Basic Questions

[Pluck]

I have a lot of questions about digital media servers. I can approach it from the top down or from the bottom up. I flipped a coin and decided to approach it from the bottom up.

First off, I want to lay out what I think I know. Take shots at that, and I'll go from there. Some of these statements will be in the realm of the obvious, but bear with me. I'm trying to get the vocabulary straight from the beginning so the remainder of the conversation can be useful.

The digital storage medium is irrelevant. What we care about is the encoding, i.e., 16 bit/44.1 KHz or 24 bit/192 KHz, etc., not whether it's stored on an optical disk, a magnetic disk or in flash memory.

From storage to the DAC, the issues are: the performance of the transport, the connections between the transport and the DAC itself.

Jitter is the main enemy. It is the result of small timing problems. Most of it is introduced by the spinning of a disk, be it optical or magnetic. Some jitter comes from electromagnetic artifacts introduced from various sources.

A typical CD/SACD/DVD player is an integrated system containing a transport mechanism, the equivalent of a computer sound card and a DAC.

Once it's out of the DAC, it's an analog signal no different than what might come out of your turntable, meaning that technical analysis of "digital reproduction" stops at that point. An amplifier or a pre-amp doesn't handle a digital signal, and neither do speakers. When we discuss digital sound reproduction, it's a conversation about what happens before and within the DAC.

My main question at this particular moment is the difference between USB and other transfer mechanisms (i.e., S/PDIF) between the storage media and the DAC. What are the various transfer mechanisms between storage and the DAC and how do they differ?

 

[MDS]

The digital storage medium is irrelevant. What we care about is the encoding, i.e., 16 bit/44.1 KHz or 24 bit/192 KHz, etc., not whether it's stored on an optical disk, a magnetic disk or in flash memory.

The storage medium affects access time. An optical disc is an order of magnitude slower than a hard disk. Flash memory is faster than a hard drive but capacities are limited and you will not find any media server that uses Flash as its storage medium.

A media server would use a hard drive and if you are going to buy one, try to find one that allows you to replace the hard drive if you need more storage.

From storage to the DAC, the issues are: the performance of the transport, the connections between the transport and the DAC itself.

The bits would be read straight off the storage medium and sent directly over whatever bus is in use. Given that these things are very much computers, the bus would likely be PCI.

A transport would only apply if the storage medium were an optical disc and a better transport would maintain a constant speed so that the bits can be delivered in a timely fashion.

Jitter is the main enemy. It is the result of small timing problems. Most of it is introduced by the spinning of a disk, be it optical or magnetic. Some jitter comes from electromagnetic artifacts introduced from various sources.

NO. Jitter is a timing error indeed but a storage medium CANNOT exhibit jitter and a cable CANNOT induce jitter. 'Interface' jitter is the timing error introduced between the time the bits are read off the storage medium and the time they actually arrive at the DAC.

A very poor implementation would have jitter on the order of a few tens of nanoseconds (billionths of a second). Good implementations exhibit jitter on the order of a few tens of picoseconds (trillionths of a second). It is a non-issue in either case. Anyone who thinks they can detect a timing error of 30 ps is seriously fooling themselves.

A typical CD/SACD/DVD player is an integrated system containing a transport mechanism, the equivalent of a computer sound card and a DAC.

The transport is what spins the disc and reads the bits. The player includes a DAC solely so you can use the player to convert the bits to analog so you can hear it. If you have any player connected digitally to a receiver, the receiver can do the D/A, so the transport 'quality' is irrelevant.

Once it's out of the DAC, it's an analog signal no different than what might come out of your turntable, meaning that technical analysis of "digital reproduction" stops at that point. An amplifier or a pre-amp doesn't handle a digital signal, and neither do speakers. When we discuss digital sound reproduction, it's a conversation about what happens before and within the DAC.

Yes and No. Sound is analog and our ears are analog. You cannot hear 'bits' (well you can but it would sound like noise). The DAC converts that stream of bits to analog which can then be amplified and sent to the speakers.

Preamps and receivers can handle digital signals - they do the D/A and some companies have introduced 'digital' speakers. Digital speakers simply accept a digital bitstream and have a built-in DAC to convert it to analog.

My main question at this particular moment is the difference between USB and other transfer mechanisms (i.e., S/PDIF) between the storage media and the DAC. What are the various transfer mechanisms between storage and the DAC and how do they differ?

This is way too hard to explain, but here is the short story:

S/PDIF is a data communication protocol used for audio data between two devices - like a dvd player and a receiver. USB is a serial communication protocol and can carry any data. The 'payload' if you will being sent over a USB connection could be in S/PDIF format.

Think about your network connection. You could have a cable modem that is connected to your computer's network card using a CAT5 cable with an RJ45 jack or you could have it connected via USB. In both cases, the 'data' is in IP format (Internet Protocol - the IP in TCP/IP).

 

[Pluck]

The storage medium affects access time. An optical disc is an order of magnitude slower than a hard disk.

In the real world, why do I care about access time? Isn't this simply the time I wait until the very beginning of the recording? Once the record starts playing, doesn't this issue go away?

Flash memory is faster than a hard drive but capacities are limited and you will not find any media server that uses Flash as its storage medium.

A 4GB flash card costs $45, and within 18 months a 16GB flash card will cost $45 as 32GB and 64GB flash cards are introduced. Within three or four years and certainly within five years, we'll have 128GB and 256GB flash cards. Isn't this an argument for at least putting flash card inputs on the front of CD players?

A media server would use a hard drive and if you are going to buy one, try to find one that allows you to replace the hard drive if you need more storage.

Wouldn't a media server have a USB input so you could connect an external hard drive? In fact, why don't CD players have external USB inputs for this purpose?

The bits would be read straight off the storage medium and sent directly over whatever bus is in use. Given that these things are very much computers, the bus would likely be PCI. A transport would only apply if the storage medium were an optical disc and a better transport would maintain a constant speed so that the bits can be delivered in a timely fashion.

How is timing regulated? Are there differences in the regulation of timing between moving disks (optical or magnetic) and non-moving (flash) memory?

NO. Jitter is a timing error indeed but a storage medium CANNOT exhibit jitter and a cable CANNOT induce jitter. 'Interface' jitter is the timing error introduced between the time the bits are read off the storage medium and the time they actually arrive at the DAC.

A very poor implementation would have jitter on the order of a few tens of nanoseconds (billionths of a second). Good implementations exhibit jitter on the order of a few tens of picoseconds (trillionths of a second). It is a non-issue in either case. Anyone who thinks they can detect a timing error of 30 ps is seriously fooling themselves.

The transport is what spins the disc and reads the bits. The player includes a DAC solely so you can use the player to convert the bits to analog so you can hear it. If you have any player connected digitally to a receiver, the receiver can do the D/A, so the transport 'quality' is irrelevant.

I want to be sure I understand you correctly, so forgive me if this is an obvious question. If transport doesn't matter and jitter can't be heard, it would seem to follow that anyone who spends more than $30 for a CD player (the price at Radio Shack) is paying too much. Is that a fair reading of your position?

S/PDIF is a data communication protocol used for audio data between two devices - like a dvd player and a receiver. USB is a serial communication protocol and can carry any data. The 'payload' if you will being sent over a USB connection could be in S/PDIF format.

Think about your network connection. You could have a cable modem that is connected to your computer's network card using a CAT5 cable with an RJ45 jack or you could have it connected via USB. In both cases, the 'data' is in IP format (Internet Protocol - the IP in TCP/IP).

Gotcha. Why do (some) audiophiles believe that USB is less prone to jitter than other types of cables? I have no preconceived opinion, but am asking the question.

 

[MDS]

In the real world, why do I care about access time?

A media server is to store a large collection of music and have random access to it; ie pick any song at any time. Think about playing a normal CD and listening to track 5. You then want to hear track 11. It takes a second for the player to seek to track 11. Add to that the fact that the capacity of an optical disc is much smaller than a hard disk.

Contrast that with having the music on a hard disk in your computer and using WinAmp as the player. The instant you click on another track in the playlist the prior track stops playing and the new one starts (because the seek time of the hard disk is much faster).

A 4GB flash card costs $45, and within 18 months a 16GB flash card will cost $45 as 32GB and 64GB flash cards are introduced. Within three or four years and certainly within five years, we'll have 128GB and 256GB flash cards.

So wait 5 years and see how your prediction plays out.

Wouldn't a media server have a USB input so you could connect an external hard drive?

Some may, most do not at this time. The biggest knock against 'media servers' is that they cost a fortune and you generally cannot replace its internal hard drive. If you already have a computer, you can buy a $300 Squeezebox and you now have a media server.

If transport doesn't matter and jitter can't be heard, it would seem to follow that anyone who spends more than $30 for a CD player (the price at Radio Shack) is paying too much. Is that a fair reading of your position?

Of course not. There are other features of players that would make one a better choice than another. The point is that Jitter is the elusive boogeyman that is cited as the 'reason' for every ill in digital audio when in reality it is inconsequential. ALL digital audio systems exhibit jitter but the levels are so small that they are meaningless.

Why do (some) audiophiles believe that USB is less prone to jitter than other types of cables? I have no preconceived opinion, but am asking the question.

Because you are talking about the type of 'audiophile' that believes magic cables that cost $200 or more per foot and a jar of polished rocks on top of the speakers improve the sound. Said audiophiles tend to have zero knowledge and experience about the technical details of digital audio.

 

[Pluck]

A media server is to store a large collection of music and have random access to it; ie pick any song at any time. Think about playing a normal CD and listening to track 5. You then want to hear track 11. It takes a second for the player to seek to track 11. Add to that the fact that the capacity of an optical disc is much smaller than a hard disk.

I'm not sure that random song playing is the reason I'd have a media server. I'd tend to want it as a repository of albums. But if being a DJ was my thing and I had to wait 1 second between songs I doubt I'd find this a problem.

Contrast that with having the music on a hard disk in your computer and using WinAmp as the player. The instant you click on another track in the playlist the prior track stops playing and the new one starts (because the seek time of the hard disk is much faster).

My impression (correct me if I'm wrong) is that a media server, be it on a PC or elsewhere, wouldn't use optical storage. I'm thinking it would use either rotating magnetic storage or flash memory. So I'm not sure that there's much need to discuss optical storage in the context of media servers anyway.

So wait 5 years and see how your prediction plays out.

Obviously, that's what we'll do. In the past 4 years we've seen roughly a 500-fold increase in the capacity of flash storage. If that rate continues for another four years, this implies a 2TB flash card for $45 in 2011. Insofar as music is concerned, this would imply two things: First, that storage capacity very soon will not be an issue, and second, that flash memory will be king. And no, I am not the flash memory salesman. I do think it's pretty much baked in the cake, though.

As a consumer, I welcome it because of its convenience and relative ruggedness. Notwithstanding your view that the motion of a CD spinner or magnetic hard drive spinner is irrelevant, my gut tells me that "no moving parts" will reduce distortion. My dealer thinks so, too.

Anyway, I asked my dealer why the CD player manufacturers don't put card slots on their machines and his answer was that the high-end audio equipment makers tend to be late adopters of new technology because most of them are small and risk-averse. A bunch of them have painful memories of getting hurt badly by SACD so they're more inclined to sit back and let new stuff shake out for a while. That's what my dealer said, and it makes sense to me.

I see that new PCs are starting to have flash card slots, so I've got to figure that audio equipment will follow suit. On a separate but related topic, I expect that the next version of Windows after Vista will be designed for flash cards. You'll carry your OS, your apps and your data on flash cards. At some point in the reasonably predictable future, your credit card will have multi-terabyte flash storage.

Why not have your personal music and video collection there, too?

Some may, most do not at this time. The biggest knock against 'media servers' is that they cost a fortune and you generally cannot replace its internal hard drive. If you already have a computer, you can buy a $300 Squeezebox and you now have a media server.

Do you think a PC can currently be a worthy component in a high-end audio system?

There are other features of players that would make one a better choice than another. The point is that Jitter is the elusive boogeyman that is cited as the 'reason' for every ill in digital audio when in reality it is inconsequential. ALL digital audio systems exhibit jitter but the levels are so small that they are meaningless.

Generally speaking, in your opinion what separates an ordinary CD/DVD player from a good or great one? Talk in terms of broad categories of their core functions, i.e., the players themselves rather than, say, their looks. I'm obviously not looking for you to write a book here. What distinctions matter to you?

Because you are talking about the type of 'audiophile' that believes magic cables that cost $200 or more per foot and a jar of polished rocks on top of the speakers improve the sound. Said audiophiles tend to have zero knowledge and experience about the technical details of digital audio.

That's one of those religious wars. I'm neutral in Northern Ireland and the Middle East, too. To be more accurate, I tilt toward analog but I'm far from doctrinaire on the subject. As for Northern Ireland, I think they're nuts. Middle East? I'm hoping for solar power so I can quit having to deal with stories about the latest battle on the West Bank.

 

[MDS]

The original question asked about various storage mediums so I simply pointed out the differences. Nobody in their right mind would want a media server that relied on optical discs for storage - we have them already and they are called 'CD players'.

Flash is the future but just not currently sufficient for large collections. I have a mere 525+ CDs which translates to 5,000+ songs and the uncompressed waves currently take up 210 GB of my 300 GB external drive.

With regard to the issue of transporting digital audio from a source device to a sink device that processes the bitstream, you have to remember that the sink device will re-clock the data so the transport is irrelevant. The difference between players is solely the features they offer and the build quality of the player. Cheap players will die in a few years (laser becomes mis-aligned, the drawer mechanism fails, etc) but in terms of sending the bitstream to a processor there is no difference.

A PC is fine as a component for digital audio as it does nothing more than store the data and make it available over the network. A media server tries to encapsulate storage and a convenient user interface for accessing the media into one box and you pay a hefty price for that convenience.

 

[stratman]

Hey MDS,

Any opinion on the Sonos system? A little pricey, but seems like great implementation of hardware.

 

[Pluck]

The original question asked about various storage mediums so I simply pointed out the differences. Nobody in their right mind would want a media server that relied on optical discs for storage - we have them already and they are called 'CD players'.

I thought I'd take this opportunity to tell you that I really appreciate your replies. I'm getting a whole lot out of this colloquy.

Flash is the future but just not currently sufficient for large collections. I have a mere 525+ CDs which translates to 5,000+ songs and the uncompressed waves currently take up 210 GB of my 300 GB external drive.

We agree about the current reality. But it's changing very fast. The "digital guy" at my dealer's place has 500GB of music. When I told him I'd seen a 1TB hard drive advertised for $350, he was ecstatic. Within a few years, it looks like we'll be carrying 1TB around in our shirtpockets. Soon after that, in our wallets. This is only a matter of time, and not a whole lot of it. I suppose I should say that I'm just shy of the age where you get a colonoscopy for your birthday, so I tend to regard four or five years as pretty short. Others see it differently, naturally.

This explosion in storage capacity reminds me very much of what has happened in telecommunications since the 1990s. Fiberoptic lines have so much capacity that it's impossible to overstate. Now, it's a matter of the electronics on each end of the fiber. It's been a few years since I looked closely, but I believe that 40 wavelengths at 10 GB/s per wavelength is standard now, and if anything I'm understating it.

The result is that fiber transmission is damn close to free; to the extent anyone pays for, say, a long-distance phone call, they are paying for switches, maintenance and (especially) the political corruption that allows telecom providers to provide 100-pound sacks of government surplus flour while charging for a catered wedding cake. Telecommunications is one of the most politically-driven businesses there is, and AT&T's various spawn have been masters at that game since before any of us were born.

I think we're right smack in the middle of the same capacity explosion in digital storage, and with no moving parts to boot. CDs and DVDs will go the way of the vinyl record, except that no one's going to get all weepy and nostalgic about their "warmth." I think the music companies are so paranoid about DRM, etc., because they know that storage -- traditionally their means of protecting value -- is vanishing as a relevant factor in their business.

Just like the telecom companies, the media content aggregators have been bribing politicians in a desperate attempt to establish new protections. I think Big Telco has a much better chance of winning their battle, partly because of their political experience and cunning and partly because, once you strip away all the hoo-hah, that game is no less about the physical facilities than it was 100 years ago. We might call 'em routers instead of switches and VoIP instead of PCM, but he who owns the network makes those rules now, just as he did then.

Media content, on the other hand, has no physical form, and that makes it a LOT harder to protect. It's proven difficult to separate a book from its paper, so that looks like a durable business. I think the same goes for many magazines. But newspapers are dying because Craig's List, EBay and AutoTrader stole their classified ads and a lot of people would just as soon read it on a screen anyway. When it comes to music, the idea that people will buy the form factor for the cover art and liner notes is a little bit like when General Motors told people in the 1970s that the Japanese couldn't make big cars. Anyway, the Big Media can prosecute as many college kids as they want, but I really wonder how they'll be able to protect their content.

(p.s.: Sorry for the digression, but this is a topic that really interests me.)

With regard to the issue of transporting digital audio from a source device to a sink device that processes the bitstream, you have to remember that the sink device will re-clock the data so the transport is irrelevant. The difference between players is solely the features they offer and the build quality of the player. Cheap players will die in a few years (laser becomes mis-aligned, the drawer mechanism fails, etc) but in terms of sending the bitstream to a processor there is no difference.

I want to ask some more clarifying questions, because I'm very much in the self-education process on some of these issues. I say that because to some extent I'm going to phrase my questions as statements right here, and wanted you to be aware of that. Pick apart at will.

I wrote above that I understand a "CD player" (by which I mean to include DVD/SACD and whatever hi-def format is used) to include three elements, but I thought about a fourth one as I walked around today. Here goes:

1. Transport
2. DAC
3. "Sound card equivalent"
4. Wiring ("intestinal issues")

My understanding is that, with respect to fidelity, you regard transport quality as being irrelevant as long as the transport is operable within its specs. I also understand you to believe that wiring is irrelevant, i.e., power conditioners are frauds and cabling differences are frauds. Again, tell me if I've misunderstood. Are there any "intestinal issues" in the real world that carry implications for fidelity, in your view?

This leaves the other two. How much difference do you think there is among DACs, and what are the relevant bases of relevant difference, "relevant" meaning having an impact on fidelity? As for "sound card equivalent," I'm really pretty fuzzy on this side of things, so I could benefit from a reasonably concise explanation of what role a computer sound card plays, and how that is recreated in a CD system. I'll probably have followup questions for you.

Once again, thanks for your highly informative and interesting answers to what I've been asking. It's difficult, if not impossible, to find this information elsewhere in any concise format. I had a long conversation with the digital guy at my dealer and we went through it step by step, but I never like to limit myself to one source of information.

A PC is fine as a component for digital audio as it does nothing more than store the data and make it available over the network. A media server tries to encapsulate storage and a convenient user interface for accessing the media into one box and you pay a hefty price for that convenience.

Am I right to think that you're not any sort of believer in the idea that PCs are built by tin-eared geeks, and as a result you think it's a stinking, steaming pile of horsesh!t that they introduce distortion into a high-end music system? If so, then what about simply feeding the DAC via 802.11 from the PC upstairs, off of which hangs a $350 1TB hard drive with all the music storage you'll ever need? What's keeping that from happening right here and now?

 

[MDS]

My understanding is that, with respect to fidelity, you regard transport quality as being irrelevant as long as the transport is operable within its specs. I also understand you to believe that wiring is irrelevant, i.e., power conditioners are frauds and cabling differences are frauds. Again, tell me if I've misunderstood. Are there any "intestinal issues" in the real world that carry implications for fidelity, in your view?

I wouldn't want to say that all of that stuff is totally irrelevant and that there are no differences that can affect fidelity, butthe differences are a matter of degree and one should concentrate on the things that matter.

There are differences in wire (see this site's articles) but they do not matter for the vast majority of people and systems. If you use the right gauge for the length of run of speaker wire, .30 per foot stranded wire will sound the same as $100 per foot exotic wire.

Power conditioners do the job they are designed for but in most cases the problem they solve does not exist unless the house's wiring is very old or in poor condition or the electrical utility is flaky. If you need one, they are invaluable; if you do not, they are audio jewelry.

How much difference do you think there is among DACs, and what are the relevant bases of relevant difference, "relevant" meaning having an impact on fidelity?

This is another contentious subject but I say practically nil. Again it's a matter of degree. I'm not comparing the dacs in the HTIB at WalMart but DACS from well known manufacturers such as Burr-Brown, Wolfson, Cirrus, etc. Without going down the long road talking about all the aspects of digital sampling, I'll just say that practically nobody can distinguish between DACS from the above manufacturers.

As for "sound card equivalent," I'm really pretty fuzzy on this side of things, so I could benefit from a reasonably concise explanation of what role a computer sound card plays, and how that is recreated in a CD system.

The sound card is the interface between the outside world and the other components of the PC. Like any other card in the PC it is connected to the PCI bus so it can communicate with the CPU. It has many functions like performing analog to digital conversion if you use it to record from an analog source, performing digital to analog conversion if you play a digital source like a WAV or MP3 file, volume control, and others. For many of those functions it needs the help of software (the sound card drivers) and the CPU although newer cards have some processing capability of their own and don't need as much help from the CPU.

The inside of a computer is a rather noisy environment (electrically) so using the PC for the audio portion of your media system is less than ideal except for the case of digital audio. In the case of digital audio, all it has to do is get the bits from the CPU and send them out the digital out. It is just one component competing for CPU time though and when the system is heavily loaded you could get dropouts as it cannot deliver the bits uninterrupted in real time.

The fidelity part comes from the device at the other end that processes the bits and does the D/A to send the analog signal to the speakers.

Am I right to think that you're not any sort of believer in the idea that PCs are built by tin-eared geeks, and as a result you think it's a stinking, steaming pile of horsesh!t that they introduce distortion into a high-end music system? If so, then what about simply feeding the DAC via 802.11 from the PC upstairs, off of which hangs a $350 1TB hard drive with all the music storage you'll ever need? What's keeping that from happening right here and now?

Distortion if used for analog signals. The approach you describe is just perfect for digital audio and is the way I approach things except that I don't bother with wireless. My house is hard wired with CAT5 cables to every room and I just use 100 Mbps Ethernet.

Wireless works but is more problematic for streaming music than wired connections. That is not due to fidelity issues but more to network issues. Lots of things interfere with 802.11a/b/g wireless (cell phones, cordless phones, the microwave) and packets can be lost resulting in hiccups in streaming the music.

 

[tbewick]

I wouldn't want to say that all of that stuff is totally irrelevant [transport, power supply etc] and that there are no differences that can affect fidelity, butthe differences are a matter of degree and one should concentrate on the things that matter.

Of course, I'd go along with this, but I do feel that it's worth clarifying some of the points to do with jitter, power supplies, etc. Jitter does reduce the performance of digital audio systems, and can cause modulation noise. Most 24 bit converters, like those in sound cards or cheaper CD players, probably don't even meet the theoretical jitter requirements for true 16 bit performance. The could have an audible effect with some playback material, like sinusoidal signals, but these probably won't be encountered in ordinary listening.

The S/P-DIF interface is susceptible to jitter because the sample clock is sent along with the audio data. It is possible for the receiver to resample the incoming audio data and reject jitter, but the highest quality components, like expensive CD players, often avoid using S/P-DIF and perform the digital-to-analogue conversion internally. To do this, they often have elaborate and expensive designs to eradicate as much jitter as possible. The power supply, it has been speculated, can affect jitter performance, with periodic jitter the major concern. On the other hand, PC sound cards, with all the component interferences, are more susceptible to jitter than high quality, separate high-fi components.

In terms of importance for ordinary listening, modulation noise caused by jitter probably is quite difficult to hear. The problem is that high-quality, subjective tests are quite difficult to get hold of (mtrycrafts has previously referred to the few controlled tests in the past). In my opinion, I'd say that the typical tests done in hi-fi magazines often exaggerate the subjective performance differences between different CD players. I don't think that these performance differences are reliably attributable to objective differences between the players, i.e. like jitter levels, and are more likely the result of not using proper controls in the listening tests.

I've included some links with more information here:

http://www.essex.ac.uk/ese/research/audio_lab/malcolmspubdocs/C27 AES lecture Introduction to digital audio.pdf
C27 INTRODUCTION TO DIGITAL AUDIO, (tutorial paper), Hawksford, M.O.J., Images of Audio, Proceedings of the 10th International AES Conference, London September,1991

http://www.clir.org/activities/details/AD-Converters-Pohlmann.pdf
'Measurement and Evaluation of Analog-to-Digital Converters Used in the Long-Term Preservation of Audio Recordings', Ken C. Pohlmann, University of Miami, Frost School of Music.

http://www.essex.ac.uk/ese/research/audio_lab/malcolmspubdocs/C41 SPDIF interface flawed.pdf
C41 IS THE AES/EBU/SPDIF DIGITAL AUDIO INTERFACE FLAWED?, Dunn, C. and Hawksford, M.O.J., 93rd AES Convention, San Francisco, preprint 3360, October 1992

http://www.ebu.ch/CMSimages/en/tec_AES-EBU_eg_tcm6-11890.pdf
'ENGINEERING GUIDELINES - THE EBU/AES DIGITAL AUDIO INTERFACE', John Emmett, 1995, European Broadcasting Union.

http://www.jstage.jst.go.jp/article/ast/26/1/50/_pdf
Ashihara et al. 'Detection threshold for distortions due to jitter on digital audio'.

http://www.nanophon.com/audio/diagnose.pdf
Julian Dunn and Ian Dennis - `The Diagnosis and Solution of Jitter-Related Problems in Digital Audio' Preprint 3868, presented at the 96th AES Convention, Amsterdam, February 1994.

 

[Pluck]

tbewick, thanks for the post. I'm going to response to MDS here, and then take time with what you wrote and come back at it. I appreciate both of you having taken the time to engage here.

I wouldn't want to say that all of that stuff is totally irrelevant and that there are no differences that can affect fidelity, but the differences are a matter of degree and one should concentrate on the things that matter.

You say that there are relevant differences and then imply that they don't matter. You've been adding a lot of value by being willing to say what you think, come hell or high water. With all due respect, you're sounding like a politician. In a high-end audio context, are the differences relevant or not?

There are differences in wire (see this site's articles) but they do not matter for the vast majority of people and systems.

We're on a site called Audioholics. Who here cares about "the vast majority?" The vast majority buys a Bose radio and thinks they've acquired a miracle. I'm not interested in the vast majority of audio systems.

It's a little like saying that the shape of a wine glass doesn't matter in the vast majority of cases. Yes, but that's because the vast majority drinks Two-Buck Chuck. I drink it myself, and when I do I don't pay attention to the shape of the glass or think about decanting it. But when I drink other wine, I might start considering those issues.

If you use the right gauge for the length of run of speaker wire, .30 per foot stranded wire will sound the same as $100 per foot exotic wire.

Okay, now we have something specific. It seems like you regard wiring irrelevant other than the right gauge. Correct?

Power conditioners do the job they are designed for but in most cases the problem they solve does not exist unless the house's wiring is very old or in poor condition or the electrical utility is flaky. If you need one, they are invaluable; if you do not, they are audio jewelry.

Okay, but how do you know if you need one?

This is another contentious subject but I say practically nil. Again it's a matter of degree. I'm not comparing the dacs in the HTIB at WalMart but DACS from well known manufacturers such as Burr-Brown, Wolfson, Cirrus, etc. Without going down the long road talking about all the aspects of digital sampling, I'll just say that practically nobody can distinguish between DACS from the above manufacturers.

Who makes the DACs in high-end equipment, for example Linn?

The sound card is the interface between the outside world and the other components of the PC. Like any other card in the PC it is connected to the PCI bus so it can communicate with the CPU. It has many functions like performing analog to digital conversion if you use it to record from an analog source, performing digital to analog conversion if you play a digital source like a WAV or MP3 file, volume control, and others. For many of those functions it needs the help of software (the sound card drivers) and the CPU although newer cards have some processing capability of their own and don't need as much help from the CPU.

If I store and catalog digital information on a PC but use an external DAC, do I bypass the sound card?

The inside of a computer is a rather noisy environment (electrically) so using the PC for the audio portion of your media system is less than ideal except for the case of digital audio.

I'm not talking about running my turntable's output through my PC. The PC would be nothing more than a storage and cataloging device.

In the case of digital audio, all it has to do is get the bits from the CPU and send them out the digital out. It is just one component competing for CPU time though and when the system is heavily loaded you could get dropouts as it cannot deliver the bits uninterrupted in real time.

Can't buffering and error correction solve this?

Wireless works but is more problematic for streaming music than wired connections. That is not due to fidelity issues but more to network issues. Lots of things interfere with 802.11a/b/g wireless (cell phones, cordless phones, the microwave) and packets can be lost resulting in hiccups in streaming the music.

Same question as before about buffering and error correction.

 

[tbewick]

[Speaking to MDS] You say that there are relevant differences and then imply that they don't matter... In a high-end audio context, are the differences relevant or not?

I'm sorry to butt-in, but I think that you must try to judge this as us all trying to decide on where to spend our money on hi-fi equipment, and where that money is best spent. I believe that MDS is trying to emphasize, as I am trying to do, that concerns raised about things like jitter or power supplies do have a reasonable basis, but their importance in affecting system performance can be overstated.

An example would be that of digital audio cables, where there is often a lot spoken by cable manufacturers about the importance of the dielectric etc. This is all well and good, but what is important is to what degree this affects sound quality. I've quoted some information from the EBU document I referred to earlier that gives you some indication as to how important the cable is:

'Whether or not a circuit is treated as a transmission line depends on the frequency of the signal and the length of the circuit. With modern equipment, the adverse effects of not matching impedances in cables are mainly caused by reflections which interfere with the wanted signal. Transfer of maximum power is not in itself very important. Thus although analogue audio distribution can suffer from transmission line effects over distances in excess of 1,500 metres, they are rarely met with in studio practice. As a consequence audio cable specifications pay little attention to parameters such as the characteristic impedance or the attenuation of signals above 1 MHz. For data signals like the AES/EBU digital audio interface, the situation is very different. Data signals start to suffer transmission line effects after only ten metres or so. This due to the higher frequencies and shorter periods involved. The system designer therefore has to modify his installation practice. This has to become in some respects quite unlike traditional analogue audio practice and much closer to traditional video practice. Fortunately, for a simple binary signal, there is no need to obey the transmission line rules anything like as strictly as for an analogue video signal.

In fact the original 1983 specification allowed up to a 2:1 mis-match of the line characteristics and this gave a certain flexibility to "loop through" receivers, or use multiple links radiating from transmitters. This concept was based on the theory that lossy PVC analogue audio cables would be used and it was predicted that:

· reflections in short cables were unlikely to interfere with the edges of the signal, due to the short delays involved,
· reflections in longer cables were likely to be attenuated so much that they would not significantly interfere with the amplitude and shape of the signal at a receiver.

In practice, however it was soon found that problems occurred with an open ended spur which happened to have an effective length of half a wavelength at the frequency of the "one" symbol. This length is also a quarter wavelength for the frequency of the "zero" symbol. This condition causes the maximum trouble for the signal characteristics on any connection in parallel with the spur.

It has been found that connectors are of little consequence since their electrical length is so short that any reflections due to mis-match are immediately cancelled out. Surprisingly, some "noisy" analogue connectors, such as brass ¼" jack plugs, work extremely well with digital interface signals. This is because all digital signals, even silence, are still represented by several volts of data signal and, by analogue standards, digital signals are very tolerant of crosstalk.'

- page 11, EBU technical document 'Engineering Guidelines - The EBU/AES Digital Audio Interface', by J. Emmett (referred to previously)

'Dispersion is a phenomena which is largely a factor of the cable dielectric. It provides a theoretical limit to the length of cable which can be used between repeaters. This limit cannot be improved by simple equalisation procedures. Nevertheless, tests have shown that for cables with polyethylene dielectric, dispersion is unlikely to be a limiting factor for lengths below 4,000 m. Therefore dispersion is unlikely to produce any practical limitations within studio premises.'

- page 15, above document.

 

[Pluck]

I'm sorry to butt-in

Don't be sorry at all. I'm glad you did. I especially appreciate those links. I've downloaded the pdf files. There's a lot of reading material there, so it will take me some time to digest it. But thanks much for the links.

but I think that you must try to judge this as us all trying to decide on where to spend our money on hi-fi equipment, and where that money is best spent. I believe that MDS is trying to emphasize, as I am trying to do, that concerns raised about things like jitter or power supplies do have a reasonable basis, but their importance in affecting system performance can be overstated.

Fair enough, and if there's one thing I want to resist it's that temptation to launch into an Internet flame war over semantics. It seemed as if MDS was trying to have it both ways, saying that it matters but doesn't matter. But I'll hold that off to the side for the time being; to call someone a politician is to call them a hypocrite, and them's fightin' words or can be. Besides, I agree with you about the desire to find the best bang for the buck. I'm sympathetic with it to a great degree. On the other hand, there is another way to look at it.

I routinely tell people that for 500 bucks or so, they can have an audio system that will deliver 85% of the performance of a $100,000 system. For $5,000, you'll get 95%. Don't hold me to specific numbers; that would be an exercise in false precision. My point is that the low end and mid-fi ends have improved tremendously. In my view, this is probably the main reason for high-end audio's shrinkage in the past 10 years. By and large, there is a whole lot more good news than bad news here, unless you're a high-end dealer who has failed to adapt to change.

At the same time, however, there is a cohort (small and shrinking, I freely acknowledge) for whom price is, at most, a secondary issue. Those people care to some degree about the cost-benefit ratio, but they care more about the absolutes. That's where I'm coming from when I wrote that MDS seemed as if he was trying to have it both ways. Yes, I do want to know the price tag for A, B or C, but my main interest is in the absolutes. When it comes to the question of whether it's worth paying for them, I don't need any help from the outside world.

And, while I'm truly no snob about high-end anything, what the vast majority does is of very little consequence to me. I fully realize that, when it comes to high-end audio, I long ago slip'd the surly bonds of earth and touched the face of God.

An example would be that of digital audio cables, where there is often a lot spoken by cable manufacturers about the importance of the dielectric etc. This is all well and good, but what is important is to what degree this affects sound quality. (snipped for length)

I think we can safely assume that audiophiles aren't interested in issues generated by cables that run for more than a quarter of a mile. So, if I screen out that part of your post, am I right to think that except for making sure to pay attention to the connector plugs on analog audio cables, you believe that there are no other relevant cabling issues? ("Relevant issue" = an issue that introduces audible distortion.)

I'm asking this not to nail you to the cross but in an attempt to accurately frame all of this in my mind. One of the big value additions in this thread is that you and MDS appear to be (please correct me if I'm wrong) at one end of a continuum on the general questions of analog v digital and whether (many, and especially the expensive) high-end audio tweaks are relevant or whether they're fashion ripoffs.

If I can establish an accurate sense of how you guys approach a variety of questions, then I have specifics to discuss with the people at the other end of the continuum, i.e., the guys who think you ought to run a green magic marker around the edges of all your CDs. Heretofore I've always found those discussions difficult because I haven't had a good sense of the technical issues behind them. This thread represents my effort to start to untangle it so I can make some specific judgments about these issues.

It's not all just theory, by the way. While I do this, I'm thinking specifically of the question of digital media servers. So I have an application in mind. I suppose I could just go buy one and see what happens, and ultimately that's what everyone has to do. But this time around I thought I'd try to really educate myself on a bunch of issues that have always interested me anyway. Moreover, unlike many high-end geeks, it doesn't thrill me to buy stuff all the time. If I can jerry-rig a media server for a while and then wait for the high-end guys to blast it out of water I'd consider doing so, but I need to know what the relevant (i.e., audible) issues are in doing that.

 

[tbewick]

At the same time, however, there is a cohort (small and shrinking, I freely acknowledge) for whom price is, at most, a secondary issue. Those people care to some degree about the cost-benefit ratio, but they care more about the absolutes.

...One of the big value additions in this thread is that you and MDS appear to be (please correct me if I'm wrong) at one end of a continuum on the general questions of analog v digital and whether (many, and especially the expensive) high-end audio tweaks are relevant or whether they're fashion ripoffs.

What I think the problem is, as I've already said, is that differences in performance, say, for example between one speaker cable having a better conductivity than another, are then unreliably used to account for supposed differences in sound quality between those cables. The most important issue when determining audibility is to look at the problem from both an engineering (objective) and subjective way, but both forms of evaluation must be comprehensive and of high-quality.

You're probably bored of looking up these sort of links, but the Audio Engineering Society website has access to a seminar on listening tests held by experts from Nokia, B&O, Harman etc:

http://www.aes.org/technical/documentDownloads.cfm?docID=168

You'll see that the listening tests done by most of the hi-fi press are simply not good enough to determine real, as opposed to imaginary, differences in sound quality. So if What Hi-Fi? say a superbly designed £6000 Naim CD player is 'worth every penny', then that's fair enough, but their method of evaluating the subjective performance will not reliably tell you whether or not that player will sound better than another player costing £200.

...am I right to think that except for making sure to pay attention to the connector plugs on analog audio cables, you believe that there are no other relevant cabling issues? ("Relevant issue" = an issue that introduces audible distortion.)

I don't know much about analog audio cables, personally I only bother spending money on getting good quality analog video cables. I'm sure that buying screened analog coaxial audio cables maybe worthwhile, but I've never bothered with them. People often say a good rule of thumb is to spend 10% of your total system budget on cables.

It's not all just theory, by the way. While I do this, I'm thinking specifically of the question of digital media servers.

Nowadays I rarely use my computer and the internet for music, so I don't know much about this.

One thing I should have said earlier about jitter is that the effect of jitter and sensitivity to jitter will depend on the design of the digital converter. I don't know what the effect of 1 ns of jitter would be on a 24 bit oversampling converter. Around 0.2 ns of random jitter is the requirement for true 16 bit 44.1 kHz noisefloor performance, but this requirement may differ for oversampling converters. Malcolm Hawksford has a fairly recent AES conference paper on jitter which raises this issue:

http://www.essex.ac.uk/ese/research/audio_lab/malcolmspubdocs/C134 Paper 121st convention (corrected).pdf
C134 JITTER SIMULATION IN HIGH RESOLUTION DIGITAL AUDIO, Hawksford, M.O.J, 121st AES Convention, San Francisco, October 2006, paper 6864

 

[Pluck]

You'll see that the listening tests done by most of the hi-fi press are simply not good enough to determine real, as opposed to imaginary, differences in sound quality. So if What Hi-Fi? say a superbly designed £6000 Naim CD player is 'worth every penny', then that's fair enough, but their method of evaluating the subjective performance will not reliably tell you whether or not that player will sound better than another player costing £200.

I've read a few reviews of audio products, but I'm not sure I've ever read a review all the way through other than the one by the guy who really went after some magazine's praise for Wilson Maxx speakers. I read it because it was entertaining as hell.

For better or for worse, I've always picked my equipment by sitting down at my dealer's place and listening. I don't do the extensive double-blind comparisons that some people favor. It's not something I really believe in. I don't follow that approach with other things that I enjoy, such as wine, malt whisky, great food and cigars, so for myself I really don't see the point in doing it with audio.

At the upper end of anything you're paying a lot of money for small differences that won't always be apparent, so to me the blind testing method is almost beside the point. I listen to a system, not to each component. I've always bought that way. I've upgraded about four times in 25 years, each time upgrading a bunch of components at once. I've never kept any particular system for less than five years. The only single-component upgrades I can remember were when I had Vandersteen 3 speakers. I sent them back to be turned into 3As, and again to be turned into 3A Signatures. Each time, they came back noticeably better than before.

Each time I got a whole new system, the new system has been more expensive than the last time and each time it has sounded noticeably, and sometimes radically, better. To the extent that I'm a fanatic about anything it's my loyalty to my dealer, who has always made it understandable and who's never tried to upsell me into trendy and expensive crap that he knows I couldn't afford. Plus, they've got a track record with respect to how everything sounds.

Trust me, I'm not holding out my "method" as the "right" method. I'm just telling you how I've done it. It seems to have worked for me so far, anyway. One thing it has allowed me to do is to not pay a whole lot of attention to the religious wars, at least as expressed in this or that component. It's never been very interesting to me, but now that I'm looking at the whole digital media server issue I'm taking a somewhat different approach and starting to familiarize myself with the various issues without turning into a pointy-headed geek about it.

One thing I should have said earlier about jitter is that the effect of jitter and sensitivity to jitter will depend on the design of the digital converter. I don't know what the effect of 1 ns of jitter would be on a 24 bit oversampling converter. Around 0.2 ns of random jitter is the requirement for true 16 bit 44.1 kHz noisefloor performance, but this requirement may differ for oversampling converters. Malcolm Hawksford has a fairly recent AES conference paper on jitter which raises this issue:

http://www.essex.ac.uk/ese/research/audio_lab/malcolmspubdocs/C134 Paper 121st convention (corrected).pdf
C134 JITTER SIMULATION IN HIGH RESOLUTION DIGITAL AUDIO, Hawksford, M.O.J, 121st AES Convention, San Francisco, October 2006, paper 6864

Thanks for that link and others. As I wrote before, it'll take me a bunch of time to digest all of this stuff because I have other things going on in my life. But I've really been enjoying this thread, and appreciate and respect what you and MDS have been posting here.