New Naim NDX

quote:

Originally posted by AMA:
...
Shall we admit they are all deaf?

AMA -

I tried multiple times to qualify my statements, so as to not make them offensive. Sorry man - I obviously failed in your case.

JS is right. As I have said multiple times here and in previous threads, am simply trying to understand why (from a technical perspective) people are hearing these differences.

If, as you say, S/PDIF receivers are inherently capable of flipping random bits, I (and apparently, Aleg as well) still do not understand how that results in subtle differences in sound quality. Would have expected pops/ticks/clicks, etc., especially if a high order bit is flipped.

But you could be 100% right. I do not know. Am curious - were you ever able to substantiate what your digital engineers told you? Is it documented somewhere that in-spec S/PDIF receivers can actually flip bits? I have no problem admitting that this is new info to me.

Thanks.

Hook

quote:

Sorry man - I obviously failed in your case.

Not at all -- I'm not offended. You are more than wellcome to keep further queries and you have the full right to disagree with my answers. I really want to give you my technical view of the subject as I was very passionate with it (now a bit less). Regarding the "deaf" argument -- it's usually used not to put someone down but to provoke the opponent to publish his own explanation of the phenomenon in thread. If you don't have them -- it's still fine. Please don't take it too close.
I wold love to hear from you on the demo of HDX/CDX2-2 against your current digital transport.
What if you hear the difference? Then you will possibly wish to revise my answers

quote:

If, as you say, S/PDIF receivers are inherently capable of flipping random bits, I (and apparently, Aleg as well) still do not understand how that results in subtle differences in sound quality.

every transmission line can have errors. the spdif specifiction has therefore a parity bit to detect such errors. thus the developer of these protocol consider that bit erros could occur.

quote:

Would have expected pops/ticks/clicks, etc., especially if a high order bit is flipped.

depending on the receiver of this signal is the result of such errors.

quote:

Originally posted by AMA:
...
What if you hear the difference? Then you will possibly wish to revise my answers

I know for sure and from own experience there are differences. I just don't understand how they come into being.

JS seems to suggest the datastream is polluted by some aspects of the data source. But why does that only lead to subtle differences? Random changes to the bitstream should IMHO be sometimes more than subtly audible.

The same goes for AMA's argument about PSU's randomly causing bitflips. That can IMO not lead to a constant but subtle sound difference.

The character of the constant but subtle difference suggests to me something influencing the analogue stage or maybe (something prompted by Jerryct's reply) in the DSP if there is any error-correction or -handling prompted by the detection of errors in the bitstream (parity bit)?

I would still like to know, but I am afraid only one party is able to and appears is not going to shed any light on this

-
aleg

quote:

If, as you say, S/PDIF receivers are inherently capable of flipping random bits, I (and apparently, Aleg as well) still do not understand how that results in subtle differences in sound quality. Would have expected pops/ticks/clicks, etc., especially if a high order bit is flipped.

Not at all. The traditional impact of what is common in vinyl cracks does not extend to the digital bistreams. It show itself differently. The "pop" or a "crack" in digital waveform is not a single bit flip -- the real "pop" or "crack" is a LONG sequence of bits of a particular shape. It's very unlikely that S/PDIF mistakes can generate such a sequence.
But the impact of the single-bit flips is possibly far inferior than vinyl pops and cracks. It deteriorate the sound in a very unnatural way resulting in edginess (hissing), distorting the instrument tonal balance, "halo" around the voices and instruments, timing mistakes resulting in poor soundstage -- actually all the attributes of digital playback which feed the analogue apologists with gloating energy. The regular PS hum/hiss and vinyl cracks are much easier to ignore for a man ear.

BTW Andy was not worried by cracks and pops -- he was mostly referring to the break of the bitstream consistency in case of bit flips. To be more specific the continuous audio bistream consists of the packages which have quite a complicated design including the header bits, checksum control bits, volume control bits, left channel bits, right channel bits etc. Breaking a bit in such a structure can result in complete fail of the D/A chip to understand the whole package. And I agree that this happens sometime resulting in D/A shut-down for this particular package which is another type of the sound distortion. It's very similar to what you can see on the PC screen when watching the video with broken bits (usually during the grabbing). It freezes, it squares, it flips with different colors and then resumes with normal video. In fact, the similar things constantly happen in shorter time frames resulting in overall dithering of original video quality so we can say some players have better quality or less. The same happens with audio bistream as well. If S/PDIF transmission is generally OK then we shall not experience the stoppage or squares but we may not notice the micro-fracturing of the original bistream keeping ourself in a blind belief that it's perfect.

There is another problem -- much bigger then single-bit flip. It's very important for the transport to avoid periodical bit mistakes which can result in audible harmonic contaminations -- such as ringing.
As you can see the bitstream consistency is a very complicated area. It's all called jitter but it has different structure and different spectrum. Some transports can have a very high jitter but innocuous in terms of audio reproduction (due to their stochastic origin). The others can feature very unpleasant audible effects although average peak-to-peak jitter can be very low. If you check stereophile you may find CD5X and CD555 peak-to-peak jitter to be the same. But the CD555 bitstream is definitely cleaner in terms of the audible effects resulting in much better resolution, microdetails, soundstage and timing.

quote:

Is it documented somewhere that in-spec S/PDIF receivers can actually flip bits?

Yes, that's why the audio bitstream protocol is designed with redundancy -- meaning it contains more bits than needed -- in order to recover the bits in case for the occasional flips. But if you read the specs on audio protocols you may find it does not guarantee 100% recovery -- it only treats the most popular mistakes

But one has to consider that under normal circumstances bit errors/flips should be rather rare. One can can even test the occurrence of such errors (for example with the weiss dacs "bit transparency check") and even a logitech squeezebox touch seems to show no noticeable problems.

So the question is if the bit errors are really the reasons for different transport sounding different?

quote:

So the question is if the bit errors are really the reasons for different transport sounding different?

jerryct, there are three different segments which break the original bit sequence from the file.
1. The transport reading mistakes, 2. Noise contamination through cable, 3. Bit flips at S/PDIF receiver.

It's important to remember that the input bitstream itself is usually not faulty -- it's still bit-perfect. Possibly visual analysis of the zoomed waveform can show distinctive valleys between the peaks The problem is that some waveforms through a certain bit-sequence can be distorted in such a bad way so that regular S/PDIF receiver can fail to resolve a bit properly. Of course, another manufacturer can design a better S/PDIF receiver which can resolve all the bits from the same bitstream. I'm not sure modern developers will spend a lot of time on further enhancement of S/PDIF receivers as the world is moving in different direction -- pulling the data from HDD through ethernet directly into the DAC.

As far as I know Naim does not use standard S/PDIF receivers and prefer to design their own. This also explains why Naim transports work so good with Naim DAC.

BTW didn't you question yourself on why Naim designed their own S/PDIF generators/receivers while S/PDIF standard provides a bit perfect transmission (as many forum members believe)?

I think those who seek for response from Naim HQ on S/PDIF bit-perfect transmission may find this fact as an official answer

quote:

Originally posted by AllenB:
This is all getting a bit technical for me as a lowly Architect , but I would say that the power supply to the deliverer of the bit stream does cause an effect before it gets to the SPDIF outlet and thereon to the DAC. Keeping it simple, that explains why transports can sound different. My recent discovery that the Qute has equivalent SQ on the dig out as the HDX supports that view because Naim are quite meticulous with their PS's right throughout their range of products.

I disagree about the Qute vs HDX dig out but what else if new? I can't seem to agree with anyone. Glad it's working for you Allen.

quote:

Originally posted by AMA:

quote:

Is it documented somewhere that in-spec S/PDIF receivers can actually flip bits?

Yes, that's why the audio bitstream protocol is designed with redundancy -- meaning it contains more bits than needed -- in order to recover the bits in case for the occasional flips. But if you read the specs on audio protocols you may find it does not guarantee 100% recovery -- it only treats the most popular mistakes

I doubt if all this is part of the SPDIF specs, or maybe part of some higher level audio coding. As far as I can read about it, the spdif is a fairly simple and straight forward protocol, according to: SPDIF-interface

-
aleg

You have to differentiate between error detection and error correction (i.e. recover). The consumer variant of SPDIF cannot do error correction on the level of the SPDIF-protocol. with the parity bit you can only detect some errors (not all bit errors). Thus after the detection there is no error-free correction.
And SPDIF with a parity does not use a very sophistacted error detection.

Hello AMA,

i could agree on your "three different segments" and would add 4. bandwith-limited link.

But you say "regular S/PDIF receiver". In this concrete case the SPDIF receiver is always same i.e. the DAC. So it does not explain why differenct transports sound different because the problem is not caused by the transport.

Regarding number 2 und 4 do you know this article?
Bits is Bits

They claim that bit errors or Amplitude errors in the digital audio interface of SPDIF are very unlikely. (but unless someone really knows the bit error rate it is only an assumption)

quote:

I doubt if all this is part of the SPDIF specs, or maybe part of some higher level audio coding. As far as I can read about it, the spdif is a fairly simple and straight forward protocol, according to: SPDIF-interface

-
aleg

Aleg, I did not say "SPDIF protocol" -- I told "audio bitstream protocol" which is for example Red Book PCM incorporating CIRC (cross-interleaved Reed-Solomon) encoding with one redundant parity byte per each three informative bytes. SPDIF is the interface which can transmit the various types of bitsreams including Red Book PCM, Hi-Res PCM or DSD or some custom codes.

quote:

So it does not explain why differenct transports sound different because the problem is not caused by the transport.

It is caused by a transport. High quality transport sends flawless waveform so that any standard S/PDIF receiver (a chipset) will resolve all the bits properly. Low quality transports send noisy and randomly distorted waveforms which challenge all modern S/PDIF receivers (some of them doing the job better than others).

quote:

Regarding number 2 und 4 do you know this article?
Bits is Bits

Yes, it's 14 years old

quote:

They claim that bit errors or Amplitude errors in the digital audio interface of SPDIF are very unlikely. (but unless someone really knows the bit error rate it is only an assumption)

Again and again -- nobody claims that S/PDIF bitstream has errors. It's bit-perfect. They are absolutely right that bitstream travelling through S/PDIF is very unlikely to get any bit flipped. But if you take a sample bitstream and distort it's waveform it will still be a bit-perfect signal, none of the bits is flipped -- but standard S/PDIF receiver may be easily fooled with particular sequence of saw-type peaks with delayed or accelerated slopes and ... make a mistake when resolving the bit at the entrance.

If your transport outputs excellent waveform (needs high quality clock and PS) and your cable is OK and your S/PDIF receiver is good and your DAC is re-clocking -- you may get and absolutely flawless bit transmission. Audiophiles will hear this as that all high-quality transports will sound the same through re-clocking DAC. Possibly HDX is already perfect -- I don't know.

quote:

Originally posted by AMA:
Aleg, I did not say "SPDIF protocol" -- I told "audio bitstream protocol" which is for example Red Book PCM incorporating CIRC (cross-interleaved Reed-Solomon)

mmh, are you sure with this?
there is no red book pcm. Red Book is the standard for audio cd and it specifies the error correction but it also defines among other things the encoding of the audio which is pcm (both are on the same level). And this pcm is send with SPDIF without the CIRC (or you send compressed data as DTS or AC3).

AMA - i'm stil not fully convinced that it's the bit errors - all your theorising is based on the premise that the nDAC is immune to jitter. Is it ?

James

PS - perhaps a seperate thread might be in order.

quote:

Originally posted by james n:
AMA - i'm stil not fully convinced that it's the bit errors - all your theorising is based on the premise that the nDAC is immune to jitter. Is it ?

James

PS - perhaps a seperate thread might be in order.

Hi James -

The answer to your question is yes...if you believe the DAC white paper:

"...In the Naim DAC the master clock is not recovered from the S/PDIF signal as usual. Instead the audio data is read from S/PDIF, stored in solid-state memory and then clocked back out to the DAC chips using a fixed-frequency local master clock. This eliminates jitter caused by S/PDIF."

The only other type of jitter is that which was created during the recording process. Of course, on playback, that jitter will be indistinguishable from the original signal.

Hook

quote:

AMA - i'm stil not fully convinced that it's the bit errors - all your theorising is based on the premise that the nDAC is immune to jitter. Is it ?

James

James, nDAC is not immune to jitter. It rejects it at a very high degree, but not completely. That's why different transports sound differently through nDAC. What nDAC is doing is it's removing time-based jitter almost completely. Or possibly completely. I mean if spacing between the bits is floating due to SPDIF generator clock instability then nDAC will still resolve them (by sampling the bitstream with matching fixed built-in clock), buffer them and then reclock with built-in high quality clock. But this procedure does not help for the distorted waveforms so that S/PDIF receiver can make mistakes on these waveforms and this is the moment when the bit is flipped. And it sounds like a veil over original sound.

quote:

Originally posted by AMA:

quote:

They claim that bit errors or Amplitude errors in the digital audio interface of SPDIF are very unlikely. (but unless someone really knows the bit error rate it is only an assumption)

Again and again -- nobody claims that S/PDIF bitstream has errors. It's bit-perfect. They are absolutely right that bitstream travelling through S/PDIF is very unlikely to get any bit flipped. But if you take a sample bitstream and distort it's waveform it will still be a bit-perfect signal, none of the bits is flipped -- but standard S/PDIF receiver may be easily fooled with particular sequence of saw-type peaks with delayed or accelerated slopes and ... make a mistake when resolving the bit at the entrance.

If your transport outputs excellent waveform (needs high quality clock and PS) and your cable is OK and your S/PDIF receiver is good and your DAC is re-clocking -- you may get and absolutely flawless bit transmission. Audiophiles will hear this as that all high-quality transports will sound the same through re-clocking DAC. Possibly HDX is already perfect -- I don't know.

Hi AMA -

Thanks again for clarifying.

Would agree that below a certain quality level, transports can be broken and things may sound like crap. In the DAC white paper, Naim warns us to "watch out for the switch mode power supplies of cheap digital players and the quality of S/PDIF, which may suffer RF noise".

Some rhetorical questions come to mind...

Where is that quality threshold set? Do we really have to spend $7850 on an HDX to hear what the DAC can do? Or does a $1000 transport that "punches above its weight" meet the quality threshold? Seems clear to me, and perhaps just a bit sad, that we are all destined to find that answer on our own.

Also, I think you might get an argument from those who hear a difference between the CD5XS and the CDX2. Cannot imagine them agreeing that the former is below the quality threshold and the latter is above!

Hook

quote:

Originally posted by AMA:

quote:

AMA - i'm stil not fully convinced that it's the bit errors - all your theorising is based on the premise that the nDAC is immune to jitter. Is it ?

James

James, nDAC is not immune to jitter. It rejects it at a very high degree, but not completely. That's why different transports sound differently through nDAC. What nDAC is doing is it's removing time-based jitter almost completely. Or possibly completely. I mean if spacing between the bits is floating due to SPDIF generator clock instability then nDAC will still resolve them (by sampling the bitstream with matching fixed built-in clock), buffer them and then reclock with built-in high quality clock. But this procedure does not help for the distorted waveforms so that S/PDIF receiver can make mistakes on these waveforms and this is the moment when the bit is flipped. And it sounds like a veil over original sound.

Hi AMA -

Claiming that the Naim DAC white paper is wrong has always been one of the possible explanations.

As far as I know, you are the first to make this claim.

Hook

Hello AMA,

wikipedia states "Jitter is the time variation of a periodic signal in electronics and telecommunications"
What do you mean with "nDAC is not immune to jitter" and "nDAC is doing is it's removing time-based jitter almost completely. Or possibly completely". This contradicts as jitter has always something to do with time.

quote:

Originally posted by james n:
PS - perhaps a separate thread might be in order.

Yes, in the Padded Cell.

quote:

Claiming that the Naim DAC white paper is wrong has always been one of the possible explanations.

... or people's interpretation/understanding of the white paper is incomplete.