Why would one digital ouput have different quality that another digital ouput?

[madeinstein]

I'm still very much confused about the fact that
people say that digital source from one device is different that
digital source from another device which "in theory" doesn't make
sense. In mean binary output should be the same regardless of the
device and why it isn't? or is it really different?

I mean the
receiver side needs to receive full signal because I'm sure just like
in any network device there are checks that make sure that is the case.
In a situation that something is missing the sender is being asked to
resend the data. Also on the receiver side there is usually a buffer so
there shouldn't be a situation that you don't have any data at some
point.
So please if anyone knows why one digital is not the same as another digital please let me know. Thanks.

 

[John Duncan]

madeinstein:I mean the receiver side needs to receive full signal because I'm sure just like in any network device there are checks that make sure that is the case.

No, it doesn't - it interprets what it's given, no more. And it's the source device's ability to read its medium which varies, and why error-corrected lossless files from a hard disk are a good idea (generalising).

 

[fr0g]

edited.

 

[madeinstein]

JohnDuncan:No, it doesn't

Ok so that is the main problem with digital out.. It's simply unreliable!

So now I understand why Linn is using ethernet port in it's DS devices and in my opinion that is the way forward. You need to ensure that your DAC receives the full signal.

 

[madeinstein]

Continuing on the same subject..

That also explains why in Stereophile's test Logitech Transporter was better than Squeezebox + separate DAC. Even if DAC in Transporter is not as good as the separate DAC (maybe it is I don't know) Transporter has the edge of receiving better signal than DAC via digital out from Squeezebox.

 

[Anonymous]

Are you sure? The OP is right as far as I know. All WiFi systems have error checking built in as part of the transmission std AFAIK. So wireless transmission of digitised information should be the same irrespective of source given the transmitter power is sufficient (i.e you are not out of range).

Now converting that digital signal to an analogue signal is a different matter, here as the OP mentions, the DAC will play a role.

 

[John Duncan]

Wifi does (it's file transfer), but the digital stream produced by a player (and received by a digital input) does not, AFAIK.

 

[Anonymous]

JohnDuncan:Wifi does (it's file transfer), but the digital stream produced by a player (and received by a digital input) does not, AFAIK.

Taking CD as the obvious example there is error correction built into the CD track using Cross Interleaved Reed-Solomon coding.

The standard is laid out in the Red Book. but to paraphrase:

The smallest entity in a CD is called a frame. A frame consists of 33 bytes and contains six complete 16-bit stereo samples (2 bytes ž 2 channels ž six samples: equals 24 bytes). The other nine bytes consist of eight Cross Interleaved Reed-Solomon coding error correction bytes and one subcode byte, used for control and display. Each byte is translated into a 14-bit word using eight-to-fourteen modulation, which alternates with 3-bit merging words. In total there are 33 ž (14 + 3) = 561 bits. A 27-bit unique synchronization word is added, so that the number of bits in a frame totals 588 (of which only 192 bits are music). Working backwards you can also see that a frame contains 1/75th of a second of music.

So any decent player will have a fault free digital output. All you need to do is get this into your DAC without corrupting it.

 

[madeinstein]

welshboy the question is what kind of signal is being send via digital out? not how it's stored on CD/HD etc..

When you're using ethernet port you have error correction build in into the protocol you're sending a data with (TCP, UDP) but as far as I understood from what JohnDuncan said earlier digital out has just raw data (bits of information) without any begining/end.

Read a bit more on this subject so yes the frames are being send rather than raw data.

Is that correct?

 

[Anonymous]

Most probably jitter is the main ******. Correct me if I'm wrong, but I believe these are tiny timing errors.

This site gives a nice explanation: http://www.jitter.de/english/engc_navfr.html

 

[madeinstein]

Found more info..

I think the problem with digital out is that the player is sending frames but only in one direction. So if the DAC misses the frame or gets invalid frame it will be skipped as it has no means to ask for this data to be resend. This is the big advantage of ethernet connection.

 

[Anonymous]

madeinstein:

welshboy the question is what kind of signal is being send via digital out? not how it's stored on CD/HD etc..

When you're using ethernet port you have error correction build in into the protocol you're sending a data with (TCP, UDP) but as far as I understood from what JohnDuncan said earlier digital out has just raw data (bits of information) without any begining/end.

Read a bit more on this subject so yes the frames are being send rather than raw data.

Is that correct?

The OP mentioned digital souce in his original post. Strictly speaking the CD or DVD etc is the digital source. Everything else in the "player" is a reading mechanism and DAC.

I wasn't really talking about CD but about the error correction. As you know CD stores PCM (Pulse code modulated) audio data. My point was that a lot of people think you can lose data from a CD during the read process - in simple terms you can't because of the fact that the CD has inbuilt error correction so the data reciever can tell if it has the correct data. Now we come to the choices. The device recovering the data has freedom within the standard to either ask for the data again which is usual or if several contiguous frames are corrupted it can "estimate" what is missed to give what is usually refered to as an inaudible correction.

The benefit (as some people see it) of digital transmission is that signal reproduction does not happen in real time as with an analogue signal. (that is to say there is normally buffering) There is therefore time (in relative terms) to correct errors from the reading of the CD transmitting of data etc. Since the time codes are also transmitted in the frames there is also time to correct the sequencing. It's not like the reproduction of an analogue signal in that regard.

You mention the S/PDIF standard (optical or coaxial digital) and you have no doubt read that this is a development of a high quality proffesional transmission standard. As such there is error correction available in time slot 31 of the sub frame as it happens. So again it's about how the equipment is designed and if error correction is enabled. Note though that it is possible to degrade the digital signal if the environment is electrically noisy. Same is of course true for any digital or analogue transmission cable. At some point with both transission systems the signal is unrecoverable if the noise is too bad. That said analogue tends to hang on longer even if what you end up hearing is unpleasant wheras digital tends to sound OK then fall of a cliff so to speak.

BTW the same factors will impact on ethernet as well - ultimately in an electrically noisy environment it can fail.

Jitter (more correctly seek jitter) tends to be caused by re-sampling errors as the Red book standard does not actually require block accurate addressing. Therefore if the extraction process re-starts a sample of a frame it may get that frame out of sequence. Most good CD players will correct this in software. Normally this is done by performing overlapping reads and sort of sliding the data blocks about electronically to find the overlaps at the edges. Think of putting 2 sheets of paper with the same thing written on them on top of each other with the light behind then and moving them about to get all the writing to line up. The differences (if any) stand out sharply. Astronomers do something similar with telescope images to see the relative movement of celestial objects.

 

[madeinstein]

welshboy:
You mention the S/PDIF standard (optical or coaxial digital) and you have no doubt read that this is a development of a high quality proffesional transmission standard. As such there is error correction available in time slot 31 of the sub frame as it happens.

[...]

BTW the same factors will impact on ethernet as well - ultimately in an electrically noisy environment it can fail.

welshboy I do not agree with your that S/PDIF is the same as ethernet. Please correct me if I'm wrong.

I agree that with S/PDIF you have information in the frame to check if the frame is correct.. But the question is what if it's not correct? The frame is being skipped.. and the next one is received just like JohnDuncan said initially.

With ethernet if the packet is incorrect receiver asks the sender to resend the data and nothing is skipped. That means that skipping is much less likely. I mean you would have to either loose transmission completly.. or several send/receive signals would have to fail which doesn't happen often even in long distance networks.

Of course I agree with you that if S/PDIF sender + cable + DAC is good quality than it should be the same as ethernet.

Ethernet protocols are much more complex and are designed specificly not to loose any data.

 

[Anonymous]

madeinstein:welshboy:

You mention the S/PDIF standard (optical or coaxial digital) and you have no doubt read that this is a development of a high quality proffesional transmission standard. As such there is error correction available in time slot 31 of the sub frame as it happens.

[...]

BTW the same factors will impact on ethernet as well - ultimately in an electrically noisy environment it can fail.

welshboy I do not agree with your that S/PDIF is the same as ethernet. Please correct me if I'm wrong.

I agree that with S/PDIF you have information in the frame to check if the frame is correct.. But the question is what if it's not correct? The frame is being skipped.. and the next one is received just like JohnDuncan said initially.

With ethernet if the packet is incorrect receiver asks the sender to resend the data and nothing is skipped. That means that skipping is much less likely. I mean you would have to either loose transmission completly.. or several send/receive signals would have to fail which doesn't happen often even in long distance networks.

Of course I agree with you that if S/PDIF sender + cable + DAC is good quality than it should be the same as ethernet.

Ethernet protocols are much more complex and are designed specificly not to loose any data.

Of course you are correct ultimatly theS/PDIF standard is design for close to real time transmission of audio data over very short distances. Ethernet is designed for buffered transmission of any sort of digital data over longer distances on a non-dedicated transport structure. My point, from which I confess I think I wandered a bit, was that it's prefectly possible to get reliable transmission of a digitised audio stream using S/PDIF standards.

My other point was related to digital sources i.e. the thing that carries the digital information be that CD or DVD or DAB broadcast. These are the sources of the digital information IMHO.

I also agree that ethernet is specifically designed not to loose data. However that does not make it bullet proof. It is subject to all of the same problems as any other sort of tranmission system.

 

[Gozaradio]

I'm going to have a go at this, but I don't proclaim to be 100% correct...

The way I've been reading this thread, I think that the original question has been a little misunderstood. I understand OP to be asking something tantamount to 'If you have a £50 CD player with SPDIF out and a £500 CD player with SPDIF out, surely there should be no difference in what comes out of the SPDIF socket'

This is something that I once thought many moons ago. As JohnDuncan said, it's about the ability of the player to read the data correctly in the first place more than anything else - you would expect the more expensive CD player to employ techniques that enable it to read as much data correctly as possible. You would expect the cheaper one to not read the data as well in the first place, but to use error correction to pick up the pieces. Therefore, playing the same disc in both machines would produce two different data streams and hence they will sound different (to some people, at least).

I understand what OP is thinking about with regard to Ethernet and error checking but as has been said, SPDIF is one way traffic; no error checking in place. However, even if CD players transmitted the data over ethernet, it wouldn't make any difference as the data coming out of the player could have errors in the first place so if the checking is done against that, then there would be no difference. It would only make a difference if the player then went back to the CD to check for errors against the original but again, there could be different read errors due to the nature of reading a spinning optical disc and you could end up in some sort of error checking loop! It of course begs the question of whether CD's are an appropriate medium to store other sorts of data!

This is why some people believe that Hard disks or Solid state memory is a better medium for reading audio data from - you can have all the error checking and the source medium isn't as likely to have the read errors associated with optical media (or is it?).

Going to lie down now...

 

[Anonymous]

I can;t be bothered reading all these lengthy posts, ask me a proper question and I'll give you an answer.

 

[Gozaradio]

Why is water wet?

 

[Anonymous]

Thanks guys, a very informative thread and very well explained by all.

Now what on earth do you expect me to do with that information ?

(Hunts for box marked ' Info to use if ever unfortunate to get stuck in conversation with old science teacher at school reunion'....Ah here it is. Oh it's empty !)

What on earth happened there. This site gets more messed up everyday

 

[Gozaradio]

Well, you know, I just started with the odd bit of useful data here and a byte of interesting binary there but then it wasn't enough. Before too long, I was scouring forum after forum just looking to shoot up any useless or trivial feeds.

Mine is a cautionary tale: Remember kids - don't do digital.

 

[Anonymous]

andyb190:
What on earth happened there. This site gets more messed up everyday

Cheers Andrew,

Don't suppose you can 'allow links/images' on my account as well could you ?

 

[Anonymous]

Gozaradio:
Mine is a cautionary tale: Remember kids - don't do digital.

Yep, too many dac heads... it's a Cyrus problem !

 

[Anonymous]

I really enjoyed this thread. Thanks very much for a real explanation of how CD data is read and transmitted.

I think the manufacturers have a lot to answer for here. It's clear that when CDs were invented, as an audio storage standard, the CD contains all the data it needs to ensure an error free read can be made. So what excuse do manufacturers have for not properly utilising this and how on earth can they excuse making equipment that doesn't broadcast an accurate bitstream from a CD?

It's a shame that What Hi-Fi don't take up the cause and get a little technical on our behalf. Rather than rating £30k CD players as the best in the world ever, they should be asking why your £200 is not buying a fault free digital read from a disk. What are they spending the money on?

A clear co-axial digital or a clear optical signal should contain as much data as is possible from the disk. Considering how much of this data is meant for error correction and time sequencing (over half from the sound of it) there is NO EXCUSE for the DAC to not reconstruct a perfect analogue sound wave.

Normally I'd say the system should have the ability to ask the transport to check the data. Digital outputs from CD players are one way only but there should be plenty of information in the stream for it not to be necessary to go back and ask again.

So What Hi-Fi, can't you use your influence to get an explanation out of manufacturers as to how they seem to engineer such difference into CD players? It's almost as if they're purposefully sabotaging the technology, that should be fairly faultless at a low cost, just to create an expensive hierarchy of CD players and make the consumer pay more.

Be very interested to hear WHFSV's views and others here. Great thread. (Sorry about this turning into a bit of a rant!)

 

[Anonymous]

Will Harris:

It's clear that when CDs were invented, as an audio storage standard, the CD contains all the data it needs to ensure an error free read can be made.

I'd say more or less the opposite - it's pretty obvious that when they designed CDs, they knew they WERE NOT going to get anything like a clean read in most cases. That's why there's so much error correction designed in, with the fallback of interpolating. We're talking mass produced lumps of cheap plastic with microscopic bumps in, which may have pressing errors, scratches, fingermarks, may be at a constantly varying distance from the read head, with a hole that may not be in the exact centre. It NEEDS all that error correction to achieve a usable end result . It's designed to survive a lot of errors; that's because it expects to get almost that many. So it's not too surprising that high end players could track and recover the signal from the CD better than cheaper ones - but even then, I'd expect the CD itself to have errors.

Will Harris:

the technology, that should be fairly faultless at a low cost

I think you've hit the nail on the head here - it IS _fairly_ faultless at low cost. That's not the same as _completely_ faultless; it just means that almost all of the time cheap players won't get so many uncorrectable errors that you'll hear an obvious click after interpolating the missing samples. And don't forget that when it first came out, there weren't any cheap players...

 

[Anonymous]

Jerry I agree with you. That is so well put!

Isn't it extraordinary though, that after 26 years of having CDs around, they still can't cheaply and reliably, read the disk and correct any errors such that a £100 player sounds different from a £10,000 player. What on earth is going on inside the high end players, and what on earth is it that £100 can't buy you in 2008!!!

 

[madeinstein]

jerryC:

I'd say more or less the opposite - it's pretty obvious that when they designed CDs, they knew they WERE NOT going to get anything like a clean read in most cases. That's why there's so much error correction designed in, with the fallback of interpolating. We're talking mass produced lumps of cheap plastic with microscopic bumps in, which may have pressing errors, scratches, fingermarks, may be at a constantly varying distance from the read head, with a hole that may not be in the exact centre.

JerryC, I think you missed the whole point about CDs.. CD information is stored in digital format, the error correction is to make sure that you've read all 0101010 correctly and it has nothing to do with the music. What you're talking about is true for old fasioned black analog disks. The quality on CD depends how it was converted to digital format in the first place in the music studio.

But anyway what in your opition is the best storage for music? A better alternative to digital format (either CD or HD).