Dear HA readers/forumpeople.

I'm currently doing some investments in my HiFi gear and would like some input from anyone who has clues about how to solve my little problem.

As it is right now, most of my collection of music is based on FLAC. So far so good.

To bring the most out of the sound, I'm buying an external DAC with both AES/EBU input and SPDif input.

Here is my little dilemma.

So how do I connect my computer to the external DAC with minimum ehhrr make that real minimum loss of data.

As far as I'm concerned, I would prefer not actually having a computer near my stereo, or at least not a biggie box, rather something like a minimac or similar (ehrm, more likely something similar).

Now here's the question... how would I do this without having any loss/minimal loss on the way? ARe there any units with say TCP-IP connection and just passes on the info with AES EBU (or SPdif)? Would this require any specific software on the computer? Is it a reasonable idea, to connect something to the network that sits just next to the DAC and have error correction all the way there, so that there's no loss to the info? WHat unit might that be?

Please bare with me here, what I'm interested in is not "good enough" devices, assuming I want it perfect all the way to the DAC, what would I use and how would I do this?

Squeezebox?

Well I actually own one, which has disappeared somewhere... dunno where. Lost it like a year ago...

But still... does Squeezebox have AES EBU out? Is it reliable enough to handle the sending to the DAC?




Ehrm... seems more like Slimdevices "Transporter" is what I'm looking for. That's for the high end it seems. But are there optional units. This sucker costs 2000$.....

From the few media streamers I've tried or demo'd- AirPort Express, linksys, netgear - all appear capable of streaming a media file decoded to wav from your computer to the device and then out to the DAC flawlessly.

I've had better success with optical outs than coax (RCA not XLR) only because there is a lot of EM interference area - poorly made cordless phones, illegally pumped up CB and car service radios, TV, etc

Which brings me to the real problem with the wireless devices - network signal interruption. I use an Airport express now and was forced to use a CAT5 Ethernet connection because of the constantly dropped wireless signals.

If you have very little interference in your area, you'll prob be ok. But if you're in a dense urban area like me, wireless streaming may have dropouts.

Interesting. I've never been a fan of Wifi anyway so that ain't a problem to me. However, what remains then is this "device" you refer too. As this device should be preferably like the Slimdevices transporter BUT without the DAC built in. why pay money for that when my intention is to use an external one anyway. To be noted, it's a Forsell Air Reference or a Bremen Licence No 1 DAC that will be used...

What about investing in a professional sound card for your computer that has AES output?

Or you can try this -

http://www.motu.com/products/motuaudio/traveler/

It's the MOTU Traveler. It has Firewire, ADAT Optical, or SPDIF digital in, and ADAT Optical, SPDIF, or AES digital out (as well as a bunch other stuff). It retails for around $850.

Or this for your computer -

http://www.m-audio.com/products/en_us/Audi...le192-main.html

24-bit/192khz
4 in/4 out
SPDIF digital I/O with PCM and surround
Retails around $199.95

Or a 15 USD cmedia card which can only output a bitperfect 96kHz/16bit stream at max, but that's more than sufficient considering your needs.

It's funny, that 'should' work just fine. And while adequately flowing bitperfect data, it may also induce noise or other anomolies to the signal. How, why? I don't know but it happens. Using a cheap audio card or the motherboard digital out still somehow transfers noise. I've compared a few different soundcards over the past couple of years and am now moving to a USB-DAC which gets the audio signal processing out of the PC entirely. I'm using an M-audio Delta DIO now which is a good card, not 'great' like a Lynx... but getting the sound processing out of the PC is where it's at.

I'm not sure if the original poster is looking for a streaming device or an alternative to the standard digital out of a PC/mini-Mac. I'm contemplating a new USB DAC developed by Scott Nixon for just that purpose. I've heard some good things about the M-Audio linked above but it's probably not 'good enough' for my audiophile tastes. I've heard good things about the Transporter... I own 2 Squeezebox 3's... and have experimented with modified versions and upgraded (linear) power supplies on them... and they all have their own 'flavor' to the sound also.... Just depends on which one suits you best.

If you want bit perfect, from source all the way through the amplifier, there are other options as well. That would be TacT gear... and enters a whole new world. Once you're in that realm, it's up to your ears to decide what you like best.. .and not what product is better/worse.

Best of luck... sorry for the ramblings.. Thanks for having me here... (first post! woohoo!)

How can this possibly be if the data stream is bitperfect and the digital-analog conversion hasn't happened yet? What you just described is impossible if the output is bitperfect.

This post makes sense, at least to me =)

but to clarify, getting it straight all the way to the AMP isn't necessary. It's getting it all the way to the DAC which is my issue here. Since SPDif and AES EBU is both one way signals, I would prefer having the unit transmitting as long as I can with error correction etc. Meaning, I want to go through say TCP IP to something which just retransmits (but does that VERY well) through AES EBU to the DAC. And yes, internal sound card is simply not "enough". It has to be external, however, it ha to send flawless over AES EBU. It's really hard to tell if Transporter would do that.... but perhaps you've tried just using the Squeezebox for the SPDif transmitions and have that work flawlessly?

The DAC is the same as seen on these pictures
http://hififorum.knaak.dk/edison/IMG_5423%20kopia.JPG
http://hififorum.knaak.dk/edison/IMG_5420%20kopia.JPG

What you are saying is extremely absurd!!!

You probably got resampled, processed (trough software equalizers), and/or analog output instead of digital, without noticing.

Why don't you go back and re-evaluate your "dealings" with your sound cards to find out where did you make a mistake. Before posting such ridiculously illogical affirmations.

Obs.: Usually audiophile = ignorant.

Correct me if I misunderstand, but how will your ears tell a difference, if the audio is always "bit-perfect"?

Killer test:

Play "ordinary" 5.1 DTS CD (44.1 kHz sampling freqzency, 16 bit, "stereo") via your digital connections, and look, if your amp/decoder can decode it properly or only noise comes.
Or: Play a HDCD through your digital connections, and have a look, if your amp/decoder shows the HDCD logo or anyhow it shows, if it recognizes the hdcd content.

If one or both of these 2 little tests are ok, no "noise" or any changes like evil internal resampling creeps over your digital connections into the digital signals, otherwise the 5.1 dts or hdcd infos would be destroyed.

And DTS/AC3 files work (bit-)perfectly on a 15$ cheap cmedia PCI card and my drivers. No interference, no crosstalk, no noise - the PCM signal gets perfectly transcoded and -mitted without any loss.

To pass your data in an untouched way to your S/P-DIF port try a look at this directshow filter:
http://ac3filter.sourceforge.net

It is able to pass even PCM data to your S/P-DIF port and leaves it untouched (http://ac3filter.net/doc/ac3filter/ac3filter_eng.html#filter_setup_system).

What can I say? I'm a nut, obsessed, an 'audiophool'. I've been comparing the effects of audio gear for over 10 years, doing ridiculous tests at a near-scientific level. I had the same assumption everyone here I'm sure does... bitperfect=bitperfect. However, I'm merely suggesting that something changes, adds noise, or otherwise degrades the presentation beyond the single requirement of 'bitperfect'.

Until you do the experiment, in a controlled manner, in a very revealing setup, you won't hear any difference... and you'll never believe me. And that's OK. You will not suffer my affliction... and live a long and happy life.

Best,
Carl

I can successfully stream DTS encoded wave files via DirectSound through the kernel mixer and through my el-cheapo chinese crap card to my receiver without any loss. I don't even have to worry about feedback loops because I use a TOSLink cable.

Other drivers process the signal (or in other words, add random noise to it ), or the hardware messes with the signal (Creative...).

Cool, I'm glad you can check that. I do not have the tools to do that... but I know someone who does and it's a big effort to be able to check my data output from digital devices. Kudos!

Using TOSLink vs. Coax makes quite a difference also in my experience. Using a Coax with BNC connections, creating a true 75ohm impedance vs. RCA-types makes a difference... and so does the length of the coax cable... Again, all of this is probably 'in my head' but that's what I've experienced... and you'll find many 'audiophools' have discovered similar things.

If you don't already, try glass TOSLink instead of a cheap plastic one and see if you hear a difference.

I know processing adds noise. And I'm not talking about loud noise that everyone can hear, I'm talking tiny differences, blacker backgrounds, that sort of thing...

I've unmapped my audio driver from the window kmixer and noticed a nice improvement in smoothness or less digital 'hash' to the sound.

Getting digital to sound like analog (vinyl) is my goal... but without the clicks and pops.

-C

If your cables aren't transmitting bit-perfect audio, that's a whole other story. Otherwise, their type or manufacture will have no bearing.

Sheer logic tells me that, if the same audio is being passed and played through the same output device under the same listening conditions, it should/will sound the same. I don't understand how anyone can believe any differently.

I'm saying their type or manufacture has a bearing on transmitting bit-perfect audio... as well as rejecting EMI and RFI (with coax, anyway).

Wouldn't something that transmits light benefit from being 'clearer'?

-C

Care to back any of that up with measurements?

Anyway, the easiest way to test if it's bit-perfect is, as mentioned before, to pass an AC3 or DTS stream to your receiver. Any change to the data will give you an earful of noise.

What kind of measurements? If you want scope read outs of the effects of various cable types, that will take some time since I don't have that equipment... but if you have patience, I may be able to arrange it.

You can measure resistance yourself with a multimeter from radio shack. There are published methods on the web. I don't often do this... but I have in the past to prove a point...

If your only means to determine bit-perfect is whether a device creates an earful of noise vs. sounds like music... then there's not much point in me putting effort into 'proving' what I have experienced. We're in different worlds.

The points I'm making deal with incremental improvements of a musical presentation... not whether a device functions or not.

-C

Probably the most important kind here: ABX or other suitable blind tests. Eliminate any preconceptions i.e. the placebo effect. Check the Hydrogenaudio Terms Of Service (yes, I have seen them before ) However, since we're talking about, or at least desiring, digitally bit-perfect streams - why not compare source and destination and see what results?

If it is a digital signal being transmitted there will be no difference. It is as absurd as claiming that the Law of Gravity does not apply on Earth.



Exactly. That’s what makes your experience irrelevant. About “audiophools”: This world is full of people with low IQ. Audiophiles are no exception.



Again: If it is a digital signal being transmitted there will be no difference. It is as absurd as claiming that the Law of Gravity does not apply on Earth.



Smoothness?! Digital hash?! Those things don't make sense when applied to sound.




Why?! If digital is proven to be better than vinyl.

You need to research topics such as: digital electronics and boolean algebra. So you can understand how absurd are your claims.

Wow... I did not see that ABX was actually encouraged here. I've yet to see an ABX that wasn't attacked because it wasn't a perfect or 'valid' test. I prefer to have meetings at my house and we do these types of tests but they're not double-blind.. just single. Someone else knows what they're doing but I don't know what I'm listening to.

I've done this a lot in the past which is why I made my comments...

Lightfire, I will go study now to see if I can be as well-versed in gravity as you. And, thanks for the laugh.

Many people don’t understand that a digital audio signal behaves completely different than an analog one.

Degradation is not a factor as it would be with an analog signal. If degradation occurs in a digital signal being it can be 100% detected and corrected. Consequently it is irrelevant what type of media is used to transmit a digital signal. That will not affect the quality of the sound (it only affects the quantity of information being transmitted by second.) The digital signal does not consist in music and/or sound. It consists in “instructions” that tell your DAC how to “make” that specific sound.

Here are some interesting links for your studies:

http://en.wikipedia.org/wiki/Digital_electronics

http://en.wikipedia.org/wiki/Boolean_algebra

http://en.wikipedia.org/wiki/Digital_audio

True, both cheap plastic and expensive glass will transmit an optical signal. However, high quality glass has very little internal reflection and does not micro fracture when handled. Both of these can introduce errors which must be corrected at the DAC. In general, I think we can all agree that the less error correction a DAC or any digital component has to do, the better.

It doesn't matter if the amount of information being transmitted takes 99% of one certain type of media's capacity and only 1% of a second (better) type of media's capacity. As far as it is enough for one’s need there will be no sound difference whatsoever. And in this case the choice is obviously the cheaper one, freeing resources to improve other areas (analog) of the system where it would really make a difference.

I think you're missing the point - the internal reflections and diffraction caused by micro fractures in the plastic can degrade the optical signal to such an extent that the DAC has to guess (error correct) the corrupted information.

I'm not arguing whether the error correction is audible on any given system - I'm simply saying that every effort to avoid putting the DAC in error correction mode should be taken, especially when it is not costly to do so.

No. I'm not. The DAC should not guess. Error detection and correction are 100% accurate. Whenever a package of information is damaged it can be retransmitted. Better quality media will allow less correction; consequently the amount of effective information transmitted in a given time frame is larger than in a media of less quality. But if the "capacity" of the media in use is already sufficient there is no need for upgrades. It would be illogical to do so.

I may be mistaken, but as I understand it, a consumer outboard DAC is a digital signal receiver - it cannot alert the CD player, Media Streamer or sound card that a PCM signal is corrupt. It can only decode what it gets at it end. This differs from the way an optical NIC card operates - a 2-way error corrected conversation.

A DAC can only interpolate (guess) based on the preceding and following packets what the garbage should have been. This takes place in buffered real time, as the audio information is a time coherent stream. There isn't much opportunity in such a system for a "do over"

Coax digital is particularly susceptible PCM stream corruption from EMF interference - and this is audible as drop outs. I've experienced it first hand.

Sometimes a broader view helps:
An 44.1/16 audio stream will have about ~1.5MBits of bandwidth, so about the bandwith of your home DSL.
Now, glass fibers are used in high-bandwidth backbone transmission. What bandwith do you think those can handle? Well, your average fiber will have about 1000 times the bandwith of an audio stream! The error rate achieved with this bandwith can be about 10 to the -12 (0.000000000001). All over a length of several miles (how does that relate to the few feet used in home cabling?).
By signal regeneration, error detection and correction and so on, minor errors are corrected automatically (in audio, if an error goes uncorrected, a frame will usually be garbage and sound like loud noise for a fraction of a second, and/or the stream will lose sync and be silent). Using the cheapest commercially available optical cables. All at a thousand times the bandwidth of audio, without huge problems. How big can the problems with audio be, then? Remember, digital audio cables were invented to ELIMINATE the (possible) problems with analog cabling.

So, if a 1 is a 1 and a 0 is a 0, how can there be a difference? It would be the same as saying the picture you sent to a friend by mail would be received "with less detail" and "less natural" if it happened to pass through a plastic fiber instead of a quartz one.

Digital 'harshness': In at least 3 courses at university (EE), you learn about sampling theory. Sampling and reconstructing will mathematically proven give a PERFECT reconstruction of the signal, with only 2 restrictions - BANDWIDTH (by sampling freq., for 44.1kHz this is 22kHz; 2kHz for the filter transition band and 20kHz for data - no human can hear beyond 20kHz physically) and NOISE (16bit ~98dB SNR, which is even more than sufficient for listening to a cannon shot right after listening to falling needles - see the thread "Why 44.1/16 is enough). Digital audio is just like analog, but without its errors and hassles.

Placebo-effect can be heavy, heavy, heavy. Just experienced this myself. I transcode my music to my mobile, and once I though I'd try a hard-coded crossfeed DSP. When I heard the new tracks, I thought "Whoa, this is great! Much better SOUNDSTAGE! Really NATURAL! Less TIRING! Next week, I'll reencode all of the tracks with this..." - but, once I found out that "use DSP" was unchecked in foobar's discwriter, all of this 'greatness' suddenly disappeared (along with a chunk of my ego). So, non-double-blind: Avoid, avoid!

So...

If the DAC becomes confused by the flaws in the cable, it will still have to make a guess about whether that bit should be a 1 or a 0. If it has enough information from oversampling and other techniques involving nearby data, it can make an educated guess about what that info should have been. That guess may be what was in the original music, it may not. In general it will be pretty close. Whether that is audible is debatable.

If on the other hand, the DAC can't make any sense of that data packet, a audible click or dropout will be heard.

What the DAC cannot do is request a resend of that packet. Under these conditions, doesn't it make sense to minimize the chances for error?

BTW, some DAC's have indicators that light when error correction kicks in, so it may be possible to do a side by side to see if it matters with your cable.

That would be all jolly and fine, except for two things:
a) the diameter of industrial grade glass fibre cables is way, way smaller than the diameter of the TOSLink stuff => the LEDs can't radiate their signal effectively into the cable => you're screwed.
b) the SNR of a 5m long TOSLink is still high enough for a perfect reconstruction of the signal. A better cable won't do jack s..t

In other words, expensive spdif cables are an effective idiot tax.

Ehrm.... I appreciate the debate or specifically Carlmans input here. Great cheers.

To me it's clear that most think of it in a science way, say stuff and don't get the facts first.

BTW, I've found my Squeezebox which has both coax and optical out. They've just recently added the transporter product as mentioned before..

Check this site: http://www.slimdevices.com/pi_squeezebox.html
FAR more expensive than a 15€ cheapo card.
Still, I like their honesty, please read under
Digital S/PDIF outputs
Intrinsic jitter: less than 50ps (standard deviation)

Basically, there IS jitter, meaning the signal is flawed. Can't judge how much 50ps is at the moment, but it DOES affect the DAC. Since the DAC, correctly, has to guess when there is jitter. As it's not like a TCP IP connection with CRCs being done, it's just one way, and real time.

This is the exact problem I'm trying to solve....
Right now that I've found my Slimmie, which will work "just fine" for now indeed, I can manage to have 2 way communication all the way to it, and from there send a One way signal to the DAC with the 50ps jitter there is.

But indeed, digital signals suffer from FAR less loss than analogue signals, but ANY one way signal WILL suffer loss at some point I'm afraid.

thanks for help right now.

Oh and Carlman... I would love a reference about how much 50ps really is etc, but since this debate is garbled a bit, don't hesitate to send a PM to me if you so like.

Cheers...

A decent receiver reclocks the spdif signal in order to avoid jitter induced degradation of sound dynamics in the high frequency bands. And the critical jitter time is roughly 250ps for CD Audio (in the sense that the jitter induced noise is lower than noise from the quantization).

Unless the diameter of the fiber optic cable is less than 1/2 the wavelength of the light transmitted, it will have no effect. As all the fiber we are talking about is measured in millimeters micrometers, this is not an issue.

The issue is fiber alignment, which is handled by the thingamaboodle at the end of the Toslink. It makes sure the fiber is lined up optimally with the LED, and there is a fair amount of leeway here.

In any case, high quality fiber comes in many diameters, and all the up market optical cables I've seen use a larger than standard diameter cable.

Why does larger help? From what I understand, It only matters in tight turns, in which the light is more likely to travel along an optimal path - as opposed to hitting the outside of tight turn and being lost to internal reflection.



Not necessarily.

High quality glass fiber is very uniform both in crystal structure and density. Coherent light is conducted uniformly with minimal dispersion.

Plastic optical cable, on the other hand is rife with imperfections - wavy structures, bubbles and areas of increased density.Plastic is getting better but at the quality we are buying, no where near the equal of glass.

From what I understand, these imperfections have two effects: to disperse the coherent light in many directions, and at density interfaces, effectively slow down the light by altering the angle if incidence (think light bending when going from air to water.)

Edit: Plastic also tends to micro fracture far more easily than glass fiber. These fractures are both density interfaces and random angle refractors. (I believe the reason for this is that glass is a liquid, and its crystals can flow to some extent, but I could be wrong)

In a real time one way system, this can introduce errors, both in data content and timing. To the extent possible, the DAC corrects these by interpolation and educated guessing.

I have also read that there is something akin to a "skin effect" with glass fiber, such that , unlike with plastic fiber, on the inside of outer surface of the glass fiber, light tends not to be reflected at random angles. This helps maintain a coherent signal, though I admit I do not understand the physics of this.

A decent receiver reclocks the spdif signal in order to avoid jitter induced degradation of sound dynamics in the high frequency bands. And the critical jitter time is roughly 250ps for CD Audio (in the sense that the jitter induced noise is lower than noise from the quantization).
[/quote]

Obviously science disagrees with you... read this to get a better understanding of the problem
http://www.tnt-audio.com/clinica/jitter1_e.html

Clearly there is jitter in the transfers to some extent.

I made a DAC with a very standard SPDIF receiver chip (CS8412), and noticed the robustness of the digital connection on first testing. As the fibre gets near the connector I start to hear bursts of sound with clicks and pops, then as the connector gets nearer the socket the sound becomes normal - still some way to go before it is properly mated. This was with a 5m fibre of the cheapest type. Clearly the SNR margin at the input was large.

It is just not a problem if the equipment works at all.

Ken

The issue is not whether the cable is able to carry the wave or not, the issue is that TOSLink uses LEDs without optical focussing instead of LDs. Therefore, a large diameter of the cable core is required (1mm). Optical communication systems use cables which are usually one or two magnitudes smaller (10..100µm) than TOSLink cables.


Quartz glass isn't a crystal but an amorphous liquid.


Dispersion != scattering.


Even though plastic fibers have a freakingly higher attenuation than glass fibers, the SNR at the end is still high enough to perfectly reconstructed the signal due to the small data rate.


OK, you didn't understand what I wrote. Please read again.

Ahh, anyway, just received my DAC finally. The model seen on this ad
http://cls.audiogon.com/cgi-bin/cls.pl?dgt...&1172247767
however, not that actual one.

The one I have is the last actually produced in the series (as far as I've been told). What a fine piece of machinery, don't you all agree? =)

This far, it makes the Benchmark DAC 1 and below to wonder if that really is a DAC or a cheapo 15€ card =)

Again, the alignment is handled by the nature of the connection, and since the high quality Toslink cables are also large diameter, it is not an issue.



True of course, but glass flows so slowly that it can optically behaves as optically as a chemically heterogeneous but structurally uniform crystal.



True, but in this case, the light is both dispersed and scattered.




The issue with error correction in this system is not bandwidth. The issue is that there is no two way communication, so there is no opportunity to use that bandwidth to get repeat packets. If the DAC doesn't understand what it is being sent - an error - it only has the data on hand to "reconstruct" the data. This means guessing.

Think UN translators at the general assembly. They repeat, as best they can understand, what the speaker is saying in different languages. It happens in real times with a short delay. If the speaker says something they don't understand, they cannot ask her to repeat it - they have to guess at what was being said and move on. Usually what they say is pretty close to what the speaker intended, but sometimes they fail.

Yes the error correction is robust. And yes, you can get music. It will sound very good. What your experiment doesn't address is the extent to which error correction continues with one type of fiber vs another when it should be unnecessary - a stable connection.

I've never said that the imperfections in plastic were serious enough to create major problems. That would contradict the obvious performance of plastic cables. I'm not even arguing that the imperfections create errors that are audible in all systems.

What I am saying is this: the lower quality fiber can introduce errors, and doesn't it make sense to use higher quality cable where these errors are not an issue and do not need to be corrected by the DAC.

Sorry, I seem to have not stated my point clearly enough: the SNR margin is high, that means that there are very few errors. The system is static on the timescale of data frames, so all that matters is the SNR. With the 5m of very cheapo fibre the SNR is big (in my case, but I guess also normally). Also the errors sound so obvious when they happen that there is no question whether they are there or not (click, pop, splat ...).



[as an aside of little importance: the incoherent light from LEDs in large plastic multimode fibres is far less likely to lead to problems when the fibre is moved etc. than coherent light used in monomode glass fibres]

Ken

What you heard are large errors that could *not* be corrected. You may not have heard the errors that were corrected - situations where the DAC was able to reconstruct the data with reasonable, though not exact fidelity.

It is the latter that may be reduced with better fiber



Interesting about the incoherent light. Though the coherence of the light is a function of the transmitter, not the fiber, and glass can and is used for both multi mode and mono mode.

"What you heard are large errors that could *not* be corrected."
Yes - that I know, thanks.

" You may not have heard the errors that were corrected - situations where the DAC was able to reconstruct the data with reasonable, though not exact fidelity."

No, exact, not reasonable - exact. As in perfect. (You do know how it works, I assume?)

You are perfectly right! I would not have heard those non-errors.

Ken, for the last time - you may be pleased to learn

This horse, I fear, is dead.

DAC error correction is "best fit/best guess". The definition of an error to the DAC is that it does not know with certainty what the bit value is. Either it sequence is nonsense, missing or over samples containing that bit don't match within tolerance.

So the DAC makes an estimate - by oversampling, and failing that, various reconstructions. It has no reference - no checksum, no resend, no parallel stream with which to compare. These algorithms typically use the samples immediately preceding and following.

Often the result is the same as the original bit. Sometimes it is not. Usually is sounds pretty good. But it is not exact. Depending on the equipment, music, size of the error and quality of the reconstruction, we may not hear the difference.

This differs substantially from the error correction employed in data transmission, in which packets are verified against known checksums and any errors are remedied using a duplicate packet in the original transmission or a request is sent to resend the data. In these cases, the error is detected, and not "corrected" per se, by remedied by replacing the questionable bit with verified sound data.

This sounds a lot like CD error correction after a c2 error and not a DAC error correction. I would be very interested in a DAC datasheet that gave details of error correction because as far as I know it processes whatever value it gets. Perhaps there's a receiver chip which has this built in but word data errors are very rare (like never with a normal length cord) at this bandwidth but jitter is a whole other matter where cable might actually make a difference. A good buffer and clock will solve most of this however so I really don't think the type of cable is going to make any difference. To prove it why don't you sample from a CD source (or preferably an actual wav/mp3 file played on a dvd player to avoid errors on the CD itself) using a plastic and glass cable then compare the files (accouting for offset) and see exactly how often word errors occur. I don't think you'll find any.