What's the deal with CD players used as transports?

[WS65711]

----------------

On 9/27/2004 6:54:04 PM D-MAN wrote:

----------------

On 9/27/2004 6:42:44 PM nicholtl wrote:

D-man, you must have extremely, inhumanly acute ears. Maybe a dog's ears! It took me 2 generations of CD burns before I can notice a loss of dynamics.

----------------

I don't think so, but I can tell. The nuance defining the soundstage goes first (the small transients). It's subtle, but a degration for sure.

By-the-way, I have a Meridian cd player - Ironic!

DM

----------------

If one was to rip a wavefile from a cut on the original CD, and a wavefile of the same cut on a copy CD several generations down, and then do a "file compare" of the two wavefiles on the computer, wouldn't that prove (or disprove) this theory about the CDR copy not being an "exact" copy?

Dave

[D-MAN]

Burning a cd is the culprit, dude, not the type of source signal. The light-sensitive properties (and short-comings) of the recording medium will degrade the source signal regardless of the source signal - UNLESS the signal is already simple enough to fit entirely into the digital threshold (i.e., the elements are completely recordable and are not analog - no ADC required - another point of loss).

One thing to consider is that in order to fix the original analog signal into digital, alot of valid information is "dropped", that is 16, 18, 20 bits are defined as "steps" and the algorythyms "fill in" the missing analog information as needed. Computer data source information do not need to fill in these steps as nothing is missing - it ain't analog (or continuous).

The more complex the source "signal" (audio is more complex than computer data as defined above) the more loss per burned copy (include the losses of the ADC as well as the laser and media, etc). Even most error correcting algorythyms cannot fill it in (why would there be ANY error-correction needed if you get a perfect copy anyway?)...

Eventually, you may not get a further reduction, as the loss has already taken place (i.e., there is no more information available below the recordable threshold). So theoretically, you could reproduce THAT generation ad infinitem, but the loss to the original source has already occured any number of generations prior to that one.

The question is, how much degradation is acceptable?

DM

[WS65711]

I'm not really sure if you were answering my question or not. My question was in regard to these statements:

------------------

1) take a familiar commercial audio cd and burn a copy of it anyway you can.

2) a/b it on your high-end audio system - maybe a slight difference, but pretty darn close...

3) burn a copy of the copy.

4) a/b that copy (gen 2) against the original commercial one- more difference?

Repeat until your convinced...it takes me only one generation. The first thing to go are the micro-transients.

------------------

Maybe (probably) I'm wrong, but it would seem to me that ripping and comparing a wavefile from each of

the CD's in step one and step four should prove or disprove any differences. As far as I know (admittedly

not very far), there is no D/A conversion involved in simply copying a CD to CDR.

Dave

[formica]

----------------

On 9/28/2004 6:54:48 PM WS65711 wrote:

Maybe (probably) I'm wrong, but it would seem to me that ripping and comparing a wavefile from each of the CD's in step one and step four should prove or disprove any differences.

----------------

Yes, that you can perform a check-sum on an original and a copy... which will prove or disprove any data differences. If you search it on the web, you'll see that it has been done... and the results I recall from several years ago showed that a few errors do slip by. Note that reading the same disk, twice in a row, also had some errors.

One should also consider the different reflectivity of the CD and CDR media... where some players have a difficult time with certain recordable media. Pressed disks have more defined pits than CDR and will be easier to read by sensitive (inferior, but independent of price) transports. Ironically gold disks are also less reflective than regular aluminum.

----------------

As far as I know (admittedly not very far), there is no D/A conversion involved in simply copying a CD to CDR.

----------------

Yes, you are correct no D/A conversion in burning a CDR from a CD.

Question is, how much of all this is audible in a blind ABX test, and which part is psychoacoustic?

Rob

[D-MAN]

My contention is that it is VERY audible. But test it for yourselves.

When blind A/B testing, take the FIRST impression only. Then take a break, set up another blind A/B and make only one selection. Repeat until you have a list of results that statistically will prove out the question.

Otherwise, the brain kicks in and will confuse your selection which then becomes selections PLURAL, which is why there is so much confusion about A/B testing! The biggest mistake that people make is that after deciding (or having no decision) they say ok, switch it back. They are now relying on "memory" of sound and each switch will reduce discernability of differences further. Avoid the "let me taste THAT one again" approach.

One listen to each version and one choice. Then rest your ears and brain. Blind switch. One listen per version - one choice. Repeat as needed, and WRITE IT DOWN.

Then do it over again the same number of times and check the results. Statistically, the more matches, the more there is validity in the choice in question, whether psychologically induced or not.

DM

[arthurs]

This is like any other discussion/argument on any audio board. There are outboard DAC's that can definitely make a difference in sonics on digital, and the transport you use can also affect it. Whether or not you personally find value or sound improvement is driven, IMO, by the quality of the DAC and the transport as well as the rest of your system. The Benchmark DAC1 and the Ack!DAC are 2 in the reasonable price ranges that I can hear audible improvements from, but hey, that's just me, your mileage and opinion may vary....

[meuge]

I don't think you're thinking of this in the right way.

Jitter is caused by the fact that information written on CDs is not organized in sectors, and thus is only clocked to 1/75 second intervals from silence. Thus, if a drive stops reading, and restarts again, it will start only within 1/75 seconds from where it left off and will not be any more accurate, thus causing jitter.

Both hard drives and memory on a computer are organized into clusters, which are far more accurate than the time signature of a CD (4kb clusters are default for windows and correspond to 1/350th second of a redbook CD playback) and have discrete start/end points which are absent on a CD (it just counts in intervals of 1/75th of a second from silence, while a computer accurately maps precise locations on the media). Furthermore, both hard drives and memory utilize very advanced buffering and error checking technologies which guarantee that the data remains bit-for-bit intact WITHIN the clusters, as long as the CPU and memory and hard drive are operating properly. Therefore, jitter is only a function of the player itself. Thus, even though it MAY be possible to get data degradation by successively copying a CD using inferior media, it's only a possibility if the data is not checked after writing. If you check your copies against a WAV you're using to write (option is available in Nero Burning Rom, as it probably is in all software) you can detect errors even in readable CD. So the simple way to avoid jitter, is exactly what I suggested before -> avoid successive reads of the same track by buffering the track to memory while using error-correction.

So, if you rip your CDs to a computer as WAVs, and use the SPDIF-out, you'll AT WORST reduce jitter 5-fold (if you think of a hard drive as a CD-like media, which it isnt'), but realistically reduce it orders of magnitude more than that. And if you're worried about clocking errors of a chip, remember that a computer CPU is a far more precise instrument than an embedded chip (such as in most CD players) and has multiply redundant error-correction and caching instructions. If it didn't it would never be able to run at Gigahertz speeds. If it made errors so large that they'd be audible, all the scientific calculations made by computers would be absolutely useless.

I am prepared to put my money where my mouth is. I can make successive CD copies, and then label them with nothing more than ID numbers (which will be related to the generation of the CD, the information which will not be provided to the tester) and send them to anyone who thinks they can hear the difference after reading this post. When I receive the results of this blind experiment I'll gladly post them here. Just to make sure you don't think I'm cheating, I'll send a file with the table of IDs vs. generation of CD in an encrypted form to the tester, and provide the key after I receive the results. Hell, I'll pay for the postage too.

P.S. On the other hand, I certainly agree that the quality of a DAC matters, especially its power, as it does with anything where either the input or the output is analog.

[D-MAN]

the MEDIUM itself is the problem!

You CANNOT BURN A REPEATEDLY ACCURATE COPY. PERIOD.

Usable, but NEVER - repeat - NEVER a BIT-FOR-BIT COPY of the original. It is not physically possible, ESPECIALLY for the LESS THAN $99 cd burner in your computer!

I have a rather expensive Sony cd burner (audio) that also does a less-than-perfect job. It is in the nature of the medium itself.

For further proof, why on earth would commercial cd/dvd companies PRESS their versions from glass masters, instead of BURNING them, which would be cheaper?!

Even better, why do CDRW's crap out after a CERTAIN number of writes? answer: the error-correction FAILS; even more: why are there any errors at all?

P.S. I can't believe that any other "Audiophiles" here have not backed me up on this...

DM

[meuge]

The reason there ARE errors on ANY media is because the size of the pits/magnetic bits/electronic gates is so small now, that simple brownian motion of molecules in surrounding material, or the media itself, can alter the bit content. However, if you research the way CDs are written you'll realize that much of the actual space on a CD is dedicated to error-correction parity data, and even more so on DVD. It is not NECESSARY to WRITE a bit-for-bit perfect copy, which is as you pointed out, virtually impossible. However, it is very possible to READ a bit-perfect copy from an imperfect CD due to the parity data and error correction. Think about this - there are only 8-bits to a byte. If I write a full CD of text, and nothing else, and there are bit errors, when I read the CD some of the text characters will be changed. That is simply NOT SO.

The reason companies PRESS the CDs and DVDs IS because of accuracy, but more importantly because the medium is different. In cheap CD-Rs, the media chemically reacts with the plastic, corrupting the data over time. In more expensive CD-Rs, this process is slower. On the other hand, in commercially pressed CDs it is far far less likely to occur. Degradation of information on media is nothing new, but that wasn't what I was talking about. The way to burn perfect copies repeatedly is by storing software copies of data, as opposed to copying them back and forth. In essence, the process of writing a CD is a way of digital-analog conversion, since you're converting digital information into a laser-heated impression on analog media. What I am saying is that if instead of burning copies directly, you first copy the information to the hard drive EVERY TIME, and then burn the CDs, degradation of data will not be an issue, as you will actually be utilizing the error-correction portions of the track, as opposed to trying to copy them.

Secondly, another reason why companies press their optical media, is because they want to ensure consistency of reading. Data problems are often the result of problems with the reader... and again, these are issues which can easily be avoided by simply caching the contents of a track... and then re-reading it several times to make sure the information is correct. I would venture a total guess, that this is how some of the higher-end transports work... and that's why they may just be CD-ROMs in disguise. They're probably an embedded computer, running an OS off a ROM chip, and have let's say 128MB of RAM for caching audio tracks (should be enough). The OS is probably running proofreading algorythms and error correction, which is how it attains much better accuracy than players that do direct conversion. They may have some EMI shielding on their output stages (most likely just CAT5 cable, which does a marvelous job of common-mode reduction), but I think that's where the differences end.

P.S. The reason why CDRWs fail is related to the physics and chemistry of the media -> after a certain amount of writes, the media has been chemically altered to such an extent that it creates more errors than the error correction can handle.

P.P.S. The way audio information is written currently, is far from perfect. The parity portion of the track is clearly inferior to the amount of parity data written on data tracks. Format a CDRW and you'll see how much space you'll lose from those 700mb. Most of this space is dedicated to 2 things -> allocation cues and parity data - the two things that are necessary for read error reduction.

[D-MAN]

meuge, EXACTLY!

I'm tryin to tell these guys that a burned audio cd copy is not the same as the original glass-pressed version.

I know that there's folk here whose systems are better than mine, and my hearing isn't any better than anyone elses...

DM

[meuge]

Well, you're right - nothing is a perfect copy in a quantum sense. The thing is that a burned copy can be read in such as way as to REPRODUCE the original perfectly.

[D-MAN]

Oh, man, not you too!

I can hear the difference, and so can my friends.

I use copies for the car, mostly.

DM

[meuge]

Hehe - let's try it. I use very crappy media myself, but I never store data in one place or in one format. If you agree that slowly ripping a CD does not change the contents, then I can burn you let's say the 1st generation CD from a WAV and then use successive generations down to let's say 10. I have never tried this myself, and I'm pretty open-minded, so I'd like to find out the results. Send me an email with your street address and I'll send you the CDs. I think 30 trials should be enough to perform a good statistical evaluation of the results. If you want, you can also send me a CD of your choice and I can rip that, and go from there.

P.S. I guess I just like experiments, regardless of the subject. It's the PhD in MD-PhD* that's speaking : )

*No, I didn't get them yet : )

[LarryC]

Gosh, meuge, are you a prodigy or a candidate for an MD and PhD at your tender stated age? What's the avatar and where does your username come from? Apologies if you've mentioned this before.

Larry

[meuge]

I wish I already had the degrees Larry. I am in the program, but have quite a bit to go (the average is about 8 years (after college), and I've started my 2nd). I didn't mean to sow confusion or pretend to have them for that matter.

My avatar is something that was a personal gift from someone : ) So it's a custom design.

My username comes from some verbal manipulations of my name and has a very long history.

Eugene

[LarryC]

I am in the program, but have quite a bit to go (the average is about 8 years (after college), and I've started my 2nd).

I figured you were in a combined degree program. May I suggest you take a swipe at public and/or private policy training and opportunities as well as medical scientific? Good luck!

Larry

[WMcD]

Advice 1. Listen to Larry. He is an individual of extra ordinary intellect and talent.

Going back to part of the thread which I hijacked a little bit. Is there a utility which will compare date on a CD-ROM versus a first, second, third, etc. generation copy. Then we'd know for sure.

I just can't buy the theory that "microdynamics" suffer. Or at least suffer alone.

"Microdynamics" implies that the least significant bits of data are the ones prone to errors in copying in the digital domain.

But it seems to me that all bits on the disk have the same chance of being mistranslated or read wrong, or being subjected to interpolation; however it works.

Gil

[Audio Flynn]

In a transport you want to get the following items in the best performance level.

1. Power supply isolation and filtering

2. EMI sheilding of DACs (video DACs in DVD players are the worst)

3. Vibraton control

If these 3 are not controlled jitter increases.

The medium is flawed like mass market CD playing products are poorly designed.

+++++++

Dampen the chassis (constrained layer damping in the inside of the lid) and get some "feet"

Ferrite cores on the incoming AC in addition to your AC management device from Panamax or Monster or whoever

Lead tape on DACs (carefully applied so you do not short out something) and transport structure

Dampen circuit boards

Add o Rings to capacitors

[mike stehr]

Encase everything in lead.

[nicholtl]

Anyone ever seen those funny little vibration dampening balls? They actually look more like pyramids. I guess to me those things are like ghosts - I won't believe it until I see/try it.