The CD transport sub-system includes the servo motor, the laser
pickup mechanism, the decoder, and the error detection and recovery
circuit. The resulting data from the CD transport sub-system is
the "corrected" or "recovered" digital data ready to be converted
back to analogue domain by a digital-to-analogue converter (DAC).
The compact discs have a high natural error rate even when their
surface has not been damaged. This is mainly due to fluctuations in
the rotational velocities as being controlled by the servo motor,
and vibration of the disc itself. In order for CD to work at all,
data reconstruction and error recovery is a vital part of the
transport sub-system. The implementation of CIRC (Cross Interleave
Reed-Solomon Codes) are spec'ed to fully recover a dropout of a
maximum of 2.47mm. Working backwards from the signal integrity
standpoint, we can see that deviations from the original bitstream
(as "printed" on the compact disc itself) will result in "loss"
of vital information which may play an important part in providing
ambience, imaging, etc... How can the resulting bitstream be
different than the original one if error correction is a vital
part of the system, you ask? Well, the short answer is that error
correction is computation intensive and takes time. In order to
keep real-time data flowing through the system, the error-correction
circuit is allowed a time budget to operate. As the error correction
approach its time limit (due to excessive errors detected in the
bitstream), it will try to play "catch-up" by kicking in its
interpolation logic or at worst, its concealment logic in order to
speed things along. It is at these data samples resulting from
the interpolation or concealment circuit that deviations occur
in the bitstream. Given a perfectly clean disc, the source for
these "excessive" errors may result from disc vibration, the error
tolerance for "constant" velocity as maintained by the servo
motor, and the quality of the auto-focus mechanism in the laser
pick-up itself. Given that the name of the game is to provide the
exact copy of the original bitstream as written on the compact
disc, everything in the transport contributes to the integrity of
the resulting digital signal coming out of the transport into the
DAC. Oh, and we have not even consider "jitter" yet but that's
another topic for another day.

_Spike

 Sorry, but this is all wrong - Werner 23:24:32 3/10/99 (3)
 Audio Comic Hi-Fi-Choice has your Orca - Edp 15:28:12 3/11/99 (1)
 Moi? - Werner 23:26:54 3/11/99 (0)

...

Show full thread outline!

 Sorry, but this is all wrong - Werner 23:24:32 3/10/99 (3)
 Audio Comic Hi-Fi-Choice has your Orca - Edp 15:28:12 3/11/99 (1)
 Moi? - Werner 23:26:54 3/11/99 (0)
 Checked out your site. It's well done and very interesting. - Audiophilander 09:10:22 3/11/99 (0)
 to JJ - Rodney Gold 20:59:11 3/10/99 (3)
 Re: to JJ - jj 09:00:38 3/11/99 (0)
 Re: Transports - Lukasz F 23:04:51 3/10/99 (0)
 Just wondering when you're gonna show up... - Spike 21:06:40 3/10/99 (0)
 Sorry. No cigar! Bad Information, too! - jj 17:40:09 3/10/99 (8)
 I stand corrected... - Spike 18:06:34 3/10/99 (7)
 You do raise an interesting point - jj 18:41:06 3/10/99 (6)
 I can't believe it's...jitter. - Spike 18:58:02 3/10/99 (5)
 Re: I can't believe it's...jitter. - Werner 23:30:40 3/10/99 (0)
 Its not placebo - Rodney Gold 21:02:21 3/10/99 (0)
 It's not a simple problem - jj 19:11:27 3/10/99 (2)
 ABing transports - Rodney Gold 21:08:13 3/10/99 (1)
 Delays - jj 09:02:48 3/11/99 (0)
 Another perspective ... - nataraj 15:25:53 3/10/99 (0)
 that's why my LP's sound so much better! well written! (nt) - dy/dx 13:21:52 3/10/99 (1)
 touché .... LOL (nt) - Stephen 13:24:10 3/10/99 (0)
 Interesting. Thanks! - Bill M 13:07:24 3/10/99 (0)

In short, the reason is that the formats are different. Data cd has a lot more space for error correction codes (a cd stores only 640mb of data, but almost 100mb more of audio ). Also it is not 'real-time' so, if an error occurs cd-rom can re-read. If you put a scatchy cd-rom you can see it really struggling to read and finally reads the files (or gives an error) after many tries.

Hopefully, mlp will correct this problem, since the format assures perfect data transmission. The new IEEE standard ('fire wire') would help too.

But, as you say, jitter is another matter. But, re-clocking in dacs should help. Lets leave out jitter at 'a to d' during recording/mastering stage.

The "error correction" circuit runs all the time, has basically constant delay, is a fixed process, and either works or it doesn't. This sort of error correction also functions in your modem, your cell phone, and many other places, and is no mystery. It does not use "extra information", rather it uses deliberately redundant information put there for the sole purpose of allowing for error-correction!

That is the real point. The "extra" bits on a CD, vs. the original 16 bits * 2 channels /per sample, is not information, it is redundant, it is derived totally mathematically from the original data, and supplies no information whatsoever, OTHER than to allow error correction.

When error correction fails you do NOT loose '1 bit' you get an entire block of entirely corrupt data, btw, and that's pretty catastrophic, even when the "mitigation" circuitry is used.

As far as jitter, if you get any substantial jitter on the OUTPUT stream from the transport, then you have a transport that isn't to spec. Sorry, but them's the facts.

Now, if a CD does have to interpolate, yes, that's audible, but in most discs, that's never. There are players in which you can measure the number of errored blocks (interpolated, whatever, anything that means that the data from a given block was lost). Most discs never trigger this flag, meaning simply that under most normal circumstances, NOTHING is lost.

Basically, 100% of the information is present up to the time that the disc can't be decoded, at which time interpolation happens, and the errors do matter.

Now, some transports may not implement the full error-detection strategy, but frankly, most do, it's a simple commodity IC at this point, you'd have to go out of your way to UNDO the full process, and of those that flag when they have failed to provide the entire data fully and without any error whatsoever (and determining that is possible to an incredibly accurate determination), very few show as much as 1 flag per disc.

Until that flag starts to be activated, there is no information lost. There is no impairment of the data. There is no change in the ambience or anything else.

Spike is entirely incorrect in ?his? claims here, in fact, especially in the representation of what information is stored on the disc. ONLY the original PCM information is stored, the rest is added, deliberate redundancy that allows for the error correction. Until the error correction FAILS, which it rarely does, the data does not change.

What bothers me about this article is the statement----
Working backwards from the signal integrity
standpoint, we can see that deviations from the original bitstream
(as "printed" on the compact disc itself) will result in "loss"
of vital information which may play an important part in providing
ambience, imaging, etc...
---

This statement is entirely incorrect, very misleading at best, and likely to launch another "audio myth" at worst.

Sorry, but them's the facts. The data that is "printed" on the CD includes the FEC (that's Forward Error Correction) information that is quite conciously and directly redundant, and loss of that redundant information causes NO loss of the decoding of the original data, in fact, that's why the redundant information is added during the process of writing the CD.

Spike's description also left out a very important step, the "EFM" or eight to fourteen modulation that is used to add more redundancy and to also aid the tracking system and modem in recovering the bits to be used for the Reed Solomon code. (The "cross interleve" is simply a data rearrangement so that bad spots are spread out so that they make the total error rate below the rate that is EXACTLY correctable by the Reed-Solomon code up to some length of drop-out.)

It is a complete and abject fallacy that information is lost before the error correction fails. The information "added" in the process of writing the CD s not actually information in the real sense of the word, it is conciously redundant information that is added, deliberately, in order to allow for error correction. The "extra information" is mathematically derived by a fixed formula from the original data, and serves no purpose other than allowing the recovery of the original data.

There is no effect on ANYTHING in a proper transport until the error correction fails, i.e. fails to return the original bits. This is rare occurance on a CD OR a CDRom. Yes, there is a SMALL amount (not a large amount, please, get your numbers straight) of extra redundancy on the CDRom, but in fact it's a small percentage of the Reed-Solomon coding and the EFM coding. The overall redundancy is not increased substantially on the CDrom. Most of the overhead on the CDrom is there to allow for accurate addressing of each block of data.

Thanks again for the info.
_Spike

Any sonic difference is due to one or the other. Clocking and data recovery are OLD arts, speaking in modern terms.

_Spike

I will say that I've personally thrown DAC's we've bought to test right out the door for having really bad dejittering performance.

Yeah, you'd think that would be fixed, wouldn't you? I would. (grumble)
We've only had that kind of processing in telco since the 1950's.

I have a truly high resolution digital system , I use Meridian digital speakers where each driver has its own dac (with dual pll "de-jittering"), amps etc . Prior to that I use a meridian 518 mastering processor (also "de-jitters"), and prior to that a z-sys digital parametric. I also have a "de-jittering" genesis digital lens.

For some time I have been testing transports extensively , from the megabuck and overbuilt ML31.5 , to the real cheap Marantz type cdp's with digital out.

Now for example , I have blind , double blind and sightedly tested a Mission Cyrus Discmaster (using a CDM9-por laser mechanism , supposedly one of the best) vs lets say , a theta data universal ( a Philips laserdisc player with a theta digital output board and upraded power supply)
The Cyrus has some serious features to read the CD properly etc ,

Using either AES/EBU , or SPDIF output from it , to be blunt , the Cyrus sounds like crap , as does an Arcam 250 , as does the Marantz type stuff. This is so regardless of what device I use to "remove" jitter after the transport , and regardless of what cable I use to link em.

This inst a subltle thing , it is noticeable immediately on listening , there is something "missing" from the music , and this was diametrically opposite to my expectation of a Drive like the cyrus , as it uses one of the best mechanisms ever produced.

The funny thing is: if you add a proprietary regulated power supply to the Cyrus discmaster it sounds a whole lot better , in fact , all the machines that sound good seem to have concentrated on power supplys.

What makes these transports sound different? , the general trend is this :
Cheap (and this seems to correlate to worse) transports sound fast and impactful , yet lack deep bass , bass pitch definition , have a narrow and closed in sound with very little note decay and "wetness"

The better transports have a more "relaxed" presentation (I think this is cos the notes seem to die off better) there is a whole lot more air and spaciousness to the sound , the bass is deeper and better defined and low level detail is markedly improved.

As an aside , the Arcam has a seperate link to its own dac for the clock signal , but it still sounds bad , if you use a better transport to the partnering Arcam DAC , it sounds good once again , leading me to discount the fact that jitter is the major cause of the "poor" sound.

This transport thing is not just confined to my system , I have tested transports in other systems and found the same thing.

_Spike

Lukasz F.

How do I know?

Firstly via a brain-boiling course at the university,
which did not concern CD, but the pure maths behind
error-correcting codes. No Mickey Mouse affair this!
Indeed this was about the toughest stuff I've ever encountered,
and I did go through an awful lot of mathematics back
then!

And secondly via a simple test, done oh so often in
the past: take a CD-player and connect an error-logging
system to its decoder. PLay a full CD. A visible scratched
sample then typically yields about 700 fully
corrected errors, and 0 up to a few (2, 3) instances
of interpolation. These are transient effects which
can not and will not harm overall sound quality.
Take a mint sample of a CD, and there are no interpolatiopns
whatsoever.

- Visions in Audio http://www.geocities.com/Hollywood/Hills/4133
- TNT Audio Magazine http://www.tnt-audio.com/int.html

• Visions in Audio

But even then bit errors would only couse transients,
and won't harm overall sound quality as such.

Jitter: as the ONLY point where jitter matters is the
actual conversion time at the DAC chips, DAC systems
could be made that are entirely insensitive to the
type of transport used.

But the industry does not do this.

One could ask: why?

$$$?

Pick the DAC/lens combo that has the most difference. Are you near anyone who has a spectrum analyzer, or a Tektronix AA501, or something of that sort (or both would be best).

Without seeing/hearing/measuring htem, it's hard to guess what's going on, although your description sounds almost like it'snot getting the full word resolution, which would be really REALLY bizzarely odd.

Then again, the power-supply thing suggests either jitter or some interaction with the AES/EBU tranciever and noise...

I've had this experience testing players. It's hard to synch closely.

Then again, it COULD be that, mind you, it's just something to try to check on very carefully.

BTW, I noticed that you're also a fan of SF/Fantasy novels. Although it's a bit off-topic (e-mail if you'ld like), who are your favorite authors? [Note: my wife and I know many of the leading SF authors (we work in the closely related fields, she as a mystery/horror author and I as an astronomical artist)]

Sorry about the digression. I just wanted to let you know that your site is excellent.

AuPh

So since this is mainstream stuff now, when are you going to sell it?

Anyway, Michell always has been mainstream. Well,
in the sense Kubrick made mainstream films, of course.

(Michell and Kubrick were neighbours in the sixties)

Maybe there's some confusion: I own an
older Michell Argo preamp, which is to house
my own preamp design. Once. If I ever find the time.

As for now: I have to finish my phonostage first!