CD Block Errors, Servos, and the Audible Threshold

[fas42]

Tim, I appreciate your impatience with those you feel go overboard with aspects of trying to improve the sound they get. Of course a lot of it is mumbo jumbo from the point of view of why they believe what they are doing is effective. That doesn't change the fact that something they've done has altered the SQ, and so it is then up to them to decide whether it was a beneficial change, and then whether good value for money. It makes a hell of a lot of sense for everyone to try and understand why, when they try something, that it makes a difference, and the big problem is that we are all still on a major learning curve understanding it all.

Even the equipment you're enjoying now has been tweaked in various ways by the manufacturer, some of it consciously and some just because they got into the habit of doing things a certain way, because "it sounded better". It wouldn't sound as good as it does if as least some level of fine tuning and fiddling with its makeup had not been applied ...

In the end, it's all just a journey -- enjoy the ride ...

Frank

 

[JackD201]

How does one tell a virgin turtle dove from one that's played the field?

 

[DonH50]

You have to look up its... well... Nevermind!

"Chatter" is what I meant by the more rapid vibrations the goop might dampen. I find it very hard to believe there is any impact on the power supply (or anything else -- we are talking about a few errors in a digital bit stream, and really nothing else changes whether those errors are present or not). Off the cuff, the only way I see all this causing a problem is if the errors cause more clock jitter, or are an indicator of higher clock jitter/discontinuities ("glitches"), which could certainly be a possibility.

 

[fas42]

"Chatter" is what I meant by the more rapid vibrations the goop might dampen. I find it very hard to believe there is any impact on the power supply (or anything else -- we are talking about a few errors in a digital bit stream, and really nothing else changes whether those errors are present or not). Off the cuff, the only way I see all this causing a problem is if the errors cause more clock jitter, or are an indicator of higher clock jitter/discontinuities ("glitches"), which could certainly be a possibility.

Again, CD player with physical bits vibrating: the servo mechanisms keeping the laser head on track have to work harder, drawing short, sharp spikes of current from the power supply on an ongoing basis. A higher error count is a sign that the CD reading mechanism is stressed doing its job, meaning that the power supplies are injecting high frequency ripple of voltage and current in various parts of the player. If the player was built like a battleship of course, as some are, this is much less of a problem, that's why people pay the money for them.

The rippling can easily become interference on the digital data stream, meaning jitter and the rest of it. Also, if the mains power supply is marginal then the interference from this could even be passed back down the cord and up to the DAC's power supply. There are a number of ways for these electrical disturbances to impact the rest of the system ...

Frank

 

[DonH50]

Frank, I hesitate to disagree because I've already got one member very PO'd at me, but would like to just to point out a few things in parallel with your assumptions:

1. The servo has to keep the laser tracking regardless, and the motor has to keep the CD spinning, but the servo is not moving at anything like the bit rate. It is a mechanical thing and is just tracking the tracks, a process much slower than the bit stream. AFAIK the servo does not backtrack or move around on a per-bit (or even per-word) basis to correct an error; the error correction data is part of the encoding process, embedded in the bit stream, and handled by the decoder before the DAC. Unless servos have gotten waaay faster since the last time I piddled with one (anything's possible).

2. A properly designed power supply with adequate decoupling and well-controlled lead/trace dressage should not be injecting noise elsewhere via mechanical or electrical coupling. Yes, i agree that few power supplies are likely to fall into the catergory... And yes, whatever the reason, if power supply noise gets into the bit stream it could modulate the clock and cause jitter. It could also get coupled into the analog output stage and wreak havoc. Actually, clock coupling into the analog output is one of the more difficult things to control.

3. Many things can cause high error rate, including a dirty,warped, or just poorly-manufactured disc. Yes, vibration plays its part, but if the transport was ideal there could still be numerous errors on the disc. Not downplaying transport noise, just noting there are other causes, too.

Please, please don't take my comments as an attack or comment on your (or anyone's) intelligence! Think of them as supplements, random thoughts whilst reading the posts... - Don

 

[JackD201]

Hi Don,

I was thinking more of the laser's tracking errors. The laser transport runs on a rail and the laser tracks at a specific angle. If a disc chatters it is conceivable (to me anyway) that the laser might not just make errors in reading before and after a data point but inadvertently read points beside it. Sheesh, how do I put this? Assuming for a second that a player reads a CD in a linear fashion like an LP. In the case of a warp where there is no stylus but a beam of light not physically in contact with the groove, that beam would vary either closer to the outer edge of the disc and toward the spindle randomly. That leaves lots of work for error correction won't it? Could these be the cause of some of the dropped bits?

 

[fas42]

Hi Don,

I was thinking more of the laser's tracking errors. The laser transport runs on a rail and the laser tracks at a specific angle. If a disc chatters it is conceivable (to me anyway) that the laser might not just make errors in reading before and after a data point but inadvertently read points beside it. Sheesh, how do I put this? Assuming for a second that a player reads a CD in a linear fashion like an LP. In the case of a warp where there is no stylus but a beam of light not physically in contact with the groove, that beam would vary either closer to the outer edge of the disc and toward the spindle randomly. That leaves lots of work for error correction won't it? Could these be the cause of some of the dropped bits?

Sorry to jump in, Jack, but to save Don the work of answering I 'll just throw in a bit ...

The reading of a CD track is a very sophisticated process: the data is not stored linearly like on LP but spaced around the track in a complex fashion to make error correction as robust as possible. An enormous number of data points, by our standards, could be mucked up or missing and the music would still be read off correctly or with barely a twitch in the sound. The chance of reading off an adjacent track and thinking that was relevant data is so damn close to zero I wouldn't think about it ...

To Don: you're one of the nicest guys around here, I have no complaints! As regards the servo, I know it is not going with the bit rate but it is trying to keep the laser positioned in a precise position with respect the disc track as it moves around: it is almost comparable to the tonearm on a TT. To follow through on that analogy, if the grooves on an LP were badly rippled sideways or vertically the tonearm bearings would be working feverishly keeping the cartridge where it was supposed to be. The servo on the CD player has to do this using pulses of current driving the coil or whatever if vibrations are effectively making the CD tracks appear rippled to the laser head.

Frank

 

[JackD201]

I know it isn't linear my friend, every bit has 4 of the same elsewhere in the disc IIRC so if there is dirt or a scratch it can be recovered. That's why I said "Let's assume for a minute". "Recovered" being the operative word here. A CDP will have a limited number of passes and if it can't recover the missing piece it moves on, again IIRC. Anything interfering with that read process logically would cause the drops mentioned by Gary. those are tiny pits and very short lands. If you think of it in their actual scale the chatter just might be significant.

 

[FrantzM]

LP and CD are very different processes .. Comparisons although attractive leads to gross misunderstanding ... The application of the AVM on CDs is too me one of these tweaks that only work when one sees or knows it has been applied... IOW Knowledge seems to be required to hear those so-called difference.. Removed it and ... Remember the "pen", "rings" and other contraption we did see when CD really sounded bad .... Call me utterly skeptical.

 

[JackD201]

But Frantz, there are measured correlations.

 

[DonH50]

Thanks guys. I have decided I am over my head on this one so will try to bow out gracefully...

I am not sure exactly how the laser servo tracks (knew once upon a time, trying to decide whether to blame age, poor memory, senility, or just the passage of time!) I think it is following an average signal level and error handling is actually all digital with minimal impact upon the transport unless there are gross errors. But, any of us who have ever heard a CD "skip" can attest that such errors do indeed happen! So in practice I am not sure how much the transport moves beyond its normal trajectory in the presence of errors, or even if it changes at all (except when the whole system loses its mind and has to start again). I suspect a lot of errors fly by without the transport really doing anything, as there may be media errors rather than tracking errors, if that makes sense. Ditto the motor, which I believe in a CD varies speed (?) I do not remember off-hand now (too early this morning) what is constant, but I think it's the bit rate, and if so the motor speed must vary to keep the data stream constant as it moves from center outwards on the disc. Is it impacted if there are more errors? One more thing to worry about...

The error recovery and correction (ERC) bits are built into the data word. I do not recall how many bit errors can be corrected, but I think it's more than one. Nor do I recall if Gary said the measured reduction in bit error rate was audible. Certainly if the laser jumps a track or loses a bunch of bits Bad Things can happen. One other thing I do not know (see why I am getting out of this?) is what interpolation may be done in the event of a series of bit errors. A memory buffer would provide the time to process the data and interpolate between the last good sample and the next, but of course that could corrupt the signal. I suppose a big enough buffer would allow time for the servo to go back and try again, but of course that is pointless if the data is simply not (readable) on the disc.

This has me showing my ignorance and wishing I had more time to read up on the fine details of all of this. I have reviewed the actual data format now and then, but rarely looked into the actual transport (i.e. the laser and disc electromechanical systems).

Interesting stuff! - Don

 

[garylkoh]

What I've been talking about are C1 errors (the Block Error Rate or BLER). They are single bit errors that are easily recovered. All CDs have C1 errors, and in the Redbook standard, 220 BLER per second averaged over a 10-second block is regarded as a "quality recording". There are other errors - C2 is a frame error, and this is where a CD player can interpolate and try to correct for the lost data. CU errors are uncorrectable, and may result in an unplayable disc. Some CD players will just stop, others will glitch.

I don't think that I can hear the difference between 5 BLER and 6 BLER. But there is a big difference between 30 BLER and 10 BLER (big in the relative sense). At the scale of bits on a CD, what the laser reads is on a probabilistic density function. Vibrations, wobbles, lack of track concentricity, etc. are all accounted for and engineered for. None of this can explain why AVM would make a CD sound better.

As Tim said, it might be me expecting to hear the difference and being able to pick out the one treated disc out of five inside a 5-disc carousel.

 

[DonH50]

Have someone else load the carousel?

I was hoping for C2 errors at least, as C1's are (should be) corrected before any processing happens and thus should not in any way change the sound. Makes me wonder about how the ERC engine is working with the rest of the DSP... I am still speculating that the extra time to correct an error is somehow "gitching" the data stream and thus injecting clock jitter. I wonder if the errors have some pattern to them? That could raise their audibility.

 

[fas42]

None of this can explain why AVM would make a CD sound better.

But would you accept that a player mechanism where there is less vibration of any form, of the components of the reading mechanism and CD could deliver better sound?

Another aspect of the scenario is that if there is a very high error rate the digital circuitry is working much "harder", doing the calculations to correct or interpolate. If you accept that high speed digital circuitry can create interference in analogue circuits and power supplies then that is another source of electrical muddying of the waters ...

Frank

 

[fas42]

Ditto the motor, which I believe in a CD varies speed (?) I do not remember off-hand now (too early this morning) what is constant, but I think it's the bit rate, and if so the motor speed must vary to keep the data stream constant as it moves from center outwards on the disc. Is it impacted if there are more errors? One more thing to worry about...

Take a rest, Don, you're right on the money ! Yes, the bit rate must be constant, how could it be other because the data must always be fed to the DAC at a specific rate. The CD dimples or pits are roughly the same distance apart from each other around the track whether it's at the innermost or outermost track, so the laser has a decent chance of always picking them up. This is of course different from the LP where the "data" is squashed together on the inner tracks, meaning the sound is never quite as good there ...

So the CD motor changes speed in a rough and ready way to make sure the buffer holding the latest bit of music data is in good shape at all times. TT fans would freak out knowing how sloppy the CD motor can get away with being!

One thing I would quibble with, though. As far as I'm aware, on normal CD players the laser head only gets one chance to read the data, if it glitches then it's up to the digital circuitry to try and compensate as best it can. Yes, I know that units using DVD drives spin fast enough to give the mechanism a number of goes at it, but my understanding is that conventional CD players won't do this ...

Frank