New Naim NDX

quote:

Originally posted by AMA:

quote:

And as was discussed for weeks on multiple Naim DAC threads:

"...And all this "noise" translates into jitter - which is removed at the DAC as it reclocks."

Which is proven to be wrong and all the heralds of "transport-does-not-matter" message have disappeared after the first serious trials.

Hi AMA -

What "serious trials" are you referring to?

Thanks.

Hook

quote:

Yep–you fell into that one–if you look at the files downloaded from the Naim Label download store, you will notice that there are additional metadata files included which, when installed into the same directory as the audio files, provide the Naim servers with extended metadata that you will NOT get with an out-of-box configuration of Twonky, sorry.

Not really David as if you re-read my post i said on SQ grounds. I just wanted to check that by a 'superior product' you were just referring to peripheral items such as meta data hanling etc etc, not the HDX being a superior UPnP server on pure sound quality grounds

James

quote:

Hi AMA -

What "serious trials" are you referring to?

Thanks.

Hook

Hi, Hook. My personal trials are a good judge for me. You may find a couple of dozens trials performed by other forum members. By word "serious" I mean that some people were very surprised at the unique nDAC ability to lift up the performance of cheapish DVD players at substantial level. But the prolonged careful listening reveals quite a big difference between the transports.

quote:

Originally posted by AllenB:

quote:

Originally posted by David Dever:

quote:

Originally posted by james n:

quote:

Why do that when you make (as an actual server) a superior product?

So David, if we take ripping out of the equation as that's a whole different ball game ...

Say you have a downloaded album from the Naim labal and it resides on a NAS are you saying ( on pure SQ grounds) that the HDX serving this file to the NDX would be a superior UPnP server to say Twonky (or any other UPnP server) serving this album out to the NDX. I'm sure you're not but i just wanted to check ?

James

Yep–you fell into that one–if you look at the files downloaded from the Naim Label download store, you will notice that there are additional metadata files included which, when installed into the same directory as the audio files, provide the Naim servers with extended metadata that you will NOT get with an out-of-box configuration of Twonky, sorry.

The structure of the server utilizes these additional relational arguments to provide a browse tree that requires no additional "custom trees" to be created on the part of the end user, i.e., it just works. (Of course, disc ripping works in the same way.)

Unlike Twonky, files which are located elsewhere from the internal storage pool can also be added (without re-location), embedded tags parsed and added to the searchable relational database–simply by enabling Network Share access to the files' location (as a shared folder) from the Naim server (can be performed using a web browser).

Again–and I've said this before on another post–ease of use is a buyer's consideration that (in spite of one's own capability to create cost-effective workarounds) differentiates products within the outer market, especially as a complete out-of-box solution.

That is the reason why the Naim servers make great demonstration tools for UPnP streaming players, and, as such, I see no reason to demonstrate using a less-suitable tool.

This is similar to the set-up I am about to venture into at home, only with a Serve & Qute. The Serve will rip all my CD's and also act as a server to the Qute.

One noticeable aspect is that the Serve (and HDX) are able to transcode files on the fly when acting as a UPnP server, which I didn't discover (on the borrowed HDX) until after I have been busily converting my ALAC files to FLAC (in readiness)!! This was because the Qute will not play ALAC, although I believe the NDX will (stand to be corrected on that last point)?

Not a big deal, I would rather not transcode on the fly if I can help it.

As this is the NDX thread, can anyone see any advantages of an NDX over the Qute in my proposed set up (with Serve and all through the nDAC), I have looked very hard at it, weighed it up and see none, but I would be interested to hear other views.

Allen

Why not just use the serve on it's own to start. It's got the upgrade PCI inside.

This has been gone over before. He wants the Qute for internet radio duties.

He's not interested in the NDX, or a non-naim internet radio solution.... just to nip those potential questions in the bud.

quote:

Originally posted by David Dever:
We did three or four different demos, depending on the crowd:

1. CDX2 via Hi-Line DIN connected to analogue outputs
1b. CDX2 via DC1-BNC connected to NDX digital input
2. UnitiServe-SSD via DC1-BNC connected to NDX digital input
3. NDX playback of UnitiServe UPnP stream across the network (wired Ethernet)

I would be very interested to hear what the consensus was, if any, on the ordering of sound quality of these four options.

quote:

Originally posted by AMA:

quote:

Hi AMA -

What "serious trials" are you referring to?

Thanks.

Hook

Hi, Hook. My personal trials are a good judge for me. You may find a couple of dozens trials performed by other forum members. By word "serious" I mean that some people were very surprised at the unique nDAC ability to lift up the performance of cheapish DVD players at substantial level. But the prolonged careful listening reveals quite a big difference between the transports.

Hi AMA -

Thanks for clarifying your comment.

So why do you think it is that nobody has ever presented a competent technical explanation about how it is possible, given the DAC's buffering/re-clocking architecture, for transports to sound different?

As I am sure you recall, while AS was active on this forum, many weeks went by while many attempted and none succeeded. So are we now to assume that simply because he has vanished, that he must have been wrong?

Sorry man, that just does not feel right to me.

Hook

iust don´t understand why the serve can´t have an simple antenna for internet connection ? why do i need to lay an extra cable when i buy the serve with hard discs on bord ?.

i thought to pare it with the ndx, but still then i need this cable extra in the room. it´s putting me so off.

can the ndx conect to the serve and give all the information since it can receive internet over antenna?

why is everything so complicated ?

quote:

Originally posted by ghook2020:

So why do you think it is that nobody has ever presented a competent technical explanation about how it is possible, given the DAC's buffering/re-clocking architecture, for transports to sound different?

As I am sure you recall, while AS was active on this forum, many weeks went by while many attempted and none succeeded. So are we now to assume that simply because he has vanished, that he must have been wrong?

Sorry man, that just does not feel right to me.

Hook

Hook,

I think you are too kind. A quick look around here and I can see threads where people describe differences between CDX2 and CD5X feeding an nDAC, let alone all the computer variations.

Andy made some quite assertive claims that fly in the face of what most people have found (even HiFace on a Mac v Optical) was clearly different when I heard it. Many others have quoted different examples.

A comparison was offered by one of the other members here - since when Andy has disappeared.

What do you make of it? I have not heard a DAC other than the Benchmark that seemed to equalise sources to any great extent, the downside with the DAC1 was that I didn't actually enjoy it. I have often wondered if it is some processing necessary to minimise jitter from poor sources, that makes for the rather sterile sound that I thought the DAC1 had. YMMV etc.

Joe

quote:

Hi AMA -

Thanks for clarifying your comment.

So why do you think it is that nobody has ever presented a competent technical explanation about how it is possible, given the DAC's buffering/re-clocking architecture, for transports to sound different?

As I am sure you recall, while AS was active on this forum, many weeks went by while many attempted and none succeeded. So are we now to assume that simply because he has vanished, that he must have been wrong?

Sorry man, that just does not feel right to me.

Hook

Hi Hook, there is a very simple explanation on why does it happen.
It's been circulating around on the forum and you can find the same on popular manufacturers web-sites.
We had big debates with Andy on this. I proposed my explanation of this phenomenon which Andy found to be wrong and I agreed on that after consulting with my company local engineer (but not a DIGITAL engineer). But my ears were still proving that nDAC is sensitive to transports and I continued searching. I promised Andy to check the same with my company DIGITAL engineers which locate in the other country and I did it this summer when visiting their office.

It turned out that explanation was on surface and already published in various places. It's all about quality of bitstream waveform. It's not a perfect square-pulse sequence. In fact -- it's far from it. The clock aside the transport's Power Supply produces a huge impact on the bitstream waveform and can distort it significantly. This leads to misreading the income bits by S/PDIF receiver and hence can not be recovered by re-clocking. Unfortunately Andy has disappeared and I failed to share my findings with him through the forum. Since that time so many forum members reported the differences between transports with nDAC that I believe the new technical insight is redundant. Anyway I would like him to come back to the forum -- he helped me a lot in my investigation.

You have a very resolving system based on 252 and nDAC/XPS. I suspect your digital transport is not up to the best. Try to loan HDX or CDX2-2 for home audition and you will possibly hear the difference. It's not huge but I would say in the high-end world many people pay thousands of $$$ for a lesser gap in performance.

IMO, It has to do with bit retrieval. What's left will be jitter free but not necessarily exactly the same. End result is clearly better and perhaps technically close to perfect but no dig in is completely noise free or a pure square. Reclocking makes them so but you could be sampling points of the waveform that may be above or below a flat because the switch point is not as verticle as it could be due to limited bandwidth. Time and noise errors are also sampled. Jitter free does not mean free lunch. How you sample and interpolte can minimize artifacts to a point of technically being in the noise but I doubt a less than stellar stream can ever be interpolated well enough to be inaudible in a top Naim setup with a DAC that is not puposely sweetening the pot, even with Naim's very powerful 40 bit processing. I'm sure it's a bit of a conundrum for them. Do we make it nice and still very detailed or do we get what we feel is the most correct actual info and boggie while letting folks hear their source? I'm glad they went with the boogie.

This scenario holds up for any interface or system including I2S. It's the details that make the difference.

There should be a point where sources do sound alike or at least equal and with the NDAC config, it should be at a comparably low threshold. Problem is that it's informative enough to hear it. That you can here difs is not a failing. If they paid less attentuion to the clock for instance, some info would get lost with the wash while it could still sound very clean. Not a good trade.

This a complete laymans interptetation so the techies can feel free to lay in. I wont debate as I know the math says it should be better than this. I could be very wrong about my postulations but I'm just trying to rationalize what I hear.

quote:

Originally posted by AMA:
[QUOTE...
It turned out that explanation was on surface and already published in various places. It's all about quality of bitstream waveform. It's not a perfect square-pulse sequence. In fact -- it's far from it. The clock aside the transport's Power Supply produces a huge impact on the bitstream waveform and can distort it significantly. This leads to misreading the income bits by S/PDIF receiver and hence can not be recovered by re-clocking. Unfortunately Andy has disappeared and I failed to share my findings with him through the forum. Since that time so many forum members reported the differences between transports with nDAC that I believe the new technical insight is redundant. Anyway I would like him to come back to the forum -- he helped me a lot in my investigation....

On Sunday, Sun 20 June 2010 13:30, js posted a number of waveform pictures and stated "A dirty source waveform is still a dirty source waveform. PSs do effect the signal itself in the same way as analog whether you think it matters or not."

AS replied "<sigh> and getting digits from the transport to the DAC is a lossless process no matter how much ringing/distortion appears on the waveform (as long as it's within spec). And don't forget the DAC is isolated from any spuriae in the signal by design (separate circuits, electrically isolated). Before anyone suggests it could affect the DAC stages, then so could the two ruddy great DSPs that are sitting next to the input. They will have a far higher influence compared to any distortions of the waveform."

If I read you correctly, you are saying that S/PDIF receivers are capable of mis-reading bits, even when they are within spec?

Everything I have ever read about S/PDIF said it delivers bit perfect (and that the only thing that S/PDIF does not guarantee is that data arrives at precise points in time). That was AS's position as well.

Hook

quote:

Originally posted by js:...I wont debate as I know the math says it should be better than this. I could be very wrong about my postulations but I'm just trying to rationalize what I hear.

Hi JS -

As you may recall, this discussion did take place. It ended with AS's explanation about how the DSP's loss of precision is well below the noise floor of the DAC chip.

Hook

quote:

Originally posted by Joe Bibb:
...
What do you make of it?
Joe

Hi Joe -

Everything starts with the assumption that transports can deliver bit perfect. If they can't, then differences are easily explained. But if they can, it gets more complicated.

The fact is that AS went unchallenged in his explanation that S/PDIF delivers bit perfect (and that the only thing that S/PDIF does not guarantee is that data arrives at precise points in time). Nobody could counter his argument that up to, but not including the DAC chip itself, that this was a digital system, and no different than any other digital system. And before anyone calls me wrong, I would suggest they re-read the old threads. I did (big yawn), and would really like to avoid any more copying/pasting the old discussions. But they are still there if anyone is still interested.

AS's rebuttals left any and all explanations about how bit perfect transports could sound different in the realm of the environment, and not in that of the transports themselves. To date, it is very hard to understand how power noise and/or RFI can somehow leap over a digital cable (much less a Toslink) to effect the DAC's analog stage. Note also that AS said multiple times that anything goes once we get to the analog side, and that anything (cables, power cords, whatever) can make a difference.

Once again, and just for emphasis, I am not saying that anyone's experience is invalid. And I am really not trying to rehash this argument. I sure do not want to take AS's place as forum lightening rod. But since you asked what I think, I will answer by saying that I am left with lingering doubts on this subject. Also, it bugged me that during all of those weeks while this topic was debated, that nobody from Naim HQ said a word. It's not like they are silent on other threads, eh?

I am left with thinking that there is a real difference between subjective and objective explanations. The other night I went to sleep thinking the bass in my system had gone all wrong. Woke up the next morning to a head cold. That previous night's experience was very real to me, but I changed my mind in the morning based on new, objective (and snotty) evidence.

I get it that nobody finds this discussion "fun". And it pisses some people off, and makes them feel like their opinions are not being respected. Not my intent at all. I like reading/posting here, and do not want to develop a bad rep. My guess is that AS left because he got tired of the name calling, whether deserved or not. My goal right now is to not follow him out the door...so I will shut my big freaking mouth and move on.

Peace.

Hook

quote:

Originally posted by AMA:
... It's all about quality of bitstream waveform. It's not a perfect square-pulse sequence. In fact -- it's far from it. The clock aside the transport's Power Supply produces a huge impact on the bitstream waveform and can distort it significantly. This leads to misreading the income bits by S/PDIF receiver and hence can not be recovered by re-clocking. ..

Though I'm not a digital engineer (and not even an electrical one), I understand that the data in spdif is encoded as a biphase mark code:
 

quote:

The digital signal is coded using the 'biphase-mark-code' (BMC), which is a kind of phase-modulation. In this system, two zero-crossings of the signal mean a logical 1 and one zero-crossing means a logical 0.
  

Reading this I can imagine Andy's position that an spdif signal is very robust and not very prone to data errors. Because it would require a full phase shift at a sampling point for a bit to change from 0 to 1 or v.v. and that is quite a signal distortion not a minor one. 

Furthermore when it should come to a data error than 
 

quote:

 2. Errors
This usually causes very significant changes in the sound, often loud popping noises but occasionally less offensive effects. Any data loss or errors in either are a sign of a very broken link which is probably intermittently dropping out altogether.

 

In this case you are talking about serious audible defects, not subtle sound quality issues.

So I doubt that it is errors in reading spdif datastream compared to the original datastream that is causing sound quality differences.

-
aleg

quote:

Originally posted by AMA:
...
You have a very resolving system based on 252 and nDAC/XPS. I suspect your digital transport is not up to the best. Try to loan HDX or CDX2-2 for home audition and you will possibly hear the difference. It's not huge but I would say in the high-end world many people pay thousands of $$$ for a lesser gap in performance.

Certainly possible. JS told me a while back that I was doing as good as possible when it came to using a PC server as source. The RME 9632 card has gotten nothing but great reviews. Mine is BNC connected to the DAC.

May do this one day just for kicks, but it is not a huge priority for me. In fact, my only complaint about the DAC/XPS2 is that it does not sound as good as my turntable. But whose does?

Hook

quote:

Originally posted by ghook2020:
My goal right now is to not follow him out the door...so I will shut my big freaking mouth and move on.

Peace.

Hook

Great post Hook, love it.

I was looking forward to sitting in on the comparison - so I feel robbed of some infotainment.

Regards,

Joe

quote:

So why do you think it is that nobody has ever presented a competent technical explanation about how it is possible

Because no one can account for it, and no one really cares. It is typical flat earth theory, let your ears do the talking. Anyway, Naim representatives do talk about sources having an effect. Regardless, there was consensus...

1) Many people hear differences.
2) Andy knows a lot of about digital whatnot.
3) Their experiences could not be correlated with Andy's interpretation of the science.
4) Others could not hear differences between transports.

The are many potential explanations for 1) and 4). And everything still points to the age old adage, "Listen for yourself".

How could there be any definitive statement from anyone regarding what "sounds better". Come on.

I would however put lots of money down that the Serve/HDX/Qute owners will say the transport matters tremendously. HTPC users and Mac-to-DAC will say it doesn't. I would also bet there is a very direct correlation between the strengths of these beliefs and the value of the transport and overall budgets.

Give the guy with the Hiface a FREE Serve and ask see if he could take or leave it regarding SQ.

-Patrick

quote:

Originally posted by jon_jh:
... an HDX (in a cupboard) was used as a server for the NDX, and this fed into the DAC. ...

can someone explain me what the added value of the NDX in this configuration regarding SQ is? what is the benefit instead of feeding the DAC directly digital from the HDX

jerry

quote:

Originally posted by Joe Bibb:
[QUOTE]Originally posted by ghook2020:
My goal right now is to not follow him out the door...so I will shut my big freaking mouth and move on.

Peace.

Hook

I was waiting for an explaination of what what was heard.

quote:

If I read you correctly, you are saying that S/PDIF receivers are capable of mis-reading bits, even when they are within spec?

Yes

quote:

Everything I have ever read about S/PDIF said it delivers bit perfect

No

quote:

That was AS's position as well.

Yes

quote:

that nobody from Naim HQ said a word

Why should they? Do you think I shall disqualify my ears just because Naim HQ says their DAC reclocks everything? Whatever they say I shall still trust my ears. And my ears say nDAC is still sensitive to transports.

The bitstream waveform is obviously not square-shape. See the picture below.



But it's not a big issue as the S/PDIF receivers are smart enough to evaluate the signal front and qualify the bits properly (that was a strong point in Andy's explanations and he was right about this). Unless .... the saw-shape waveform is contaminated by noise and the front slope is blurred so that S/PDIF receiver is not smart enough to pick up the bit change and ... set up the 0 bit by default. Obviously whatever S/PDIF generator received will be flawlessly translated further along the chain.

BTW the transport can mess the bits on its side as well (when reading the data from CD or when sending the data through S/PDIF with poor clocking and noisy PS) -- and Andy agreed on this. The whole bout was around bit-perfect transport (say HDD streamer with high quality S/PDIF) into nDAC which can still rise the errors.

Hook, how do you explain that forum members are very certain about the difference between INT202, HDX, nServe, Qute, HiFace, Evo? Some forum members are raving around that different PS (!!!) affect the Evo output (which run into nDAC) and batteries sound better. Shall we admit they are all deaf?

I don't think Hook is debating what we hear. Just wants to know why.

As for batteries, in this case, I suspect it's the lesser of 2 evils and the right linear supply will top it.

quote:

Originally posted by pcstockton:
[QUOTE]...
Because no one can account for it, and no one really cares....
-Patrick

Patrick -

From the heart: buy whatever you want and enjoy your purchases. Life is too short to do anything else.

I love AMA and all he contributes to this forum, but I couldn't resist the temptation to throw out a friendly challenge to his comment that "serious trials" have "proven" the objective thinkers wrong on this issue.

No offense intended, so in the immortal word of Stoik...

Bye.

Hook

I think I forgot how to place pics.
I cut off the pic with ideal and real waveform from famous Stereophile article "Bits is bits".
But it does not go into the post (60 K, jpg). Why?

quote:

As for batteries, in this case, I suspect it's the lesser of 2 evils and the right linear supply will top it.

I have the same feeling. I was very surprised to hear what Dr. Jazz reported. But ferenc's case is different -- he uses a car battery which can supply a lot of energy and very fast. Still dedicated PS could potentially do the job better.