. They are designated T3 to T11 according to their lengths. A CD player reads these longer pits and lands far easier than single bits. The pits on a CD are arrayed in a long spiral sequence resembling an LP’s groove. Besides the T3 to T11 pits, there is room for a sub code to provide laser guidance along this ‘groove’. Another sub code contains motor instructions for speed adaptation After all, a CD does not run at constant speed. When reading near the center hole, the CD spins at 495RMP. When reaching the outer edge, it slows down to 212RPM. This is necessary to obtain the exact constant read speed of 1.2 meters per second.
While reading a CD, a raging stream of pits of variable lengths whirs past the laser reader whose response currents in the light diodes registering the reflections from the CD are completely analog in nature. Each time a pit transitions to a land or vice versa, it gets interpreted as a digital 1. Each time—here’s that clock again—a clock tick registers no transition, a 0 is read. The CD thus contains only symbolic zeros and not singular 1s per se (nor true 0s for that matter).
It must be clear by now that the timely detection of a pit transition resulting in a digital 1 is a matter of critical scrutiny. Even the smallest deviation in detecting the precise transition results in timing errors called jitter. Causes of jitter are many. It could be bad pressings where the pit walls aren’t perpendicular to the surface but crooked. Other optical problems prevent the laser beam to properly hit the reflective layer with the pit pattern through the polycarbonate. This could arise when a CD is still greasy from the stamping process. Other optical problems are the result of light scattering in the plastic layer rather than getting a clear lock. Wobbly or eccentric discs are also detrimental to correct data capture as the laser needs to physically adjust hundreds of times a minute to track the erratically spiraling pits. Poor power supply design compromises matters further as the laser’s motor power is diverted by other circuits. The large currents of the laser servo (which assists in tracking the pits) can interfere with the tiny response current in the optical read diodes. And so forth.
The upshot simply is utter confidence that calling the audio CD a mostly analog medium is correct. By the way, a data CD for computer use is truly digital by design. It operates with different error correction schemes and sports no spiral groove but is addressable directly to any given word and capable of re-reading certain sectors as needed. Is it any surprise then that the audio market is overrun by ‘after-market’ tools meant to adjust or correct the quality of audio CD data read-in?
Analog CD tweaks, a necessity in an imperfect world: We’ll list merely a few examples to drive home the point that CD is mostly analog by nature. Green markers color the edges of a disc to prevent stray light exiting the disc through its side where it can interfere with the laser’s focus on the actual data. Circumcision/beveling of the outer edge again counters stray light issues. Then there are mats made from plastic, carbon, graphite or even Corian aimed at stabilizing a CD and/or to increase its reflective properties. Various cleaning products champion clean surfaces. Then there’s Nespa "light therapy" where high-intensity photon bombardment alters the polycarbonate’s molecular structure (or eliminates gas pockets) to improve the sound. Then there are the various demagnetizing and anti-static treatments. All have audible effects. The issue is not whether those effects are positive or negative. The issue is their audibility. If 1s and 0s were just 1s and 0s, none of these addresses should make any difference whatsoever. Yet as we well know, they do.
Even playing a CD twice in a row changes things since the second time static charges have built up. We’ve often ripped a CD with EAC to make a copy and burn that at the slowest speed on an excellent MAME Gold CDR blank. While time consuming, the improvements can’t be argued with. In short, a CD has so many analog aspects to it that calling the medium digital is a profound error in our opinion. Perpetrating it invites grave misunderstandings which are epitomized in the throwaway line “it’s just ones and zeros”. Far from it.
https://audiophilestyle.com/forums/...ng-up-xxhighend/?tab=comments#comment-1131638
