ND5XS to NDAC - which digital interconnect?

Thanks totemphile - the DC1 is a similar price to the Chord, so at that price level, these seem to be the obvious alternatives, so I need to do some listening.  I'd be interested to hear which interconnect other people with this combination are using.

NDX -> NDAC I use DC1 with ferrite chokes clamped on to minimise the transformer effect

I stream to my nDAC with a squeezebox.  I have borrowed a number of coax cables for testing, including DC-1 (RCA->BNC), Vitus, AQ Hawkeye, Analysis Plus Oval & Zu Ash.  I think DC-1 gives the most natural sound.  However, at a New York City naim dealer, i was told Transparent makes a good coax cable that is a better match than DC-1.  Unfortunately, the HK dealer doesn't offer a loaner for testing. Has anyone tested this?

It doesn't matter.

The nDAC is ambivalent to the interconnect (RFI through coax excepted). I hook up my PC streamer to my nDAC/XPS/52/2x135s/DBLs with a £3 3m long optical (so definitely no RFI as it's optical). I have a mate who streams a similar system with cheapo coax through a Y splitter. We can't tell the difference between interconnect methods. Full stop. Period. End of.

I can (and have described in minute detail on this forum) why it doesn't make an iota of difference which cable you use with the nDAC. Have a search for posts by AMA on here if you think I'm B/Sing.

That really you Andy?   What's it been, a year?   You been lurking?

Regardless, I hope you had a nice sabbatical, and are back to posting here regularly.

If you really want to have some fun, look for a good post from Simon-in-Suffolk, and then challenge what he says.  Who knows, maybe it could launch another marathon thread!  I do think it would be good entertainment to see you two fountains of knowledge debate some aspect of digital replay!

Haven't done extensive testing in this transport-to-nDAC area.  When a PC was my streamer, I assumed the worst in terms of noise, and used a 5M optical cable (a nice one -- a Chord Optichord) to achieve maximum separation.  It sounded really good.  When I ordered the NDX, I also ordered a DC1, and it sounded even better.  

In theory, it is easy for me to imagine how RFI and/or vibration through coax, and even EMI through the air (from a really broken source) could create differences in sound quality.  At the very least, I would not question any forum member who claimed to hear such differences.  I can also imagine how these negative effects could become even more annoying in the most revealing systems.

But what I really don't understand about your post is why you think agreeing with AMA would lend validity to your statement?  

Hook

PS - Just teasing Arthur...you know I dig your act! 

Hi Andy, your reputation precedes you :-).   But contrary to Hook's post, I see reason  to challenge and agree with your points. To support and build on your points..

1) ndac is ambivalent to the SPDIF interconnect except for conducted RFI, I agree

2) additionally I say and can view reflections with a scope, bad reflections from a characteristic impedance mis match causes distortion of the waveform and increases the chance of false detection of a transition.

3) with Toslink, I agree for regular non super hidef bandwidths  over very short lengths there is not much difference over fibre types ie glass and plastic. Longer lengths and inferior fibre startto distort the waveform and increase chance of false detection.

4) Toslink is however potentially sensitive to mechanically  induced distortion of the waveform,unless carefully cushioned. Again this can impact 3).

Hook perhaps you are referring to how I  hear that transport clocking jitter can ultimately compromise the performance of a DAC through coupling? But that's another  matter..

Simon

Guys - I am genuinely in awe of your technical Chutzpah but do you think you may be over thinking the problem - just a little?  There has to be a balance surely between finding the best quality and - you know - actually buying stuff and enjoying it?

Tog

Yep you are right, it's about enjoyment first, but when you find your enjoyment is not where it should be its nice to know why  or have a convincing theory why, otherwise life becomes one long magical mystery tour. Personally that would drive me insane.

Simon

Ahh... the impedance mismatch and false detection argument. Assuming a working interconnect, you simply don't lose bits in S/PDIF interfaces. You can't have false detection because if you did, the system just wouldn't work at all. It wouldn't lock onto the signal because you would be as likely to lose the framing bits as you would the data bits and the data stream would fall apart. Basically, if the stream is playing, you aren't losing bits.

Impedance mismatch just increases jitter - i.e. you just can't precisely time when you get that data back. However with the nDAC, it completely removes that jitter by decoupling the DAC section through a buffer and high precision clock.

The only way it can sound different (assuming bit perfect playback) with different transport mecs/streamers and or cables is if the S/PDIF signal and it's associated noise affects the analog stages as the signal travels from the input across (actually above if you look at the nDAC internals) the circuit board to the DSP in the nDAC. If this happens, Naim really have dropped the ball as they could have got rid of this level of interference quite easily. 

This only holds for DACs that are designed as the nDAC is. I can easily accept - and explain - why different transports/cables can and will sound different through different "standard" DACs. 

As to why I've been absent - primarily because I'm a non-believer - or rather I'm a believer in a different church. Yes, I have a high-end Naim system and yes, I believe (actually I know!) it sounds better than 99% of systems out there. It is now a single source system - all my music and all my films and timeshifted TV are streamed through the nDAC - in fact it gets more use for video rather than audio. The source is always the same - a £200 HTPC running xbmc connected via a £3 3m optical cable. I hung up my anorak a good few years ago - the system simply works and plays beautiful music. To make it any better would be serious, serious money (active with 135s or 500 series components in the analogue sections). It certainly wouldn't be in more expensive streaming components. 

Hi Andy -

If I understand your post correctly, you left the forum for a very good reason:  you reached your "saturation  point".  Back in October of 2010, our good Canadian friend Stoik said something (in a discussion about whether a HiCap was a good upgrade) that I thought was very interesting, and it has stuck with me:

"...Is it worth it, considering what your actual source is? The best answer I could give you is to try out a couple of those top flight DACs from Naim, Weiss, dCS, etc... and see if you're able to appreciate the difference of not, preferably on an extended trial period.

The goal is not to notice a difference, because it's going to happen. It's more about if it matters to you, if you're able to really appreciate it or not.

That's what I'm calling finding your saturation point. It's a point where your expectations are met, you can't be any happier. And where better equipments could do things better and differently, but you're not able to appreciate it yet, so it won't make you any more happier. At that exact point or above, you're saturated."

Of course, you are certainly not the first to take concept this one step further.  After all, if you are truly satisfied with their stereo setup, then the decision to go expend your intellectual energy someplace other than the Naim forum can certainly make sense! 

So whether you stay, or leave, or just do annual cameos, is up to you.  Suffice it to say that I have always enjoyed your contributions, be they on this, or other topics.  After all, there is a lot more to this forum than just the "why is the nDAC so cheap" discussion!   I hope you decide to make music recommendations or, if video is your thing, would love to see you describe your adventures in the Home Theatre forum.  But back to our favorite topic...

It was two years from the time I bought the DAC to the time I bought the NDX.  During that time, I tried two PC's (a DIY server with an RME 9632 card, and a laptop with a M-Audio Transit USB converter), an Oppo universal player and a Denon DVD player as sources for the DAC.   The differences among them were negligible.  In fact, I enlisted the help of Mrs. Hook to do quite a bit of blind a/b testing.  Among her many endearing qualities, there are two that are very relevant here:  1) she has seemingly unlimited patience for helping me with a/b's, and 2) she has very sensitive hearing.  When we watch TV together, either she wears soft ear plugs or I wear headphones -- there is rarely a good compromise on volume level.  I digress, but I think it is for a good reason....

She, too, could not reliably tell the difference between these sources.  And when I first posted this, AMA was among the first to suggest that my system may not be revealing enough to allow me to hear the differences...and he may have very well been correct.   During those two years of DAC ownership, I was very lucky financially, and my system grew from a Nait XS to 252/300.  And my DAC received first an XPS2, and then a 555PS upgrade.  Then, one day, my dealer sent me an NDX to play around with...

Our a/b testing resulted for the first time in a clear preference for the NDX as a digital transport for the DAC.  Mrs. Hook picked the NDX either 9 out of 10, or 10 out of 10 times in each and every test I constructed versus the other sources.   These tests were reinforced by my longer term testing over the next few weeks, where I found myself spending more and more time listening to digital rather than vinyl.   Of course I appreciated the NDX's reliability, and all of its convenience features, but it also just sounded....better.   So out the window went my firmly held belief that all sources for the DAC sounded alike.

Funny thing is, since the NDX went into place about 9 months ago, my desire to discuss this topic has gone way down.  I read, but rarely contribute to the many threads about Amarra, JRMC, and so on.  I ignore Noogle's bits are bits posts.  In fact, my interest level in digital replay has only recently been rekindled by news of this month's HDS announcement, and the thought of replacing both the NDX and DAC with a single box.

But as you say, we can reduce the conversation to different churches.  We are certainly all free to worship wherever and whenever we like, and we can simply agree to disagree.   In fact, I hope you and Simon will forgive me -- it wasn't very nice of me to prod you guys into a schoolyard brawl.   If this really has become a religious issue for you, then like with most religious issues, there is probably no point in pursuing it. (e.g., it's kind of like arguing women's health issues with a pro-lifer.....it never goes anywhere, and it is certainly never any fun).

Hope you choose to stick around Andy, and hope you decide to contribute on a lot of different topics.  But if you don't, that's cool too.  It was nice to see you again.  Good luck!

Hook

I will shortly need a Digital I/C for CD5XS to DAC and was going with DC1 but spoke to a dealer yesterday

who advised not to but use Chord Signature instead. He claims the DC1 imparts too much coldness in addition to the already analytical DAC. For the DAC to 202 intend to use Cadenza. 

Hi Hook,

Yes, I was (and still am) happy with my understanding and my point of view. A little background if I may...

My day job means I have to have a fairly complete understanding of digital A/V systems and how they work at the lowest of levels. I project manage set-top boxes from low level driver code to production which includes everything from board design through complex system software stacks. I understand how a digital broadcast system works - both the algorithms used to encode the data as well as the system level of how data is muxed, transmitted and then pulled apart at the box end before it is decoded and presented by the TV/audio chain. I've been involved in this for the past 20 years and even if I say so myself, I'm quite good at my job - you have to be to avoid the B/S that some pass as fact and that can seriously derail a project if you don't expose it. That software and system knowledge is backed by a degree in Electronics (far too many years ago now) so I'd say I've got a fair understanding of how these systems work - even down at the transistor level (well, maybe 25 years ago I might have but... )

In all of that, everything has cause and effect. Everything is understood - there is no black magic. There is not one element in the chain that isn't understood as to why something behaves as it does. If something is not understood, it is investigated until it is understood and then (normally) corrected.

And then we come to high-end audio. Here, there is the unerring belief in the ability of the audio manufacturers to conjure seemingly impossible feats that no other engineer is capable of producing, let alone detecting and therefore explaining. Do not get me wrong, once a signal is in the analogue domain all bets are off and the personal cunning and knowledge of the audio engineer is king. Having said that, the good ones will know why things sound how they do. But drop back to the digital domain and the cause and effect are much easier to predict and understand.

Clearly, the nDAC has analogue electronics in it - it produces an audio signal at its output, but all the clever digital processing is there to deliver precisely timed data to the conversion. The data itself is always correct, it's just how it is presented to the DACs that matter. And the nDAC takes very great care to present the data to the DACs at exactly the right time. Save micro effects of hash in the S/PDIF interface affecting the analogue output stages the transport is irrelevant in the nDAC or it doesn't work as Naim claim. If that hash is noticeable, the electrical design should be better!

Back to a point you made about the HDX. I don't doubt you hear a difference between different transports with the nDAC. Having said that, if I were a betting man, I'd put money on the HDX doing something to the data before it is output through the digital outputs. Either that or the nDAC hasn't been designed optimally and really is affected by RFI derived from jitter in the S/PDIF reception circuits. To have put so much effort into the design and then have it fail because they didn't negate the effects of RFI at this point in the replay would be an incredible oversight - wouldn't you agree?

Exactly -  either through rininging or low pass filtering the transition can't be determined accurately at a point in time.

To have a complete misfire is somewhat rare but I can can count with some equipment and non 75ohm CI wire - and can be easy to blank when parity is used for the frame, if parity is not used then you will have bad data. However the odd bit of bad data, depenmding on manifestation, can be hard to detect by ear.

Yes you are looking at transport jitter and sample jitter being ideaslitically seperated. Indeed they are with the nDAC and many designs. However switching EMI casued by the incoming trasnport clock can cause perturbations on the sample clock through coupling - of ususally the powerlines or ground currents. Therefore the two in the real world are not completely decoupled.

You can see this on a sesnitive spectrum analyzer and  for my sins was an horrendous learning expieirience when I was designing video frame stores at a London University many years ago. I had ignored the real world couplings even though my clocks were completely decoupled and reclocked.

And of course the big school boy error when considering sample data, its not about 1 or 0s or the sequence of the 1s or 0s being correct. Its about point in time of the 1s or 0s - and of course that is the cornerstone of digital signal processing.

Simon

Umm... to a point. If the link is broken enough to lose data, then it can lose ANY bit of data in the stream. Given the way S/PDIF is implemented as a set of frames and subframes which has less than half of the bits allocated to data bits (each 16 bit sample is transferred as a 32 bit word plus there is frame packing) if data errors were causing this, the effects would be far from subtle. I.e. in the real world, S/PDIF is bit perfect.

So... let me rephrase what you're saying to see if I understand it.... What you're saying is that any jitter coming through the digital circuits get transmitted back up the powerlines and affect the regulation of the power supplies of all the other components. In the nDAC when powered from an external power source (which powers the DAC section as far as I'm aware), this would mean that this jitter is transmitted back up through the mains and back into the separate power supply powering the A/D conversion process. I would think with modern household distribution and what goes on in most peoples houses, there would be significantly more for the power supply to reject other than a slight modulation induced by one tiny bit of electronics some feet away.

I can't contribute anything to the technical discussion, but I can relate an interesting incident.

I have a CDS2/XPS and a UnitiServe/nDAC, that are exchanged occasionally between our city apartment and our country home, as sources for two different systems.

For the past 3 months, the CSD2/XPS was used as the source in the country home (rest of system : 82 / CB HiCap / 250, NAC A5, 2-way 8” speakers using Peerless components) but last Friday evening I replaced the CDS2/XPS by the UnitiServe/nDAC, using the DC1 interconnect (BNC-BNC) and a new power cable on the nDAC. The sound just wouldn’t gel ; thin and uninviting. Lack of warmup ?, mood ?, the new power cable ? Replacing the new cable with the stock one didn’t solve the issue, neither did warm up, as the problem was evident the following morning. I was starting to wonder what exactly I had seen in the UnitiServe/nDAC that had incited me to move on from the CDS2/XPS and I was really starting to miss the warmth of the latter combination. Then the penny dropped. I checked the DC1 and noticed that I had inadvertently installed it the wrong way around. Reversing the cable brought back everything I love about the UnitiServe/nDAC combination ; the CDS2/XPS is finally up for sale.

BTW there are no WiFi networks in my neck of the woods, nor had I installed one to control the UnitiServe.

I have tried another well-regarded digital cable also using BNC-BNC connectors, but found that it accentuated brightness. Repeatably. The DC1 just makes the UnitiServe and nDAC perform like a single component, to my ears. The importance of cable directionality was a bit of a surprise though.

FWIW

Jan 

@Andy S:

Bravo!

It's a dirty job, but someone has to do it.....

Meanwhile the snake oil manufacturers and salesmen make a decent living. If I were one of them, I'm sure I'd want it obfuscated too.

I also have nothing to contribute on the technical side, and honestly it really doesn't matter.  You ears will tell you everything you need to know.  You dont listen to music with a white paper or an advanced degree.

I have always wondered why/how I could hear a click when I minimized my monitor for the first time after reboot.  Subsequent minimizing and maximizing of the screen does nothing as it should.  Closing and re-opening the media player didn't change anything.  ONLY after rebooting, then opening up J River or Foobar (didn't matter which player), the first time I minimized the player I would hear a clear audible click.  Not that loud but obviously there.

I am sure there are plenty of explanations for this phenomena. 

Regardless of the nDACs capabilities, this PC source obviously sends sounds that other sources may not.  Somehow some video driver is overriding ASIO, and the muted system sounds settings, and sending audible information to the nDAC.

Understanding that, why is it UTTERLY IMPOSSIBLE for all of these various "sources", all of which have some degree of "computer" inside, to do strange things like sound different.

Maybe someday someone with a high level of knowledge will explain in layman's terms how and where in the nDAC White Paper it says that the nDAC is 100% perfect in every possible way regarding source leveling, clock selection, jitter handling etc.

It always seems like these discussion come back to "if the nDAC works like the White Paper says it does..."

I have read that thing at least three times.  Of course most of it goes right over my head.  But I have yet to find any absolute language as I read in posts here.  There is no mention of TOTALLY eliminating source jitter, ALWAYS having a perfect clock, etc.  It seems to use language like "reduces", "maximizes", "attempts" etc...

It says "significantly reduces RF noise, NOT "eliminates ALL RF noise" 

Things like that..... anyway.

Cheers!

Patrick

To quote the white paper:

"In this way the data entering the DAC chips is completely isolated from the incoming jitter."

I think completely is pretty definitive. I'm now expecting someone to say that this is only "cable jitter" and not "source jitter". As far as the DAC is concerned, it doesn't know nor care where it comes from, it sees jitter and just gets rid of it.

Part of the problem is the white paper was written by a techie who understands how the system works. As a techie who understands the underlying technology myself, I can read what is written and infer how the system works and why it does what it says as I work on a day-to-day basis with similar levels of detail and in the same subject areas. Even if the language isn't definitive from your point of view, there is only one conclusion that can be reached if you understand the underlying technology.

Andy, not quite,it's the EMI coupling from the switching logic. I gues it's quite counter intuitive unless you have had the expierience of having to design through it and measure it with test equipment I realise it might be hard for some on the forum to follow, but as I am sure you as a computer and electronics design engineer like me you reguarly appreciate EMI coupling, and earth loop  and Powerline decoupling issues and how clocks can be modulated, albeit well within spec with extremely low level perturbations and probably have your favourite design patterns to mitigate it ;-)

Simon

Yes, I appreciate how EMI/RFI can affect these things. HOWEVER, the board is totally separate electrically and the spectrum of extra RFI introduced because there is further jitter on a 1-bit bitstream will be negligible given everything that is going on with the DSP there chattering away on the memory system. Additionally, if this were the case, you'd get variations between samples as even the tolerances will vary between different units.

If this is the case, you should be able to eliminate it by putting the circuit in a Faraday cage. They're quite cheap these days for circuit boards. Again, something that would be easy and cheap to design in and remove the effects of the problem. 

Andy, yes Faraday cages, or more commonly these days thin bent aluminium is often used in much of the equipment I use, as well as twisted power lines and multilayer PCBs can achieve a faraday effect and reduce capacitive effects. In the perfect world these measures are totally effective. I don't want to be critical on this forum, but the decoupling on current Naim equipment is not foolproof.

Perhaps in the forthcoming NDS, these area have been tightened.. Well have to wait and hear...

BTW you  are right about the DSP  the streamer boards, and this why some of us hear differences between FLAC and WAV andeven TCP window size.

I suspect the decoupling on the DSPmicro controller is quite effective, and the synchronous aspects of the DSP is driven by the master DAC clock, albeit the i2s gated.

Simon

Hi Andy -

NDX, not HDX, but I am pretty sure I would have heard the same improvements over the RME 9632 with an HDX as well.   But to your point...

I do not know if the NDX is doing something to the digital stream before handing it off to the DAC, but I certainly wouldn't be surprised to learn that it does.   Unfortunately, whatever takes place inside the NDX is proprietary, hidden from view, so all we can really do is speculate.  My thinking is that there is something akin to what's going on with the various player software packages...

In the Mac world, I (along with many others) have long wondered why Amarra, Pure Music, and the others sound better than iTunes.  If they are all delivering a bit perfect data stream, then what differentiates them?   Then I read on asylum and some of the other fora that the simple answer is:  it’s the math. Amarra's Jonathan Reichbach says:  "You really do need to know when to round, when to truncate, when to use 32 bit vs. 64, when a multiple vs. a bit shift, etc. You need to get it all right."

Maybe it's just me, but I fail to see how the math can be THE most significant issue with bit perfect playback.  I would have thought that stuff like this was handled inside the computer's CPU.  But this guy has a lot of street cred in the software development world, so maybe I'm just not understanding which algorithms he is really referring to.  If it is some form of DSP, then they shouldn't claim "bit perfect", right?

Instead, as you suggest, I think it all comes back to noise.  The nDAC aside, it is obvious to me that, in general, DAC products are all over the map in terms of their ability to deal with RF.   So if an Amarra can, for example, do a much more efficient job of managing the data flow from disk to output buffer, this should help to minimize the electronic and mechanical activity inside the computer and, therefore, reduce the RF noise (that I think we all agree) can effect downstream DAC performance.

I think that most of what the NDX probably does falls into this area as well.   I think of it is an very well-optimized computer that does a great job of making the processing path operate as quiet as possible. This allows the nDAC on the other end of the S/PDIF connection to do its thing with a minimum amount of effort and, therefore, a minimum amount of internal noise effecting its own analog output circuitry.

I may have missed a step or two in a very complex chain, so don't beat me up too bad if I did.  But until told otherwise, I think the Amarra analogy for why the NDX sounds better probably isn't too far off.

As far as the nDAC goes, I'm guessing that Naim has done a good, but not perfect job of rejecting RFI in its S/PDIF circuitry.   My other guess is that Naim has done a good, but not perfect job of shielding the nDAC's analog circuitry from the the internal noise generated by its own digital processing.  So, I am guessing it sounds better when it has to work less hard to process its digital input.  Also, and this is something that never gets mentioned, but the nDAC is, after all, a first generation product!   To assume that there were no oversights, to assume it has no bugs, to assume that its design has absolutely nowhere to go for improvement seems like a really bad assumption to me.  Wouldn't for example, removing the internal PS, and powering the DAC strictly from an external PS (with separate rails for analog and digital) make for a quieter working environment?

For me, the NDX is about more than just an improvement in sound quality -- it is about reliability and ease-of-use.  I have a limited amount of time to listen to music, and now I no longer have to think about keeping a computer OS updated, installing the latest patch for its playing software, or for the S/PDIF card or USB converter driver, and so on.  I don't have to worry about some link in a complex, multi-vendor chain breaking.  In my ten months of ownership, I have had but one NDX downtime:  45 minutes for me to load the new firmware and pop in the new 24/192 streaming board.  That process was easy, and worked perfectly, and I would bet the farm that I am now looking at (at least) another 10 months of zero downtime, and of zero faffing about with getting my player to work.   Sorry for the Naim commercial, but I do tend to get myself a little worked up on the issue of the NDX (and ND5 XS) value prop (versus Macs and PCs).

Back to the nDAC:  the question going around in my head is whether or not the next iteration of the nDAC (the one that will improve on the limitations and anomalies I suspect it really does have) will make it even more source agnostic.   I would think it probably will, so this leads to a bit of a purchase decision dilemma.  If one invests in the new NDS (with its internal DAC *and* digital output), what does one do when the gen 2 nDAC comes out?  If one is convinced that Naim will be delivering this gen 2 nDAC in the next year or so, it would seem to me to make more sense to wait and see how it sounds with the NDX as its source...

I digress, but in my mind, all of this stuff seems interrelated......and it makes my head hurt to think about it.  More fun to just listen to tunes.  Am going to go record shopping this morning so, hopefully, it will be a vinyl evening!

Cheers.   And oh yeah, am so happy to see that you and Simon have been properly introduced. 

Hook

Allen interesting post. I woulnt however think about separating the DSP from the DAC. Certainly with Naim for example the DSP and D to A converter chip are all part of the 'DAC' function.

The reason is because mathematically in the process of going from a discrete to continuous domain you will alter the information, specifically the ratio sum of all the frequencies that were in the original continuous or analue signal. ( this is because you can't look at infinite number of samples at once)

The electronics designer or DtoA chip designer will try and compensate for this using a transformation filter. This may exist partly in the digital, analogue domain or both.

This filter function is subjective, and usually adds to the 'sound' of a DAC. 

Where the filtering is done in the digital domain it's done by the DSP. Many dac chips have a built in DSP or you disable it in favour of your own which I believe is what Naim does with the NDX, ndac and almost certainly with the forthoming NDS. This transform filter function is intrinsically linked to other techniques used to avoid frequency distortion such as oversampling. 

Simon