NOS DAC sound more natural

[VenturaRV]

I tried to understand about this topic you mentioned. Is the problem happens when the sine wave generation started (the gray area on attached pic)?

 

[ATTACH=CONFIG]33472[/ATTACH]

No by no means.

I just tried to show a contradiction: if computer generated square waves and single transient impulses were not acceptable in DACs evaluation, why equally computer generated sine waves would be "legitime" for the same purpose.

I understand that a square wave has, in theory, an infinite odd order harmonic content, but the amplitude of each harmonic decreases 1/nf, being "n" the order (3, 5, 7,...) and "f" the fundamental frequency, so, in a digital system the highger order harmonics rapidly reach the LSB level, loosing any consequence. In the case of a pure sine wave, ideally, there are not harmonics at all, BUT as in the example you show, at the beguinning of the wave the level increases very fast (almost as a transient: the system cannot and must not make "predictions") and the unstability saw in the graphic is some sort of pre-ringing.

My point was (and is): what is the proper method not subjective to. evaluate a DAC performance? Test CDs? No, if we do not accept them 'cause their signals have not taken into account the ADC conversion method. A wave generator passing first throw the ADC converter? No, because we will never know WHAT actually we are measuring (the only way to know with this method how the DAC acts will be to know how the ADC acts in the first place, but for this purpose we need a DAC, so we'll be moving into a circle).

 

VenturaRV

[VenturaRV]

Interpolation, upsampling, upscaling, they're all the same thing. It doesn't matter whether the axis is time (audio) or space (images).

"Interpolation, upsampling, upscaling are all the same thing"... Three errors in one single stroke, and all three items implies quantification and, consequentely, their noise and errors.

By the way: sound evolves in "time axis", true; but moving images evolves in, at least, THREE axis: two spatial (high. and wide) and one in time.

As far as I know, Shannon theorem does apply to periodic functions, has nothing to do with digitazion (sampling is completely reversible, quantization is not), and cannot be applied to imagen.

 

VenturaRV

 

 

VenturaRV

[VenturaRV]

For the love of God, stop arguing. Please.

 

I know I am a new member and perhaps should just keep my own counsel, but I can't help but feel that you guys are losing sight of the big picture while arguing over these very technical issues. It's about our music and the enjoyment of it ultimately; please don't lose sight of that.

Welcome, George.

Don't panic.

You'll must see how things goes in the topic on cables (specially if they're digital and USB type).

Just take our arguiinng with a grain or two of salt.

Enjoy your love for music and sound (as all of us do, despite the appearances) and give your opinion or your impressions any time you like.

Cheers.

 

VenturaRV

[jhwalker]

For the love of God, stop arguing. Please.

 

I know I am a new member and perhaps should just keep my own counsel, but I can't help but feel that you guys are losing sight of the big picture while arguing over these very technical issues. It's about our music and the enjoyment of it ultimately; please don't lose sight of that.

 

The name of the site is "Computer Audiophile". Of necessity, the topics will often involve very technical discussions and arguments. IOW, it's not *all* about the music.

[Jud]

For the love of God, stop arguing. Please.

 

I know I am a new member and perhaps should just keep my own counsel, but I can't help but feel that you guys are losing sight of the big picture while arguing over these very technical issues. It's about our music and the enjoyment of it ultimately; please don't lose sight of that.

 

Hi George. There is actually no argument. Ventura's posts may give you the impression things are very technical, but the reason you can't understand them is because they are meaningless word salad. If there's anything you would like to understand better, please ask; there are plenty of people here who are well able to explain. And if there's nothing you want to understand better, there are lots and lots of discussions of subjective impressions and music, or you can start a thread yourself.

 

 

Sent from my iPhone using Computer Audiophile

[VenturaRV]

For the love of God, stop arguing. Please.

 

I know I am a new member and perhaps should just keep my own counsel, but I can't help but feel that you guys are losing sight of the big picture while arguing over these very technical issues. It's about our music and the enjoyment of it ultimately; please don't lose sight of that.

Hi again, George.

Please do not be mislead by Jud's comment about me. It comes from a personal disencounter of small or no consequence at all.

Effectively, here are a good bunch of folks that have a lot of technical speech and knowledge (more than me), but it is not neccesary a bad thing, but normal in a site like this, as it would be in, say, a forum dedicated to cars.

Of course you can ask anything you want to know or understand better, and many will help you, included Jud, whose opinions, despite his pathetic obsession against me, are usually ponderated and well informed (no sarcasm in this). Ask and comment, and enjoy a little bit of healthy discussion on a subject that obviously interest you.

Cheers! [emoji4]

 

VenturaRV

[VenturaRV]

The name of the site is "Computer Audiophile". Of necessity, the topics will often involve very technical discussions and arguments. IOW, it's not *all* about the music.

The name of the site is "Computer Audiophile"..., yes.

The topics will often involve very technical discussions and arguments..., yes.

It's not "all" about music..., yes and no.

Music is just sound organized following cultural history, tradition and so forth. Because of this, it is logic to suppose an audiophile is, prior to it, a music lover. And a music lover that wants to hear at home the music (the organized sound) as faithful to reality as possible, the only "reason d'ètre" of High Fidelity (the second word is very important).

If someone (no matter if a newcomer) remember us that maybe we're putting out of focus the first and last goal of our debates, perhaps he or she worths a little bit more attention from our part... and a small post of welcoming.

 

 

VenturaRV

[One and a half]

Which NOS DAC plays DSD128++?

[Jud]

Which NOS DAC plays DSD128++?

 

 

My impression is there are a number of DACs that won't further convert the signal prior to the final analog filter (the final conversion from digital to analog output) if they are fed an input at DSD128, -256, or -512.

 

 

How many DACs advertised as NOS or capable of being switched to an NOS mode are able to receive input at DSD128 or above I'm not certain.

 

 

 

Sent from my iPhone using Computer Audiophile

[lmitche]

Jud,

 

Do you mean so called discrete DSD DACs that use a delay line with shift registers sent to weighted groups of switches? This is the design used in the T+A DAC 8 DSD and the TI DSD1793 chips in the IFI microIDSD and IDAC2 products to name a few.

[Jud]

Jud,

 

Do you mean so called discrete DSD DACs that use a delay line with shift registers sent to weighted groups of switches? This is the design used in the T+A DAC 8 DSD and the TI DSD1793 chips in the IFI microIDSD and IDAC2 products to name a few.

 

Isn't there also a Lampi DAC that's supposed to do "NOS DSD"?

[lmitche]

Isn't there also a Lampi DAC that's supposed to do "NOS DSD"?

Yes, I have heard that as well about several Lampizator DACs, GoldenGate and Atlantic come to mind.

[buonassi]

On 8/6/2015 at 9:51 AM, Jud said:

That said, I have done a lot of experimentation with oversampling filters in Audirvana Plus, and have set up some filters that *did* allow aliasing. These didn't sound bad to me.

 

Nor do they sound bad to me.  But that's me and I don't hear like others do, as you've also mentioned.

 

I think the argument in favor of NOS can be summarized:  The aliasing present due to not having a reconstruction filter in place is the lesser of two evils.  The timing anomalies created by using filters (ringing, no matter how little pre or post) is more detrimental to the musical enjoyment.  Any images mirrored into the ultra sonic region are of no consequence because our ears act as a natural filter.  Any intermodulation distortion injected into the audible region are, likewise, of no consequence because their amplitudes are sufficiently small and are masked by the music being played (approx 50db below the music). 

 

Those who argue in favor of OS believe just the opposite of the above and have objective measurements on their side.  Further, they believe that these ultrasonic images not only affect the audible band, but the ultrasonic energy can alter the physical response of their transducers/amplifiers etc.

 

The one thing I'd like to understand is the phase response of NOS.  I think I read that the phase is altered because the 1x passthrough process of D to A produces a very benign filter of its own; something like -3db per octave?  This is inherently a minimum phase filter, like analogue.  Therefore, the phase has to be altered somehow, right?  @Archimago modeled out an OS filter similar to NOS and it reeked havoc on the phase (section 5 of the link).  But I'm not sure if this is the same phase shift that would be observed with true NOS?

[Jud]

I don't know why you think an analogue filter must be minimum phase.  It's not the case.

 

You should consider that the IM distortion might indeed be audible, and might in fact be responsible for some people liking filters that allow it.

[opus101]

5 hours ago, buonassi said:

The timing anomalies created by using filters (ringing, no matter how little pre or post) is more detrimental to the musical enjoyment.  Any images mirrored into the ultra sonic region are of no consequence because our ears act as a natural filter.  Any intermodulation distortion injected into the audible region are, likewise, of no consequence because their amplitudes are sufficiently small and are masked by the music being played (approx 50db below the music). 

 

My own experiments with filtering (passive filtering, not opamp-based filtering) the output of multibit DACs with LC filters tend not to confirm these ideas. Filtered definitely sounds better (clearer) than unfiltered, and this is what theory would predict. The presence of images at higher than audio frequencies increases IMD in both the DAC's analog output stage and downstream pre/amp and hence increases the dynamic noise in the system.

[buonassi]

3 hours ago, Jud said:

I don't know why you think an analogue filter must be minimum phase.  It's not the case.

 

You should consider that the IM distortion might indeed be audible, and might in fact be responsible for some people liking filters that allow it.

 

Regarding the analogue filter, I had read that somewhere, I think on CA actually.  Can't link to it, however.  If I recall, the explanation was that analogue EQ has no delay, so must introduce a phase shift, unlike "group delay" of linear phase.  But I'm very open to learning facts.  It seems I may have either misunderstood what I read or just picked up on something that wasn't factual.

 

And for IMD audibility, I agree that this could very well be the case.

 

I'm not in either camp, yet (pro/against NOS).  But I have been playing around quite a bit with my iDAC2, which allows you to switch to NOS and linear phase slow rolloff oversampling on the fly.  The biggest difference I can hear is the top octave droop when in NOS.  I do here other subtleties, but this one is easiest to pick up on.  This can also be a reason why some prefer NOS, simply because of the gentler high treble response.  But my gut tells me it isn't that simple. I hear the biggest difference between the linear fast rolloff OS and NOS settings.  I'm not going to say that it's ringing, but those two settings are easiest to tell apart.  

 

 

 

 

[buonassi]

1 hour ago, opus101 said:

Filtered definitely sounds better (clearer) than unfiltered, and this is what theory would predict.

 

I also find that this is the case.  But I also think that while clearer, it sounds a bit more holographic and TOO perfect, if that makes sense.  It's hard to describe.  (Keep in mind I'm a headphone listener, whatever that's worth)  It's not better, not worse, just different.  I still can't decide which I like the best.  I find that I can relax more and not get fatigued as easily with NOS mode switched on.  

 

For the most part, I tend to like analytical sound and critical listening, but can only take so much of it at a time.  My other DACs (both Sabre) are so surgical and precise I can hear details much more easily - but it's also harder to "get into" the piece of music I'm listening to.  Every micro detail seems forced at me sometimes.   

[opus101]

18 minutes ago, buonassi said:

I also find that this is the case.  But I also think that while clearer, it sounds a bit more holographic and TOO perfect, if that makes sense.

 

What kind of filtering are you using? I agree with the holographic description, I'm at present mostly a headphone user. I'm not a fan of Sabre DACs - I have an ES9018 in my phone and it corrupts the higher frequencies too much for my taste, infusing them with noise. The newer Sabres are likely better in this regard though.

[opus101]

37 minutes ago, buonassi said:

Regarding the analogue filter, I had read that somewhere, I think on CA actually.  Can't link to it, however.  If I recall, the explanation was that analogue EQ has no delay, so must introduce a phase shift, unlike "group delay" of linear phase.  But I'm very open to learning facts.  It seems I may have either misunderstood what I read or just picked up on something that wasn't factual.

 

My understanding of this is that, while its possible for analog filters not to be minimum phase, they'd need to be specifically designed to be that way, through adding all-pass sections. An all-pass section is flat in terms of frequency response but adds group delay, meaning that the phase could not be determined from the FR which is what minimum phase normally means.

[buonassi]

21 minutes ago, opus101 said:

What kind of filtering are you using?

 

Well I have several DACs: ES9018K2M standard presets, fast/slow/min.   ES9028PRO standard presets, too many to list.   my burr brown DSD1793 standard presets fast/slow.  The burr brown doesn't sound as holographic, but it's not quite as detailed either - not as "forced".

 

I also play around quite a bit with the iZotope SRC in software and model out various reconstruction filters.

 

23 minutes ago, opus101 said:

My understanding of this is that, while its possible for analog filters not to be minimum phase, they'd need to be specifically designed to be that way, through adding all-pass sections.

 

Thx for explanation.  Can you confirm that the zero order hold inherent in delta sigma modulation does introduce a phase shift and is not a linear phase system?  If this is the case, I'm pretty sure I know why I've been enjoying the NOS setting. 

[opus101]

1 hour ago, buonassi said:

Well I have several DACs: ES9018K2M standard presets, fast/slow/min.   ES9028PRO standard presets, too many to list.   my burr brown DSD1793 standard presets fast/slow.

 

Ah then it sounds as though we're not talking about the same kinds of filters. These DACs are all containing digital filters right, for upsampling? Whereas I'm talking about post-DAC (reconstruction) filtering.

 

The ZOH question is an interesting one, I'd have to study more theory to be able to answer you.

[buonassi]

19 hours ago, opus101 said:

These DACs are all containing digital filters right, for upsampling? Whereas I'm talking about post-DAC (reconstruction) filtering.

 

I'm probably using the term incorrectly.  I should be saying "interpolation filter" for oversampling DACs or even software outside the DAC.  

 

Can you give me an example of a reconstruction filter post DAC?  What do you mean by this?  Analogue domain? 

[Miska]

Analog filter is always needed for reconstruction... Oversampling/upsampling digital filters are there to help the analog filters. By moving the images that the analog reconstruction filter is supposed to remove, apart from the  actual signal. Otherwise there is really not enough space for the analog filter's transition band.

 

If you want accurate reconstruction of 16-bit digital source, you need to have at least 96 dB attenuation by the first image frequency component (for RedBook that is at 22.05 kHz). If you want accurate reconstruction of 24-bit digital source you need to have 144 dB attenuation correspondingly...

 

[opus101]

11 hours ago, buonassi said:

Can you give me an example of a reconstruction filter post DAC?  What do you mean by this?  Analogue domain? 

 

By example do you mean a schematic for a filter? If so then there are various ways of going about it, either active (normally with opamps, in several possible configurations) or passive.

 

Yeah the reconstruction filter's going to be in the 'analog domain'.

[Jud]

Miska will hopefully correct me if I'm wrong: From a couple of sources I've read, the phase characteristic of a zero order hold filter is "linear, with jump discontinuities." I wonder whether the filter would be constructed so as to avoid the points of discontinuity?