JIMV
JIMV's picture
Offline
Last seen: 6 years 1 month ago
Joined: Jan 31 2008 - 1:46pm
A fundemental question
dcstep
dcstep's picture
Offline
Last seen: Never ago
Joined: Sep 16 2007 - 4:59pm

Optical transports having reading errors that your Benchmark can't correct. A better quality transport will have lower reading errors. The DAC's error correction can fix some of this. Ideally a disc is read to a buffer and errors are corrected by re-reading prior to playback or patched with an alogarhythm.

Flash drives are closer to ideal because of the lack of moving parts. Wire's don't matter near so much as errorless reading of the data, but improvements can be made by improving analog I/O devices and the quality of the DAC and the accuracy of the clocking.

You're rolling around in the low end of the possibilities and still readily hearing differences. As you scale up in quality you'll be able to hear more differences, but it will not necessarily correlate to cost. There are better DACs than the Benchmark, but it's very good for the money. You'll likely have to spend way more money to improve on it.

Dave

JIMV
JIMV's picture
Offline
Last seen: 6 years 1 month ago
Joined: Jan 31 2008 - 1:46pm

Thanks...that makes sense. Another question, is there inherently a difference between a DVD playing CD and a CD player when using the digital out? Ie; for music, should I invest in a CD player or DVD player (at the under $300 range of the spectrum)?

Elk
Elk's picture
Offline
Last seen: 3 years 6 months ago
Joined: Dec 26 2006 - 6:32am

There is a difference only in that a DVD player that also plays CDs as a bunch of extra components so that it can play both. CDs and DVDs are different technically and thus involve different wavelength lasers, etc.

Everything being equal, an identically priced CD player and DVD player should favor the CD player - for CD playback.

(A DVD laser has a wavelength of 640 nanometers; a
CD laser, 780 nanometers. The smaller DVD laser wavelength enables it to focus on the smaller pits on a DVD, partially explaining a DVD's much greater disk capacity.)

JIMV
JIMV's picture
Offline
Last seen: 6 years 1 month ago
Joined: Jan 31 2008 - 1:46pm

Thanks

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

Hey Jim,

Interesting test and not the outcome I would have expected. Dave is mostly correct although the error correction and concealment are implemented in the transport (often in the CD 'module' itself) and not in the DAC, so if the transport can't read or reconstruct the right bits there won't be anything the DAC (any DAC) can do about it.

However data error bursts usually manifest themselves as much more noticeable artefacts like clicks and pops, it is generally timing errors that produce the more subtle sonic degradations that result in subjectively poorer sound quality. The Benchmark claims very high levels of jitter rejection which should in theory mask some of the differences between transports, although obviously not in this case.

I have heard really profound differences when comparing a really expensive transport with one twice the price, both feeding a top-of-the-line DAC. My expectation bias was that at this exalted level there probably wouldn't be any significant difference, if indeed there was any audible difference. How wrong I was! Don't ask me to explain it, all I can say is that the datastream only has two axes: amplitude and timing. If there is a difference it must be in one or other or both.

That said, I am still surprised that Philips DVD player/Benchmark combo gave such poor results. You've got me scratching my head here. Very interesting.

Jan Vigne
Jan Vigne's picture
Offline
Last seen: Never ago
Joined: Mar 18 2006 - 12:57pm

In the $300 range there's not much to listen to. Try the new NAD players.


Quote:
If a junky computer produces an acceptable data stream, why would not a junky CD or DVD player acting as a transport do the exact same thing and have the same result when played into a good DAC?

I bought a used Benchmark DAC1 and hooked it up with a very inexpensive Phillips DVD player as a transport. I used a good Transparent digital cable, insured I had clean contacts everywhere and played the new hookup into my existng amp. The result...my 8 year old Adcom 750 CD player sounds far better than this new setup. I changed the Philips player out and used a 10 year old Optimus portable CD player with a digital out and that sounded as good as my old Adcom. If bits is bits, why do changes in units acting as transports change the sound quality?

I can't say that what you heard should have been completely unexpected. A junky player is a junky player. For one thing, the Adcom has no extra cable between the transport and the DAC. If you begin with a junky quality signal, a cable will only make matters worse. If the signal is corrupted at the disc reading stage, then no amount of error correction or buffering can compensate for those errors and not be somewhat audible. In my experience, the longer the signal is held in the error correction circuitry - a 60 second correction vs. a 10 second correction, the worse the sound. What's in a junky DVD player? Junky parts. I would expect the Adcom to have a better transport than the junky DVD player and therefore also likely to have a better error correction system. I think you are confusing the issue of bit retrieval with the issue of junky bit retrieval.

At the time one of the rationalizations for the RS's exceptional sound was the transport. If you look at the three balls that clamp the CD in place, it is very similar to the data retrieval transports and very unlike the flimsy mechanism found in most players and certainly in a low end DVD player sold for features rather than sound quality. Look at the current range of Rega players, all good value items and starting at the $1k price range, and you'll see the same sort of three ball arrangement.

The Optimus was a hot set up in its day used with an outboard DAC. My player stopped working years ago. I still have the PS audio DAC however - somewhere.

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

Hey JV,

Couple of points:


Quote:
If the signal is corrupted at the disc reading stage, then no amount of error correction or buffering can compensate for those errors and not be somewhat audible.


You don't distinguish between error correction and error concealment. Many common error types can indeed be corrected completely inaudibly because the correct data can be computed by the error correction circuitry (using redundant data on the disc) with complete accuracy. Error concealment (using, typically, repetition or interpolation) is an approximation and might therefore be audible, at least in theory.


Quote:
In my experience, the longer the signal is held in the error correction circuitry - a 60 second correction vs. a 10 second correction, the worse the sound.


Given CD's frame rate of 7350 frames per second at 1X a 10-second error would cover 73,500 frames and a 60-second error would cover 441,000 frames. A disc with a BLER that high would be completely unplayable. Remember a BURST error is defined as just 7 or more contiguous C1 frames containing an error and a CD conforming to the Red Book standard should contain no BURST errors.

Also, the error correction circuitry doesn't 'hold' the audio, it processes the errors much faster than the stately 1X replay speed. In fact it works pretty much the same way whatever speed the disc is read at. If you have a CD ROM drive capable of reading at 48X the error correction/concealment is being performed 48 times faster than real time. You'd need bat ears to hear it in operation.

JIMV
JIMV's picture
Offline
Last seen: 6 years 1 month ago
Joined: Jan 31 2008 - 1:46pm


Quote:
That said, I am still surprised that Philips DVD player/Benchmark combo gave such poor results. You've got me scratching my head here. Very interesting.

I was not expecting it either as the cheap Phillips DVD player was only 2 years old and the the portable at least a decade. The explanations to my question made sense to me. I have since had the DAC factory refurbed/updated and have sold the Phillips transport so I cannot repeat the initial tests. I can say the Optimus/DAC1 combo sounds better than my Adcom GD750 player.

JIMV
JIMV's picture
Offline
Last seen: 6 years 1 month ago
Joined: Jan 31 2008 - 1:46pm

Thanks...again good information.

Jan Vigne
Jan Vigne's picture
Offline
Last seen: Never ago
Joined: Mar 18 2006 - 12:57pm

I will say I am not a digital specialist. It doesn't interest me. I read about the process and it sounds far too much like putting the toothpaste back in the tube.

From my somewhat uneducated viewpoint, there is little audible difference between error "correction" and error "concealment". I understand that technically there is a vast difference. But in the end poor error correction results in what I would consider concealment of the musical signal. Not technically correct but that's what I hear.

Regarding the 60 sec. "hold" on error correction, I was using an extreme example which I have seen on several portable CD players. Lately I've seen what would appear to be enomrous amounts of time for error corection to take place on portables. I actually don't know what the average or the limit would be on error correction in a home audio player from today's market. I do know that I heard fewer good sounding players as the average amount of time the signal was held in a buffer increased. To say the poor sound was always the result of increased amounts of errror correction would not be completely logical since other changes in the players were also taking place at the same time. However, increased amounts of error correction were being touted as the benefit of many players and the resulting sound became less and less agreeable. It would also seem unreasonable for me to assume that increasing the amount of error correction did not have some effect on the player's sound quality.

With the oversall intent of the mass market players being one of lowest possible cost, increased error correction seems to be a side "benefit" of such design. The average Joe doesn't want to take home his new $39 player only to find it skips whenever he tries to play that disc he forcefully retrieved from his child/dog's mouth.

With that in mind, the results reported here are not surprising to me. I could certainly be wrong about this. Like I said, digital doesn't hold my attention.

Just as an after thought, the three ball arrangement found on the old RS player and the new Rega's is also found on the original Sony PlayStation that has received mixed reviews but that AD found to be superior to several high priced units. It does appear getting the information off the disc in an organized manner is of the utmost importance while relying on error correction to do the work the transport should have accomplished is a fool's game.

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

JV,

No worries, I thought it was a superb piece of 'policy-based evidence making'.

I'm not trying to ding you, I know you contribute a wealth of great experience on these boards, but I think you have a few things mixed up when it comes to digital audio.

I am certainly no expert on portable players but I do know that some of them offer 'electronic shock protection' where a certain number of audio samples (maybe a couple of seconds' worth) are buffered so that if the laser mistracks momentarily due to a physical impact there is no dropout in the music. This really has nothing whatever to do with error correction or concealment. And it may or may not be audible, and if so it is equally debatable whether it would manifest itself as a SQ degradation or improvement.

Misunderstand me right, I am not dismissing what you have heard out-of-hand here. I know it is generally unwise to dismiss empirical observations just because they don't fit with accepted theoretical wisdom. However it is ironic that some modern players are incorporating RAM buffers specifically to improve SQ. I would offer the Boulder 1021 as a case-in-point.

My own view is that sonic differences between CD players are far more likely to have to do with other factors such as traditional audio engineering preoccupations (e.g. power, grounding, shielding etc. - all of which give rise to different jitter spectra in digital circuits) as well as the quality of the analog circuitry, than error correction or concealment per se.

Cause and effect are notoriously difficult to link in digital audio circuits, not least because so many digital distortions manifest themselves with the same sonic signature, non-correlated noise. I have no doubt that many folks can hear the deficiencies of digital audio, however I believe there are no more than a few hundred in the world who can with any degree of precision tie what they are hearing back to specific aspects of circuit design or implementation. For most of the rest of us there are just way too many variables we can't eliminate. That's where the temptation to 'imagineer' causes to explain effects (and vice versa), as well as draw spurious linkages comes in.

My own opinion fwiw is that most error correction and even concealment is likely inaudible by most people in the case of most music, not least because the real-time duration of most errors is only a few thousandths of a second and the audio frequencies most susceptible to distortion are the ones to which the ear is least sensitive. However I am not one of the above 'few hundred'. That assertion is based on my understanding of digital audio theory and not empirical experiment, so I have absolutely no way of proving it one way or the other.

Jan Vigne
Jan Vigne's picture
Offline
Last seen: Never ago
Joined: Mar 18 2006 - 12:57pm

I agree with that. There are numerous places where a digital signal can be degraded beginning back at the recording studio should the engineer consider digital a flawless or at least a faultless system. Maybe I didn't make my point clear when I said junky bit retrieval is to be expected from a junky player. Along with junky bit retrieval I expect junky error correction and many other flaws in the system. The idea that a junky player of any vintage didn't retrieve the information properly isn't much of a surprise to me. A DAC can only deal with the data it is provided.

I wouldn't use portables as my ideal for sound quality, the old RS unit being the exception. The 60 second buffer was just a number, not a true comparison point. I've heard many players with unknown-to-me error correction buffers that I didn't like and that I did like. When I purchased the Rega Apollo I didn't care or ask what the error correction figures were.

The implication was that as "junky-ness" increased more BandAids such as larger amounts of error correction were used in an attempt to simply make the player operate at an acceptable level to the buyer who relied on their $39 purchase. It's not that error correction is the only fault in such a design but that error correction is used by the manufacturer to overcome other problems upstream.

However, thanks for the clarification.

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

Don't wish to appear argumentative JV (honest!), but I think your thesis is fundamentally flawed. The error correction/concealment curcuitry is designed principally to deal with faults in discs, not laser sled assemblies. For most players, timing errors are the dominant form of digital-related distortion and error correction/concealment, occurring as they do in the digital domain, have no effect whatsoever on timing.

As I proposed earlier, I suspect most of the deficiencies you hear can likely trace their roots to traditional analog circuit issues (poor circuit design, poor component choice, power supply instability/noise, noisy grounds, RF interference etc.) that you are undoubtedly familiar with, just manifesting themselves in unfamiliar ways.

Jan Vigne
Jan Vigne's picture
Offline
Last seen: Never ago
Joined: Mar 18 2006 - 12:57pm


Quote:
The error correction/concealment curcuitry is designed principally to deal with faults in discs, not laser sled assemblies.

Then correct me if I'm still wrong. You refer to error correction as principally used for faults in the disc. By this I assume you are referring primarily to physical imperfections such as scratches, smudges, physically unreadable locations due to damage, etc. If that's the case I also assume those errors would be something rather easily avoided and an assiduous audiophile could for the most part prevent error correction from entering the picture in even the lowest priced players. But my clean discs still sound bad in a cheap player despite the amount of disc hygiene I apply to the physical surface of the disc.

My understanding of digital playback would suggest disc hygiene and handling are not involved in the day to day sound quality of most players. That's not to say we can use a CD as a coaster but that there are more important matters at play here. Is it not the case that error correction is always taking place to some extent no matter the obvious condition of the disc? Similarly to LP's I understand there are no perfectly centered CD's/DVD's. And somewhat unlike LP's but still having a common theme most digital transports present the reading mechanism with an "unflat" or constantly varying horizontal surface to work with. This is the result of the discs spinning at as much as 500RPM and the distance from the laser head to the actual surface of the disc constantly varying up and down causing the laser to constantly correct itself and resulting in lost data or misread information. As such the servo mechanism of the laser transport is constantly making corrections for mistracking of the disc.

One of the finer points of a "better" player is then a better servo mechanism with less ambiguity in its "search and correct/search and correct" nature. This keeps the laser on track more of the time than a cheaper system where the servo - acting as servos must - only responds after the read error has occurred and then often overcorrects to the point of creating further and possibly greater errors in the reading of the disc which is still spinning off center and wobbling up and down. The compensation network for such sloppy transports (the BandAid) is higher levels of error correction/concealment with the previously read data, the "signal", sitting in the buffer for longer and longer periods of time while more incorrect data is fed to the system and interpolation of the good data is filtered through the bad data that results from sloppy construction. This keeps the error corection operating on a near constant basis with the already retrieved data being stored in a buffer awaiting correction or concealment and mixed with incorrect data. The problem is compounded by the minimal number of chances a digital system has at retrieving high frequency information. Higher rates of error in the laser/transport assembly mean higher rates of servo induced errors which then place incorrect data into the correction circuit. Since digital playback is based on algorithms with a finite number of bits of information placed in the system and error correction interpolation based on what has been read before, the higher the error rate at any one point in the disc, say, from an off center disc being misread, creates a situation where the player never truly catches up to itself until the 0's and 1's have long since passed through to the anaolg sections.

This becomes a never ending source of misread information/misinterpolated error correction that only contributes to the error correction circuit working more (most) of the time rather than just when a scratch needs to be filled in. In other words, one of the benefits of a better player would be I am hearing the error correction circuit in operation less of the time and that therefore produces far fewer timing errors as a result.

You stated earlier the error correction circuit operates primarily at frequencies that most of us can't truly hear. That's all well and good but I would still think missing data is missing data and the effects of error correction filter downward into the audible spectrum just as rolling off an analog signal at 150kHz will affect the overall preception of realism. Is that not true?

And I do agree the analog section of any player is still vitally important to the overall sound quality. But correct data retrieval from the source disc is where it all starts and a junky player is still a junky player only with larger BandAids applied to the wounds it inflicts upon the signal. This has always been my understanding of how to consider the functional quality of a transport when paired with an outboard DAC. Junk is still junk.

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

Wow JV, that's a lot of questions for someone who confesses that digital audio theory doesn't interest him. I'll try to address your points as best I can but if you are interested I would really recommend reading up on this rather than just speculating it together. You are obviously experienced in the audio field so I am sure you have all the prerequisite knowledge. There is plenty of information out there on the internet but if you prefer good old-fashioned books, two good places to start are Bryan Brewer's The Compact Disc Book: A Complete Guide to the Digital Sound of the Future and Ken Pohlmann's The Compact Disk Handbook.


Quote:
Then correct me if I'm still wrong.

I'm afraid your assumptions are wrong and consequently all the conclusions you base on them.


Quote:
You refer to error correction as principally used for faults in the disc. By this I assume you are referring primarily to physical imperfections such as scratches, smudges, physically unreadable locations due to damage, etc. If that's the case I also assume those errors would be something rather easily avoided and an assiduous audiophile could for the most part prevent error correction from entering the picture in even the lowest priced players. But my clean discs still sound bad in a cheap player despite the amount of disc hygiene I apply to the physical surface of the disc.


By faults I am referring to faults manufactured into the disc. There are literally hundreds of critical parameters that define a Compact Disc according to the Red Book Standard. These range from simple physical characteristics of the disc itself like inner (hole) and outer diameters, weight, thickness, eccentricity, flatness etc. to optical characteristics of the polycarbonate such as reflectivity and birefringence to the physcial characteristics of the substrate such as the correct location of lead in and lead out areas, track pitch and, perhaps mostly important of all, pit geometry. In order for a disc to play, the reflected laser beam must generate track-following and focus servo signals so that the player can follow the track and stay focused. Generation of the proper servo signals is completely dependent on the size and shape of the pits on the disc. There are millions and millions of pits on the disc and a certain proportion of them are misshaped, some 'fatally' so. None of the above factors are "easily avoided by the assiduous audiophile" for obvious reasons.

Another fallacy is that damage to the 'playing' side of the disc is most detrimental to playability/sound quality. The 'playing side' is completely out-of-focus to the laser and therefore many small defects there are transparent to it. Only a (relatively) serious defect here will cause an issue. However any damage to the substrate itself is likely to be catastrophic; it is the label side of the disc you need to worry about!


Quote:
My understanding of digital playback would suggest disc hygiene and handling are not involved in the day to day sound quality of most players. That's not to say we can use a CD as a coaster but that there are more important matters at play here. Is it not the case that error correction is always taking place to some extent no matter the obvious condition of the disc? Similarly to LP's I understand there are no perfectly centered CD's/DVD's. And somewhat unlike LP's but still having a common theme most digital transports present the reading mechanism with an "unflat" or constantly varying horizontal surface to work with. This is the result of the discs spinning at as much as 500RPM and the distance from the laser head to the actual surface of the disc constantly varying up and down causing the laser to constantly correct itself and resulting in lost data or misread information. As such the servo mechanism of the laser transport is constantly making corrections for mistracking of the disc.

This is mostly quite true. Laser assemblies are designed to cope with exactly these problems and as you point out the cheaper ones probably do a worse job than the more expensive/better ones. However the point I think you are missing here is the scale of problems that can and can't be recovered. A serious error (one affecting more than a handful of adjacent frames) will either cause an audible pop, or, if serious enough, will cause the disc to stop playing. It will not just cause a disc to sound subjectively poorer. However a poorly designed output stage in the CD player certainly will.


Quote:
One of the finer points of a "better" player is then a better servo mechanism with less ambiguity in its "search and correct/search and correct" nature. This keeps the laser on track more of the time than a cheaper system where the servo - acting as servos must - only responds after the read error has occurred and then often overcorrects to the point of creating further and possibly greater errors in the reading of the disc which is still spinning off center and wobbling up and down. The compensation network for such sloppy transports (the BandAid) is higher levels of error correction/concealment with the previously read data, the "signal", sitting in the buffer for longer and longer periods of time while more incorrect data is fed to the system and interpolation of the good data is filtered through the bad data that results from sloppy construction. This keeps the error corection operating on a near constant basis with the already retrieved data being stored in a buffer awaiting correction or concealment and mixed with incorrect data. The problem is compounded by the minimal number of chances a digital system has at retrieving high frequency information. Higher rates of error in the laser/transport assembly mean higher rates of servo induced errors which then place incorrect data into the correction circuit. Since digital playback is based on algorithms with a finite number of bits of information placed in the system and error correction interpolation based on what has been read before, the higher the error rate at any one point in the disc, say, from an off center disc being misread, creates a situation where the player never truly catches up to itself until the 0's and 1's have long since passed through to the anaolg sections.

Again, error correction and concealment are not there to deal with errors on the scale you describe. Yes, error correction may be active almost constantly. This is not a problem, it is 'invisibly mending' the data and the following circuitry and your ears are none the wiser. Error concealment can mask bigger errors in a way that I would assert is largely inaudible, but we're still talking about numbers of samples 2, 3 or 4 orders of magnitude below what you are implying. These circuits process data so fast that they are still idle most of the time, compared to their processing speeds the real world looks like a movie being played back at a rate of a-frame-a-day.


Quote:
This becomes a never ending source of misread information/misinterpolated error correction that only contributes to the error correction circuit working more (most) of the time rather than just when a scratch needs to be filled in. In other words, one of the benefits of a better player would be I am hearing the error correction circuit in operation less of the time and that therefore produces far fewer timing errors as a result.

I'm afraid your last assertion is another fundamental misconception. The only points in the digital audio chain where timing is relevant are at the points of ADC and DAC. Error correction and concealment work entirely within the digital domain and have, to a first approximation, no effect on jitter. Note this statement is true in theory and probably largely true in practice, although of course one can never rule out microscopic interactions through power supply rails etc.


Quote:
You stated earlier the error correction circuit operates primarily at frequencies that most of us can't truly hear. That's all well and good but I would still think missing data is missing data and the effects of error correction filter downward into the audible spectrum just as rolling off an analog signal at 150kHz will affect the overall preception of realism. Is that not true?

At risk of repeating myself error correction is completely inaudible. It restores the sample to its original value and nobody is any the wiser. Error concealment deals, per above with small errors (samples representing thousandths or hundredths of a second of audio). In most cases an interpolated value is likely to be very close to the original. However this breaks down for higher and higher frequencies. An audio signal with a frequency of 22.05 kHz will indeed have different values in all adjacent samples, so interpolation would be completely unable to reconstruct this signal. However how good is your hearing at 22.05 kHz? The error concealment circuitry knows nothing about the frequency of the audio signals encoded in consecutive samples (you only find that out after DAC) so it interpolates missing values unaware of whether the music at that point contained high- or low-frequency content. It is working across the entire audio frequency band, but the magnitude of the distortion it is introducing rises with the frequency of the encoded signal.


Quote:
And I do agree the analog section of any player is still vitally important to the overall sound quality. But correct data retrieval from the source disc is where it all starts and a junky player is still a junky player only with larger BandAids applied to the wounds it inflicts upon the signal. This has always been my understanding of how to consider the functional quality of a transport when paired with an outboard DAC.

I think I agree with most of this with the exception of your casting error correction and concealment as "Band Aids" for a poor circuit design or implementation. It is not the digital equivalent of 'rolling off the top' to mask a shrill treble, hopefully the above rationale makes it clear why.


Quote:
Junk is still junk.

Amen!

Jan Vigne
Jan Vigne's picture
Offline
Last seen: Never ago
Joined: Mar 18 2006 - 12:57pm

As a salesperson during CD's rollout I read the Pohlmann book when it came out and additionally Nakamichi sent out a very informative primer on digital playback when they introduced their first player in the mid/late 1980's. By that time there had already been quite a bit of discussion about the virtues and failings of CD, outboard DAC's were just around the corner of time, and the Nak book did a fairly decent job of addressing both areas. Surprisingly only a small portion of the Nakamichi book length technical essay on digital's claimed superiority was concerned directly with the performance of a Nakamichi player. (Possibly this was in part due to the fact Nakamichi was also rolling out their Dragon "self centering" turntable at about the same time. No need to kill the goose before you harvest the eggs.) Both of these books were something I read twenty years or more ago. Today there are numerous specialists and sub-specialists in digital that far exceed the knowledge base required to be well versed in turntables or amplifiers on a sales floor. I have basic concepts in my head though they don't always connect with each other after two decades. I try to read as much as possible when a review of a digital product is in front of me but sometimes it just doesn't interest me enough to make a concerted effort to integrate what lurks in the back of my memory with what is now being presented. At a backyard gathering I'm not going to sit down and strike up an in depth conversation about digital is what I'm trying to say.

Like you I'm not trying to be argumentative but what you've laid out appears to me to still operate mostly in the theory of digital and not necessarily in the reality of digital - especially digital playback at the sub-$500 level. Introducing issues such as power supply drain caused by shared circuits is a matter that would appear to dominate lower priced players. This is a shared issue with both the digital and the analog circuits is it not? One affecting the other?

If what you say is how the real world operates, what then is the role of the transport in sound quality? Should a junky transport sound as good as any other?

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm


Quote:
Like you I'm not trying to be argumentative but what you've laid out appears to me to still operate mostly in the theory of digital and not necessarily in the reality of digital - especially digital playback at the sub-$500 level.

Haha, fair point! I don't have any practical experience of digital audio circuits, so in that sense I am a pure theoretician. However I do have a lot of practical experience of digital electronics in general, I designed and built a math coprocessor for an early home computer for instance, as well as practical experience of analog audio, designing and building amplifiers as a hobby. What my experience tells me is that practice is generally close to theory in the 'purely' digital domain, most of the problems come when you try to transition back to analog. We spent a little bit of time designing and implementing (non-audio) real-time circuits back at university and it was enough to teach me a healthy respect of the difficulty of preserving clock integrity.

So I would propose (again, just my opinion) that digital theory and digital practice are pretty close when you are talking about the 'digital' (i.e. amplitude) domain ('the bits'), and that largely speaking that holds true at any pricepoint. This is why printing a word file from a cheap computer and an expensive one will generally give identical results! When you introduce the time domain (i.e. at the point you transition back to analog) that all changes. Bigtime. That is where I think you will find that the majority of compromises in cheaper equipment manifest themselves and where the majority of the sonic degradation occurs.


Quote:
Introducing issues such as power supply drain caused by shared circuits is a matter that would appear to dominate lower priced players. This is a shared issue with both the digital and the analog circuits is it not? One affecting the other?

Yes it is, but I wasn't talking about 'drain' per se, I was talking about the possibility that the error correction circuitry could pass noise to the DAC by modulating the power line thereby causing jitter with a spectrum correlated to the incidence of errors. However you're right, it is conceptually similar.


Quote:
If what you say is how the real world operates, what then is the role of the transport in sound quality? Should a junky transport sound as good as any other?

Great question. As I related above I have heard the (significant) difference between a very expensive transport (an mbl 1521, $10,950) and one more than twice as expensive (an mbl 1621, $24,000) so empirically I know that the transport can make a difference. So how do I reconcile that with my theoretical understanding?

Well, given everything I know I would have to say that I am skeptical that all of the design and implementation differences between the two contribute equally to the resulting difference in quality. Indeed it may be that the $24,000 transport is grossly over-engineered for show and to justify its extremely high price tag. I strongly suspect that there are one-or-two key differences that really make the difference. That is why there are still screaming bargains out there in digital audio and why, for instance, the best $2-3k DACs can sound pretty much as good as the best $10k+ DACs. Doesn't quite work that way in analog.

It may very well be that the error concealment circuitry has less to do in the more expensive transport, however I am convinced that the lion's share of the sonic difference lies in its ability to generate a low-jitter digital signal. Fwiw I also fundamentally believe that this is easier to do starting from solid state RAM memory than from a rapidly spinning polycarbonate disc, hence there is no CD transport chez Struts. I rip all my music to hard disk and play it back using a sound card (headphone system) or a music streamer (big rig). At the end of the day it is not a patch on the sound of my tt, but I haven't given up yet!

Jan Vigne
Jan Vigne's picture
Offline
Last seen: Never ago
Joined: Mar 18 2006 - 12:57pm

Interesting post, thanks. I'm of the opinion the theory of digital and the operation of digital are far closer throughout the price ranges than the theory of analog and the operation of analog can be at widely disparate price ranges. So, if that's a good thing, digital wins on that account. Yippee! A $39 CD player does sound "better" than a $39 turntable.

Still I have to think there is some deviation from ideal theory when a manufacturer can produce a digital disc player that can sell at retail for $39.

I believe your example of computer drives isn't totally accurate since the majority of the difference in price is not in the hard drive's build quality - as far as I can discern - but in all the peripheral gadgets that go with a higher priced computer. To my knowledege, the hard disc is still very much the same between most consumer CPU's. I don't know if the trend is stll prevalent with DIY'ers and tweakers but using an outboard data transport as a CD player was quite popular among that set not that many years ago.

From what I understand the early Sony PlayStation transports were taken from a Sony computer drive and this in large part accounted for the sound quality many people find to be quite musical. Years ago when outboard DAC's were beginning to sell, the superiority of a better transport could easily be demonstrated IMO. One of the best transports we had in the shop at the time was an ES Sony that was quite solid. It gave the impression you could pick up the 30+ lb. player by the open disc drawer and swing it around the room without damaging the player. I know the "feel" of the disc drawer has nothing to do with the transport used inside the player but this Sony was exceptional in its overall build quality.

This is where I started to get my impressions of how a quality transport affects the performance of the digital playback. We also had an Onkyo Integra that was built along the same lines as the Sony ES and it also gave similar results when used as a transport. Then came the Radio Shack portable with its unique (for the time) computer sourced transport and my empirical experience began to say "transports matter".

Unfortuantely, everything matters. Jitter certainly matters. Analog circuits matter. IMO error correction matters. And isolating a digital source has proven to matter even when digital theory suggests it should not to the degree it apparently does.

I've done a few of my own recordings where the information gets stored on a SD card with no moving parts. The transfer to disc seems to subtly loose something - even though digital theory tells me it should not. There are, in my mind, too many locations where such loss could occur so I don't have a clue why the transferred signals are slightly dissimilar to the original but they are when I listen close enough. Possibly it is my imagination but doing away with the mechanics of the transport would seem to be the best approach to digital reproduction. I just don't believe I'm ready for a music server.

haroon
haroon's picture
Offline
Last seen: 3 months 1 day ago
Joined: Sep 5 2005 - 12:01am

I am sorry if someone has already pointed out the issue.

A dedicated CD player has a higher chance of being better transport for CD playback than a DVD player because of clock accuracy difference. A CD player just has to generate one word clock of 44.1 kHz from its master clock while DVD player has to generate two word clocks of 44.1 kHz and 48 kHz by using a PLL and thus a compromise is made. I believe this might be the reason that your older portable CD player sounded better than the newer DVD player. However, this will not hold true if a DVD player has two dedicated crystals for each master clock to generate word clocks of 44.1 kHz and 48 kHz instead of relying on PLL.

For more insight on clocks, check this thread

http://forums.slimdevices.com/showthread.php?t=54162&highlight=word+clock

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

JV,

I suspect you have a point, in fact I wouldn't be surprised if the truth lurks somewhere in between the positions we've each set out.

One of the problems is that very few people know how different CD modules compare on error rates and error correction/concealment or are in a position to easily test for themselves. The folks at AccurateRip for instance have an extensive database of ripping data from computer-attached CD-ROM drives, but presumably not from the CD-DA/CD-ROM modules typically used in 'consumer electronics' CD players.

As I said, my personal preference is to try to avoid the whole issue by ripping all my CDs to hard disk using dbpoweramp "secure mode" (which rereads each track up to six times if it doesn't get exactly the same result as in AccurateRip) and then send it to the DAC using a streamer with a nice big FIFO buffer. However as I also said, I am a computer geek and I realize that this approach is not nearly mature enough for the mass market.

I agree with what I think you are saying insofar as I don't think the current state of our knowledge of what is going on here is nearly complete. We can laugh about it over a beer in future years if someone discovers that bits have some hitherto undiscovered property other than their ordinal value, such as 'spin' or 'charge' or some quantum-level mischief, which neatly explains why exactly the same bits can produce different sounds in different circumstances.

I don't know if I think it is likely to happen, however I wouldn't rule it out completely!

andy_c
andy_c's picture
Offline
Last seen: 5 years 4 weeks ago
Joined: Dec 25 2007 - 12:48pm

There are also issues with S/PDIF outputs of cheap devices related to resampling. A while back, I was reading in the Roku forum about the evolution of different versions of their SoundBridge product (a lot like the Squeezebox). In order to reduce cost, later versions of the product resampled everything to 48 kHz for output to S/PDIF. I'm pretty sure this happens with some cheap DVD players as well. Some of the resampling algorithms (and their implementations in silicon) can be pretty bad too.

struts
struts's picture
Offline
Last seen: 2 years 9 months ago
Joined: Feb 1 2007 - 12:02pm

I hear you Andy, but I think that is an understatement. There are issues with S/PDIF period. Mixing clock with data is IMHO clutzy in the extreme and it is a shame that I2S never gained more traction as a consumer standard. I am convinced that for optimum sound the right architecture is to slave the source to a high quality clock sitting right next to the DACs.

As for bad SRC, that can make just about any resolution of source material sound like mp3 as anyone who has done battle with kmixer will tell you. Don't even get me started on undithered digital volume controls, phase-mangling digital filters, etc., etc.

The simple truth is that it is both easy and cheap to get surprisingly reasonable sound from digital, however it is extremely difficult, and often expensive, to get truly great sound.

Jan Vigne
Jan Vigne's picture
Offline
Last seen: Never ago
Joined: Mar 18 2006 - 12:57pm


Quote:
We can laugh about it over a beer in future years if someone discovers that bits have some hitherto undiscovered property other than their ordinal value, such as 'spin' or 'charge' or some quantum-level mischief, which neatly explains why exactly the same bits can produce different sounds in different circumstances.

If digital as we know it today sticks around then I think that might well be the case. I look at the new generation of turntables - each time there is a new generation of turntables which seems to be surprisingly often since the advent of CD's - and I am amazed at the knowledge gained over a few years time into exactly how a closed loop system built around propelling a circular groove past a stationary chunk of mineral has progressed. It has done so primarily through a greater understanding of that closed loop system and how to achieve the ideal of the entire system remaining truly closed to all outside forces. I suspect digital will have such an epiphany also, though digital is a system with many more variables than my understanding of turntables would allow. However, in the end, both turntables and digital players work together as closed loop systems and in such a system each piece is dependent upon the whole for its good fortune and success.


Quote:
The simple truth is that it is both easy and cheap to get surprisingly reasonable sound from digital, however it is extremely difficult, and often expensive, to get truly great sound.

Certainly that's true of most audio products. There have always been the giant killer products available to the "average joe" throughout the last fifty years of consumer audio. The laws of diminishing returns applies to all subjective evaluations and the rules of mass production distort the progress of too many otherwise successful products.

Log in or register to post comments
-->
  • X