Why is the nDAC so cheap?

quote:

Originally posted by Andy S:

quote:

Originally posted by Joe Bibb:
Best tried by people for themselves, I think. I have iTunes, Pure Music and Amarra and there are definite sonic differences using the same file. All produce bit perfect output, so the story doesn't end with that for me.

Joe

How do you know they produce bit perfect output - have you captured the output streams and done a bit-for-bit comparison? Digits is digits at the software player level - there is no magic - this I'm certain of

Just 'cos you have 44.1kHz out of the back of the PC does NOT mean you are guaranteed bit perfect output - the player could have messed with the 16 bit word before it is output.

Hi Andy, Yes, Pure Music and iTunes are both bit perfect. I haven't confirmed Amarra since uninstalling, but I'll assume it's the same.

Peter

quote:

But I can hear differences between different cables hooking up the DAC to the 52 - immediate and not to my ears subtle

Hey, Andy, maybe because you 'buy' the idea of post-nDAC cables affecting sound quality your placebo effect is functioning. The placebo effect is pretty strong you know.

You raise a fair point about bit perfect. I know that my HiFace/DC1 sounds better than my Optichord cable. I think both are bit perfect because everyone says they are with a mac (here and computeraudiophile) but I don't know this. How can I investigate this? What do you do to prove a bit perfect transfer?

Gav.

quote:

Originally posted by Andy S:

I'll quote again from the White Paper:

quote:

What is jitter exactly?
Jitter is variations in the time separation of digital audio samples. All S/PDIF induced jitter coming in to the Naim DAC is removed, except for the encoded jitter caused by the mastering analogue to digital conversion done in the studio where the CD was mastered.

All jitter is removed. Period. Except that encoded at A->D conversion. Crikey, talk about being overly detailed

Andy

You assured me you now understand this. You’ve obviously lapsed into a state of misunderstanding.


ALL S/PDIF INDUCED JITTER COMING IN TO THE NAIM DAC IS REMOVED.

This does not mean, “all jitter period”.

quote:

Originally posted by Andy S:

Whilst I'm here, if anyone is finding that different software players sound different, the only way this could happen is if they aren't producing bit-perfect output.

Hmmm.

You’re treating the Naim dac white paper as gospel, yet treating the equivalent for the CDS2 as marketing bollox. Is there any reason apart from the fact it suits your argument to do so.


CD transports don’t output bits they output a waveform. The receiver extracts the bit data from the waveform by periodically measuring voltage. If the waveform entering the DAC contains noise, harmonics, dark matter or whatever, as long as the receiver can extract the voltage information, correct bit data will result. The point is the noise, harmonics, dark matter or whatever have still entered the DAC.

and the dark matter is really really black

quote:

Originally posted by fatcat:
You assured me you now understand this. You’ve obviously lapsed into a state of misunderstanding.


ALL S/PDIF INDUCED JITTER COMING IN TO THE NAIM DAC IS REMOVED.

This does not mean, “all jitter period”.

Sorry, you're missing the point. You can't tell where the jitter originated. You either remove it or you don't. Please, tell me where else the jitter comes from that you can't remove?

I'm interested to find out where in the audio replay chain the laws of physics are violated

No.. really... I am.....

quote:

Originally posted by fatcat:
CD transports don’t output bits they output a waveform. The receiver extracts the bit data from the waveform by periodically measuring voltage. If the waveform entering the DAC contains noise, harmonics, dark matter or whatever, as long as the receiver can extract the voltage information, correct bit data will result. The point is the noise, harmonics, dark matter or whatever have still entered the DAC.

Oh.. really.. I'm glad I've spent the last 15 years in my line of business totally misunderstanding how digital audio (and video) work. Thank you for putting me right.....

BTW, the receiver doesn't extract bits by periodically sampling the input waveform... it actually looks for transitions. It needs to do this to extract the clock back out of the SPDIF signal...

quote:

Originally posted by Nero:
and the dark matter is really really black

And makes up 90% of the universe....

Allegedly....

quote:

Originally posted by Andy S:
Please, tell me where else the jitter comes from that you can't remove?

Well, there is potential for adding jitter in everything that goes on from the buffer (from which the DAC's own clocks pull data) to the triggering networks in the DAC-chip. The question is whether it is correlated with the signal received over S/PDIF...

quote:

Originally posted by bhaagensen:
Well, there is potential for adding jitter in everything that goes on from the buffer (from which the DAC's own clocks pull data) to the triggering networks in the DAC-chip. The question is whether it is correlated with the signal received over S/PDIF...

OK.. so how would it be correlated? I have an input buffer that is clocked by one clock, and the output buffer clocked by a different uncorrelated (if you are to believe Naims white paper) clock. Other than the clock selector switching around (which you'd expect to cause a hiccup in the output) how does input jitter produce output jitter - which Naim claim to remove....

quote:

Originally posted by Andy S:
OK.. so how would it be correlated? I have an input buffer that is clocked by one clock, and the output buffer clocked by a different uncorrelated (if you are to believe Naims white paper) clock. Other than the clock selector switching around (which you'd expect to cause a hiccup in the output) how does input jitter produce output jitter - which Naim claim to remove....

Well, firstly I probably should have written 'dependent' rather than 'correlated'. Secondly, I don't really know. But the point is that, ultimately the exact time at which a 'bit' gets converted (on the chip), is determined by exactly when the resistors are triggered to let current flow or not (on the output side). Now, this is no longer only abstracted data values such as bits. These are real-valued, continuous, realtime, electrical signals. So are the energies that are stored in the (capasitor or transistor based) buffer represting the 'data'. Moreover, at each point in the system where 'data' is converted or moved, the fact is that the output bits are not only a function of the input bits. But rather that a continous, real-valued, realtime signal (representing a bit) depends on another one of the same kind. This is really one of the crusial observations that separate pure digital from what goes on in S/PDIF as well as internally in the DAC.

Now unless one can argue that in fact the electrical quantities involved at each stage are 100% independent of those present in the previous stage (starting from the S/PDIF receiver), there is a possibility that given slightly different input signals, the voltages ultimtely reaching the resistor triggers on the chip could differ slightly in time and/or space - thus affecting the output of the chip and ultimately the DAC itself.

Finally, instead of saying that input jitter produces output jitter, I'd rather say that different input signals could produce different output jitter. Thats the connection as I see it.

quote:

Originally posted by bhaagensen:
Now unless one can argue that in fact the electrical quantities involved at each stage are 100% independent of those present in the previous stage (starting from the S/PDIF receiver), there is a possibility that given slightly different input signals, the voltages ultimtely reaching the resistor triggers on the chip could differ slightly in time and/or space - thus affecting the output of the chip and ultimately the DAC itself.

But that's what they say they have done in the white paper.... Assuming you get the bits in correctly, the output buffer is clocked independently of the input buffer and the two sections of the circuit (input/DSP from DAC section) are electrically isolated via transformer/couplers. Can you think of a mechanism whereby the slightly different input signals (which, if we assume bit perfect transport & SPDIF can only be jitter) can affect the DAC stage? I'm at a loss to explain any method that could allow the input stream to affect the DAC stage from my understanding of the white paper.

For the interest of science I decided to do my own cursory comparison of different sources feeding the DAC.

My configuration:
nDAC+ 3rd Party PSU – NAC282+HiCap+NAPSC – NAP200 – Harbeth Monitor 30s

Sources tested:
1. USB Memory Stick
2. Macbook Pro (iTunes) – Hiface – 5M Gotham GAC-1 75 Ohm coaxial w/BNC X 2
3. Squeezebox Touch – QED 75 Ohm coaxial w/BNC & RCA

Music Used:
- Carol Kidd – When I dream 24/96 from HDTRACKS (ALAC for the digital clients and WAV for the USB)
- Rebecca Pidgeon – Spanish Harlem from HDTRACKS (ALAC for the digital clients and WAV for the USB)
- Used these two because their simplicity makes it easy to discern differences. I find orchestral music more difficult to tell apart

Provisos:
- I am no “golden ear”. Not good at telling between equipment if switched between A/B. I usually need long term listening of up to weeks to judge equipment.
- This is a personal choice of a layman and is nowhere near definitive
- My bias was that the USB stick should be best since it is the shortest chain and also played by the DAC player which eliminates possibility of grounding, jitter, electrical noise etc…
- I fully believe in the significance of placebo effect. So take everything I heard with a grain of salt because the differences if any were small.

Listening method – I listened to each song in full and went on to the next source. Order was USB/Macbook/Touch. After the test I switched back to USB and Macbook just to double check.

Result:
The differences were small. To my surprise, the Macbook sounded the best to me compared to both the USB and the Touch. Sounding more intimate, more “in your room” on the female voices. Other than that the three sounded very close, with the USB sounding better than the Touch. But overall, all three sounded very good and I could easily live with any of them.

Which begs the questions:
Surely we should at least treat Naim’s claim of jitter immunity as credible. (And from my own cursory test the differences between *my* sources are small.) ---- My questions are:

- Assuming that there are differences between sources. How big a difference would a Naim source make?
- Is it big enough to justify buying a perfectly good Naim CD player and trash its DAC section? That would be a big waste of a CDX2S which cost something close to US$6000, considering that its digital section should be very good for that price range.
- Since the CD5X and CDX2 *looks* like they have the same physical transport, it may make more sense to use a CD5X instead of the CDX2?

Hi JYOW,

Thanks for that - a worthwhile test I think - if only to get some of these things straight in your own mind. The interesting thing is the effects you describe of the differences in sound are not the ones usually associated with jitter (which tends to be a fuzziness in the higher frequencies and inability to place the sound spatially as accurately as a lower jitter source).

Do you have a cheap CD or DVD player with digital out available to try that in comparison? Obviously, you'd need to set it so that volume out was at maximum to avoid any mixing effects, but that too would be an interesting comparison IMHO. In my (short) tests with a cheap DVD player, I couldn't tell any difference between it and the ripped music file.

Dear all,

I too tried different sources on the nDAC and to answer JYOW´s question:

The use of the CD5XS is a perfect match for the nDAC.

I tried the Ipod = good, but not very engaging.

The Harman Kardon DVD 37 via DC1 = excellent value for money, but just not the foot-tapping spark sprang over...

WAV-files via diff. USB-sticks = way above both of the above. IMHO very very musical...fantastic sounding.

CD5XS = IMHO the best solution for at least my humble system. More resolution, more punch and fatness, but sometimes, with some records, I prefer the slightly "laid-back", more "musical" presentation of the USB-stick...

quote:

Originally posted by Andy S:
But that's what they say they have done in the white paper....

I guess you agree that the whitepaper, while somewhat detailed, remains closer to being a conceptual design paper, than a genuine schematic. It leaves many black holes which must be filled by anyone attempting to evaluate the claims put forth. Whence, this thread is now on its 13th page and accompanying many others debating the very same topic...

quote:

Originally posted by Andy S:
Assuming you get the bits in correctly, the output buffer is clocked independently of the input buffer and the two sections of the circuit (input/DSP from DAC section) are electrically isolated via transformer/couplers.

To my knowledge, even the best isolators are not perfect. AFAIR they even state in the whitepaper that the electrical separation between the digital and analouge side is not perfect. This is really the springing issue. In order to achieve the independence they claim, the signal received through S/PDIF must be 100% electrically isolated from the signal seen at the chip. Is this positively the case? AFAIK, the answer is no.

quote:

Originally posted by Andy S:
Can you think of a mechanism whereby the slightly different input signals (which, if we assume bit perfect transport & SPDIF can only be jitter) can affect the DAC stage?

If you are asking for the technicalities of such a mechanism, the answer is that I can not. On the other hand, can you think of an electrical design that would theoretically isolate the components in the sense that I wrote in the previous paragraph.

Btw. while the bit perfect assumption (statistically) holds, the rest of the assumption does not hold - and I am moreover convinced it is too strong. In this context one can not speak of bits and jitter alone. Any considiration must, unless a waterproof electrical isolation property holds, include that fact that the bits are not '0' and '1', but continous, real-valued, realtime signals. Slighly different input signals then obviously means much more than 'jitter'.)

quote:

Originally posted by Andy S:
I'm at a loss to explain any method that could allow the input stream to affect the DAC stage from my understanding of the white paper.

As is evident I am, in contrast to you, at loss of any explanation that could prove that the DAC is jitter-immune in the sense that you (and I too) interpret the writings in the whitepaper.

Added: Thanks to those who contribute with their time and effort in testing and reporting the effects observed by using various sources with the DAC. I posed a related request for experiences in the hi-fi part of the forum - so it is indeed interesting for me to read.

quote:

Originally posted by Andy S:
....from my understanding of the white paper.

I think that be the crux of the biscuit right there.

Tons of people are hearing differences. You are simply regurgitating ad nauseum, the white paper.

They OBVIOUSLY are not jiving with each other.

The only way I can reconcile the differences is to assume you are NOT understanding the white paper.

Since you are the ONLY only one (save maybe a few others) who are hung up on this, I suggest you either leave it alone, or start emailing with Naim HQ for more clarification of the White Paper which is clearly either wrong or easily misunderstood.

Since you are COMPLETELY unwilling to address the problem from a constructive point of view, nor are willing to address my specific reasons aside from jitter for the SQ deltas, I suggest you take it up with Naim directly.

-patrick

pcstockton: I don't know the backgrounds of your post. But AFAIC the topic is interesting.

(If you think otherwise, why not simply abstain from reading it?)

quote:

Originally posted by bhaagensen:
Btw. while the bit perfect assumption (statistically) holds, the rest of the assumption does not hold - and I am moreover convinced it is too strong. In this context one can not speak of bits and jitter alone. Any considiration must, unless a waterproof electrical isolation property holds, include that fact that the bits are not '0' and '1', but continous, real-valued, realtime signals. Slighly different input signals then obviously means much more than 'jitter'.)

Hmm.. no, I disagree with this. Whilst I agree that getting 1s and 0s off a CD is an analogue process, once it is in the digital domain, it is digits until you convert it back to analogue. As to isolation, a single bitstream - especially if connected via optical - has minimal impact to the digital noise that could be present where a DSP exists. Lets put it this way, if the SPDIF input causes 2% degradation of sound then the associated DSP must account for 32%-64% degradation (depending on if it is being used at 16 or 32 bits wide memory interface). That is, the SPDIF signal is lost in the noise. Let me ask it another way around. If the transport is connected via optical how can it electrically influence the nDAC?

Whilst I wouldn't like to misquote AMA on this, but he's just done some testing that, to him, shows that the nDAC is immune to cable differences. Now he is able to tell differences between sources... My quandry is this... If different cables do sound different on other DACs the only reason I can think of this is that they add to a higher or lesser degree to the jitter in the system (especially if optical is used). This can be the only real reason they sound different. Now, if the nDAC is cable independent, are we not saying that actually the nDAC does remove jitter from the cable, and since we can't distinguish between cable and source jitter it renders all sources the same.

I know people hear things differently, but there are enough people starting to say they can't hear differences that make we wonder if it really is a mass placebo effect... Yes, I know that isn't a popular position around here, but I say what I find...

quote:

Originally posted by pcstockton:
Since you are COMPLETELY unwilling to address the problem from a constructive point of view, nor are willing to address my specific reasons aside from jitter for the SQ deltas, I suggest you take it up with Naim directly.

-patrick

Sorry Patrick,

Ask away and I'll explain where I'm coming from. I don't recollect leaving any argument unanswered - other than "it just does" which doesn't wash with me...

BTW, the more I look into this and as other people report stuff back, the more I think I am understanding the white paper correctly...

Andy,

You know you have either dismissed or outright ignored my various, and quite plausible, reasons for sources sounding different.

I am not about to go back through the pages and find them again.

I listed them more than once, and never did you address them.

You only once said that the my reasoning could be addressed by using a toslink, or the ground switch.

I dont even know why I came back into this thread, especially consdiering I am going to use the Naim DAC with one source only for a long time.

This discussion is painful.

Out, really this time.
Patrick

quote:

Now, if the nDAC is cable independent, are we not saying that actually the nDAC does remove jitter from the cable, and since we can't distinguish between cable and source jitter it renders all sources the same.

Transports induce much more jitter into the bitstream than cables.
If nDAC can reject all jitter-related contamination from cables this does not mean it can do the same clean job on any transport.

I will keep an eye on this

quote:

Originally posted by bhaagensen:
pcstockton: I don't know the backgrounds of your post. But AFAIC the topic is interesting.

(If you think otherwise, why not simply abstain from reading it?)

Really? OK.

13 pages of getting nowhere? Yeah thats a blast.

Enjoy.

From Naim's 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*."

This to me sounds like all jitter introduced by the transports becomes a non-factor.

But in the white paper, it went on to state:

"How to overcome the problem of S/PDIF noise entering a DAC
The Naim DAC’s high-speed DSP (digital signal processor) front-end is electrically isolated from its high-resolution DAC and analogue circuits. Also, the two sections are run from separate power supplies. Together these measures *significantly reduce* the digtal RF noise which could affect the analogue stage."

To me, this means that there could still be differences between transports in how much RF noise they introduce to the DAC. But that, according to Naim, is also "significantly reduced".

I guess the ideal transport/DAC interface would be something similar to the "memory player" concept. I can picture a DAC that has a huge buffer, reads the whole track and shut down the electrical link to the transport and start.

But does not the USB stick accomplish that task?

For my own personal assurance only, in my comparison I preferred the Macbook Pro/Hiface to the USB playback, I guess whatever deficiency of the S/PDIF transport does not really bother me that much.

I guess I won't lose any sleep over not having a mega transport. Being a tin ear does have its advantage.

quote:

Originally posted by AMA:
Transports induce much more jitter into the bitstream than cables.
If nDAC can reject all jitter-related contamination from cables this does not mean it can do the same clean job on any transport.

Hi AMA,

Sorry, but yes it does (within the definition of a "reasonable" transport). Jitter is about presenting the same information at slightly different times. We are talking about the same clock (on average) just presented either slightly early or slightly late. Higher jitter transports (lets move them from 200ps -> 1000ps) just have a little more early and a little more late... The nDAC rejects this level of jitter - period (as you've found out with your cable trials).

I think people are missing the fact that jitter is both quicker and slower... If a clock is at 44.1kHz minus xHz, and it is ALWAYS at that frequency, it means it has zero jitter. A jittered clock is 44.1kHz+/-xHz. The average clock will be at 44.1kHz but can be up to 44.1+xHz and as slow as 44.1-xHz... It can change on a cycle by cycle basis..