Konnichiwa,
Someone deliberated upon his personal pet delusions as to what he thinks should happen.
The FACTS are as suggested (from the Philips SAA7327 Datasheet):
So, please get your facts right. I have corrected you at least ten times on the subjects actual facts and suggest you remedy your ignorance of them before posting again, clearly you refuse such good advise. It seems that your insistance on the results of you counterfactual thinking borders on clinical. I repeat, you may wih to seriously evaluate your position in general. I cannot suggest anything more usefull as the moderator suggets my good advise is namecalling, so please look up the older post on more far reaching recommendations.
Nope, the whole point of error concealement is to make them as little audible as possible. The scheme of interpolating missing samples from adjecent samples is quite effective. If there are too many interpolations the signal is distorted to a certain degree, not neccesarily one that is notable as drastic audible distortion or skipping, merely as a degradation in quality.
Funny. Please list one specific point where I am wrong in this discussion. ONE will do, but make sure you research it well, you may find I have the full backup for my point ready at hand.
You suspect a lot of stuff and if you worked for police Squad with Frank Drebbin you would still find it very hard to keep your job.
But you will not find them anyway.
I make you an offer. List at least 3 Consumer CD Players (eg not High End gear from Meridian or the other few exceptions) that allow multiple reads and use a large buffer (eg more than 16 to 64 Frames) and I will actually admit that you may actually have a point.
Sayonara
geewhizbang said:
Someone deliberated upon his personal pet delusions as to what he thinks should happen.
The FACTS are as suggested (from the Philips SAA7327 Datasheet):
So, please get your facts right. I have corrected you at least ten times on the subjects actual facts and suggest you remedy your ignorance of them before posting again, clearly you refuse such good advise. It seems that your insistance on the results of you counterfactual thinking borders on clinical. I repeat, you may wih to seriously evaluate your position in general. I cannot suggest anything more usefull as the moderator suggets my good advise is namecalling, so please look up the older post on more far reaching recommendations.
geewhizbang said:
Nope, the whole point of error concealement is to make them as little audible as possible. The scheme of interpolating missing samples from adjecent samples is quite effective. If there are too many interpolations the signal is distorted to a certain degree, not neccesarily one that is notable as drastic audible distortion or skipping, merely as a degradation in quality.
geewhizbang said:
Funny. Please list one specific point where I am wrong in this discussion. ONE will do, but make sure you research it well, you may find I have the full backup for my point ready at hand.
geewhizbang said:
You suspect a lot of stuff and if you worked for police Squad with Frank Drebbin you would still find it very hard to keep your job.
geewhizbang said:
But you will not find them anyway.
I make you an offer. List at least 3 Consumer CD Players (eg not High End gear from Meridian or the other few exceptions) that allow multiple reads and use a large buffer (eg more than 16 to 64 Frames) and I will actually admit that you may actually have a point.
Sayonara
If it had a buffer it would not be used for re-reads of the disk, only for jitter suppression like in the ML.
Those things are for Discmans !
/
Those things are for Discmans !
/
So there is error correction on CD:s, and the signal to the DAC does not come synchronuosly from the laser, but passes through a FIFO.
The question then is: Are the error rates on good CD:s high enough that reasonable vibration could cause the error correction to fail? I don't know.
And if the error correction is sufficient, does jitter from the laser affect sound?
Rune
The question then is: Are the error rates on good CD:s high enough that reasonable vibration could cause the error correction to fail? I don't know.
And if the error correction is sufficient, does jitter from the laser affect sound?
Rune
On this whole subject of CD Audio versus CD Data formats.. My jaw hit the floor on the past couple of pages. The sheer lack of knowledge on how even basic ECC functions by KYW blows my mind.
So far, the only person who's stated things correctly was the one (ALW? sorry, i'm lazy and don't want to reread those pages.. my stomach can't handle it) who referred to them by their proper formats (Red Book, Yellow Book, etc).
How CD's are made:
http://www.ee.washington.edu/conselec/CE/kuhn/cdaudio/95x6.htm
CD Error Correction along with the Alogrithm:
http://www.ee.washington.edu/conselec/CE/kuhn/cdaudio2/95x7.htm
Various CD Standards:
http://www.ee.washington.edu/conselec/CE/kuhn/cdrom/95x8.htm
If these links aren't enough, some googling for the IEC 908 standard can give you more information that validates these links.
The second link covers how Error Correction works and how Error Concealment works. Error Correction != Error Concealment. They are two separate functions. From this page:
"CD players use parity and interleaving techniques to minimize the effects of an error on the disk. In theory, the combination of parity and interleaving in a CD player can detect and correct a burst error of up to 4000 bad bits -- or a physical defect 2.47 mm long. Interpolation can conceal errors up to 13,700 or physical defects up to 8.5 mm long."
Also:
"SO! The final frame (which started at 6*16*2 = 192 data bits) now contains:
1 sync word 24 bits
1 subcode signal 14 bits
6*2*2*14 data bits 336 bits
8*14 parity bits 112 bits
34*3 merge bits 102 bits
GRAND TOTAL 588 bits"
ECC is a function of hardware. This happens in real time, most at the controller level. In order to correct these errors, I would assume that ALL cd chipsets employ at least a small buffer to verify the parity information and then correct it. This would not be audible as it most likely happens as a function of the chipset before the data is output to the next stop in the signal chain.
If someone here has actually worked with CD Chipsets, it'd be great if they'd post something here. If someone here has information contrary to what I've posted, please feel free to correct me. Heck, if someone knows where I can read datasheets on CD Chipsets, please post a link! I did some cursory googles for it and came up empty (too much CD-RW stuff screwing up the results).
So far, the only person who's stated things correctly was the one (ALW? sorry, i'm lazy and don't want to reread those pages.. my stomach can't handle it) who referred to them by their proper formats (Red Book, Yellow Book, etc).
How CD's are made:
http://www.ee.washington.edu/conselec/CE/kuhn/cdaudio/95x6.htm
CD Error Correction along with the Alogrithm:
http://www.ee.washington.edu/conselec/CE/kuhn/cdaudio2/95x7.htm
Various CD Standards:
http://www.ee.washington.edu/conselec/CE/kuhn/cdrom/95x8.htm
If these links aren't enough, some googling for the IEC 908 standard can give you more information that validates these links.
The second link covers how Error Correction works and how Error Concealment works. Error Correction != Error Concealment. They are two separate functions. From this page:
"CD players use parity and interleaving techniques to minimize the effects of an error on the disk. In theory, the combination of parity and interleaving in a CD player can detect and correct a burst error of up to 4000 bad bits -- or a physical defect 2.47 mm long. Interpolation can conceal errors up to 13,700 or physical defects up to 8.5 mm long."
Also:
"SO! The final frame (which started at 6*16*2 = 192 data bits) now contains:
1 sync word 24 bits
1 subcode signal 14 bits
6*2*2*14 data bits 336 bits
8*14 parity bits 112 bits
34*3 merge bits 102 bits
GRAND TOTAL 588 bits"
ECC is a function of hardware. This happens in real time, most at the controller level. In order to correct these errors, I would assume that ALL cd chipsets employ at least a small buffer to verify the parity information and then correct it. This would not be audible as it most likely happens as a function of the chipset before the data is output to the next stop in the signal chain.
If someone here has actually worked with CD Chipsets, it'd be great if they'd post something here. If someone here has information contrary to what I've posted, please feel free to correct me. Heck, if someone knows where I can read datasheets on CD Chipsets, please post a link! I did some cursory googles for it and came up empty (too much CD-RW stuff screwing up the results).
Hmm.. Looks like there were a number of replies while i was constructing my post.. Sorry if my information is redundant or points that I made have already been answered.
runebivrin said:
Hi Rune, Meitner states in the article referenced above on cryogenics that laser movements modulate the power supply, causing jitter. Extrapolating from what he said, any vibrations causing movement in the spinning disc theoretically could cause jitter.
I should add that Meitner's statements in the article regarding jitter are almost certainly based on actual measurements of jitter. I recall reading an interview of him where he discussed developing jitter measuring devices in-house to sensitively probe the many sources of jitter and track them to their source. FWIW, Meitner is probably among the best experts on jitter, and probably owns highest calibre jitter measurement devices.
Keep on Truckin! Thorsten.
This is like going back 20 years and hearing it all over again. As I remember, CD was "Perfect sound forever!". Dr. Doi's book 'Digital Audio Technology' 1983 should have set us straight. It is virtually all there! Why bother listening, or 'improving' perfect sound already?
Unfortunately, for me, it didn't sound 'perfect' far inferior to a record, analog tape, or even an FM tuner broadcast.
Then, among other things, we discovered 'jitter'. But, how to measure it, and why should such a small change in the time domain of micro, nano, or even picroseconds make any difference? Impossible? Well, I don't think so. In recent years, primarily do to the efforts of hi end designers like Meitner and Bob Stuart, we have seen CD reproduction 'improve' beyond 'Perfect sound forever'.
I guess that these improvements were there all along. 🙂
This is like going back 20 years and hearing it all over again. As I remember, CD was "Perfect sound forever!". Dr. Doi's book 'Digital Audio Technology' 1983 should have set us straight. It is virtually all there! Why bother listening, or 'improving' perfect sound already?
Unfortunately, for me, it didn't sound 'perfect' far inferior to a record, analog tape, or even an FM tuner broadcast.
Then, among other things, we discovered 'jitter'. But, how to measure it, and why should such a small change in the time domain of micro, nano, or even picroseconds make any difference? Impossible? Well, I don't think so. In recent years, primarily do to the efforts of hi end designers like Meitner and Bob Stuart, we have seen CD reproduction 'improve' beyond 'Perfect sound forever'.
I guess that these improvements were there all along. 🙂
Hi,
Basically, yes.
Now what is a FIFO? Ooopsie...That's a buffer.
Vibration of the spinning disc causes laser scatter which in turn leads to other problems.
Let's not forget this is optical media storage.
The data on the CD (after all it's just a file) includes already a substantial amount of extra information to be used by the error correction mechanism provided it can be read by the laser.
The laser uses the lands on the disc to position itself so if high vibration occurs the laser will see a scattered reflection and will have difficulty reading where it should position itself.
The pits contain the actual data.
To answer your question: it takes quite a large amount of vibration (does anyone know how much?) alone to make the entire ECC system trigger the mute transistors.
But it's not that simple: ideally a laser transport should operate as independently from the rest of the system as possible. In reality it hardly does so it causes interference (jitter) and all kinds of spuriae.
Same here but the guys in my backyard are working overtime to "Make Things Better"...
I wish they told us what those things are...maybe just their bankaccounts...Who knows?
Cheers,😉
Basically, yes.
Now what is a FIFO? Ooopsie...That's a buffer.
Vibration of the spinning disc causes laser scatter which in turn leads to other problems.
Let's not forget this is optical media storage.
The data on the CD (after all it's just a file) includes already a substantial amount of extra information to be used by the error correction mechanism provided it can be read by the laser.
The laser uses the lands on the disc to position itself so if high vibration occurs the laser will see a scattered reflection and will have difficulty reading where it should position itself.
The pits contain the actual data.
To answer your question: it takes quite a large amount of vibration (does anyone know how much?) alone to make the entire ECC system trigger the mute transistors.
But it's not that simple: ideally a laser transport should operate as independently from the rest of the system as possible. In reality it hardly does so it causes interference (jitter) and all kinds of spuriae.
Same here but the guys in my backyard are working overtime to "Make Things Better"...
I wish they told us what those things are...maybe just their bankaccounts...Who knows?
Cheers,😉
Konnichiwa,
As I have claimed. And the error correction covers up to samples (bits) out of a whole frame. Not much.
The Signal at the DAC passes through any number of digital stages. What is read of the CD by the laser is an analog pattern which is then passed through a trigger circuirt to derive "pit/land" or one/zero symbols. These are then held in a small FIFO buffer, usually no more than 64 Bytes, often less, 16 Bytes in the case of my example consumer CD Player Chipset.
BTW, I never denied that a buffer is present, I KNOW one is present. However it is a VERY small buffer and does not allow multiple attempts on reading the same data (which EAC and other Ripper Software CAN do).
They are high enough to shift from error correction to concealement readily on absolutely immaculate CD's, freshly pressed. Adding micro surface scratces and vibration to the mix would make a switch to concealement on a pretty regular basis more or less mandatory.
In other words, a microscratch that might not cause an error if the CD is not exposed to vibration might cause a read error if vibration is present. Equally, a pressing imperfection which might just pass on it's own may cause an unrecoverable sample f combined with microscratches and/or vibration.
I would be interrested knowing which transport RHoch employed in his test. Associates of mine found problems when comparing ripped and burned CD's to their originals played on consumer grade CD Players and originally argued extensively that the Copies made where not bitidentical to the original, when in reality the "originals" where actually compromised through the poor error correction of standard CD replay and ripped and copied very accuratly on PC Hardware.
But that as they say is another story.
IF the error correction delivers accurate data from the FIFO prior to the interpolator, the Jitter source is only the clock and the PSU line noise induced jitter (IIRC a single cmos gate with 50mV PSU noise translated to around 5ps additional jitter) from all the circuitry between FIFO and the DAC chip.
If the PSU feedthrough can be minimised and the clock lines are reclocked prior to the actual DAC the Jitter on the DAC should depend only on the Clock Jitter and any induced through the PSU in the reclocker and internally to the DAC.
These are many IFs.
My personal preference would be ripping the CD to a (PC) Hard-Drive and replay from there or asyncronous replay using multiple re-read attempts and the using a large buffer, that then combined with a system that provides a clean conversion master clock to the DAC (preferrably from a seperate clock generator also clocking the interface in the PC or the output of the Transport, basically studio style.
However we look at it however, using conventional CD Player systems suffer from many hidden problems that are usually not divulged to the great unwashed. These problems result at least potentially in audible differences where it would seem they should not exist. To pelicanise these problems will NOT DO and to attempt to argue them away if ignorant of facts causes egg all over faces.
Sayonara
runebivrin said:
As I have claimed. And the error correction covers up to samples (bits) out of a whole frame. Not much.
runebivrin said:
The Signal at the DAC passes through any number of digital stages. What is read of the CD by the laser is an analog pattern which is then passed through a trigger circuirt to derive "pit/land" or one/zero symbols. These are then held in a small FIFO buffer, usually no more than 64 Bytes, often less, 16 Bytes in the case of my example consumer CD Player Chipset.
BTW, I never denied that a buffer is present, I KNOW one is present. However it is a VERY small buffer and does not allow multiple attempts on reading the same data (which EAC and other Ripper Software CAN do).
runebivrin said:
They are high enough to shift from error correction to concealement readily on absolutely immaculate CD's, freshly pressed. Adding micro surface scratces and vibration to the mix would make a switch to concealement on a pretty regular basis more or less mandatory.
In other words, a microscratch that might not cause an error if the CD is not exposed to vibration might cause a read error if vibration is present. Equally, a pressing imperfection which might just pass on it's own may cause an unrecoverable sample f combined with microscratches and/or vibration.
I would be interrested knowing which transport RHoch employed in his test. Associates of mine found problems when comparing ripped and burned CD's to their originals played on consumer grade CD Players and originally argued extensively that the Copies made where not bitidentical to the original, when in reality the "originals" where actually compromised through the poor error correction of standard CD replay and ripped and copied very accuratly on PC Hardware.
But that as they say is another story.
runebivrin said:
IF the error correction delivers accurate data from the FIFO prior to the interpolator, the Jitter source is only the clock and the PSU line noise induced jitter (IIRC a single cmos gate with 50mV PSU noise translated to around 5ps additional jitter) from all the circuitry between FIFO and the DAC chip.
If the PSU feedthrough can be minimised and the clock lines are reclocked prior to the actual DAC the Jitter on the DAC should depend only on the Clock Jitter and any induced through the PSU in the reclocker and internally to the DAC.
These are many IFs.
My personal preference would be ripping the CD to a (PC) Hard-Drive and replay from there or asyncronous replay using multiple re-read attempts and the using a large buffer, that then combined with a system that provides a clean conversion master clock to the DAC (preferrably from a seperate clock generator also clocking the interface in the PC or the output of the Transport, basically studio style.
However we look at it however, using conventional CD Player systems suffer from many hidden problems that are usually not divulged to the great unwashed. These problems result at least potentially in audible differences where it would seem they should not exist. To pelicanise these problems will NOT DO and to attempt to argue them away if ignorant of facts causes egg all over faces.
Sayonara
"Yes, this "Buffer" stores each bit after read. Nevertheless the data extraction is still sequential and essentially syncronous."
I would argue that when we are talking ps jitter, the data can not be seen as even "essentially" syncronous. In this realm, its, at is best, pleisynchronous.
/
I would argue that when we are talking ps jitter, the data can not be seen as even "essentially" syncronous. In this realm, its, at is best, pleisynchronous.
/
john curl said:
John,
Indeed. Due to better filtering, higher interpolation and DAC bit rates, better understood and implemented analog post-processing and a few other bits and pieces. It has NOTHING to do with the discussion on error correction and the subtle effects that purpotedly are caused by the lack of pointy feet. It is bad enough that Thorsten tries to pull the discussion off-track to save whatever he can.
But can I ask you, in the interest of actually lesarning something, NOT to confuse us and try to introduce irrelevant issues thereby muddling the discussion? Thanks.
Jan Didden
Konnichiwa,
Granted. Where jitter is concerned the Buffer forms a barrier that simple Eyey Pattern Jitter cannot pass.
My point about syncronous, sequential extaraction related to the general operation, compared to using DAE modes or other such "read ahead & reread when error" strategies that some people seem to think are inherent to CD, when they are not.
At any extent, PSU current related crosscontamination is difficult to stop. Once you have managed to deal with that on both Ground and Supply lines you will have to wipe a lot of swaet of your brow in standard consumer CD players.
Sayonara
TNT said:
Granted. Where jitter is concerned the Buffer forms a barrier that simple Eyey Pattern Jitter cannot pass.
My point about syncronous, sequential extaraction related to the general operation, compared to using DAE modes or other such "read ahead & reread when error" strategies that some people seem to think are inherent to CD, when they are not.
At any extent, PSU current related crosscontamination is difficult to stop. Once you have managed to deal with that on both Ground and Supply lines you will have to wipe a lot of swaet of your brow in standard consumer CD players.
Sayonara
john curl said:
Don't drag out that old canard again. "Perfect sound forever" was nothing more than a marketing blurb coined my the marketing deptarment. Not by engineers and researchers.
No, "we" didn't discover "jitter." As is too often the case in high end audio, what "we" did was rediscover something which had been known of decades before Compact Disc ever hit the market.
*sigh*
se
Kuei Yang Wang said:
Clever, but you are not getting away with this. You clearly stated that subtle component changes like vibrations can lead to subtle audible changes because it would cause misread bits and influence the error correction. This is not true as noted by several. As long as error correction can cope, there are NO audible differences due to this because THERE ARE NO MISREAD BITS. Once error correction cannot cope anymore, error concealment tries its best to limit the damage, but it will generally be gross errors and be VERY audible.
Casually remarking that you lump the two together is really an insult to all of us who read your posts.
Jan Didden
janneman said:
Jan, if pointy feet reduce vibrations reaching the spinning cd, won't we also see reduced jitter?
Am I reading you right here?
Tom
Steve Eddy said:
Allow me to suggest John was saying we discovered that jitter creates audible effects. Sigh back.
john curl said:
Not exactly.
Back in the early '80 our research group bought several digital scope Tek and LeCroy. Their specs were 8bit - 100Ms/s (Megasample per second)
Their cost (each) was about 35,000,000 Italian lira (about 20,000 USD)
With the LeCroy we got a lot of literature and there was an amazingly well written white paper about time base jitter and its relation with vertical resolution.
If the CD players were built by LeCroy nobody had to worry for the jitter since the '80 🙂
john curl said:
Yes John, it DOES completely turn me off. It is a deliberate attempt to introduce issues that have nothing to do with an intellectual discussion. I am afraid that I am not accustomed to forming my opinion on hearsay from someone who was on the phone with someone who knows somebody he had dinner with. Sorry.
As for the SE bit, I am not sure how to react. Is this another attempt to discredit one or the other or both? Fot the life of me, I cannot detect anything constructive in that last part. But maybe I just missed it?
Jan Didden
What experiments have you done and/or what sources back up this claim? In my personal experience, new CD's of reasonable quality typically produce bit-identical streams on multiple reads and through multiple transports. Vibration does not change that behavior until a substantial level is reached, after which large errors (muting and long sample hold times) are observed, not single bit interpolation errors.Kuei Yang Wang said:
Do your experiments show otherwise? Are there any easily pointed to sources that confirm your assertion that good quality new CD's routinely produce read errors beyond the ability of the CIRC error correction scheme to rectify? My observations are consistent with the theoretical scratch limits on correction and concealment. I'd be interested to read the data showing that redbook CD is fundamentally incapable of producing bit-perfect playback, even with absolutely immaculate freshly pressed CD's, as you state.
Over the past few years I have performed versions of this experiement using a first gen Denon DVD player, a low-fi consumer Sony DVD player, the Marantz SACD player, a Marantz 300-disc DVD changer, and a ten year old 5-disc CD carousel player (either Yahama or Panasonic, I don't recall now). If you really must know I'll provide model numbers. I once even performed such an experiment using a 1984 Technics CD player, and was surprised at how often successive plays of a track were bit-identical using quality discs.
- Status
- Not open for further replies.