NOS DAC sound more natural

[Solstice380]

But you would be wrong. I oversimplified what I wrote before. It upsamples everything below DSD64 rates first to DSD128 and then in a separate process to 50Mhz before sending to the filter. If you feed it DSD128 it goes directly to the filter. If you feed it DSD256 it undergoes just the conversion to 50Mhz and then to filter.

 

Thanks for clarifying your "oversimplification", or as it is really called, wrong information. Do you have a definitive reference for the clarifying info or just going from memory?

 

 

Sent from my iPad using Computer Audiophile

[manisandher]

Even the Phasure is not really NOS IMO, even if the OS is being done in software, there is still OS.

 

Yep, I agree that the Phasure NOS1a shouldn't be considered a NOS DAC, in the generally accepted sense, because it's been designed specifically to accept 16x fs rates. But technically, the filtering done in XXHighEnd isn't OS, as far as I know. There will still be imaging, but pushed so far out from audibility that it shouldn't matter. That's my understanding, but perhaps PeterSt might chime in at some point.

 

Mani.

[wisnon]

Not to pick on you, but the Golden Gate was mentioned previously as well. A dac with an FPGA is not NOS as I understand it. Maybe the OP should specify what they mean by NOS. Even the Phasure is not really NOS IMO, even if the OS is being done in software, there is still OS.

 

To my mind, NOS "sound" means NOS period- such as a TDA 15xx or PCM 1704 running straight redbook. Then again, I got in trouble for suggesting a single ended triode amp ought to contain a triode tube.

 

Lampi chipless DSD does not use an FPGA and it does not change the native rate of the bitstream input.

[Jud]

But technically, the filtering done in XXHighEnd isn't OS, as far as I know. There will still be imaging, but pushed so far out from audibility that it shouldn't matter. That's my understanding, but perhaps PeterSt might chime in at some point.

 

Mani.

 

 

If the rates that come out are higher than the rates that go in, it's interpolation (which avoids any of the nonsensical "upsampling" vs. "oversampling" marketing terminology arguments).

[4est]

Lampi chipless DSD does not use an FPGA and it does not change the native rate of the bitstream input.

 

Yes, but the PCM version does use an FPGA. Many running the DSD only variants are still likely converting PCM>DSD or DSD64/128>DSD256, and I do not see that as NOS either.

[Jud]

Proper ADC-side anti-aliasing should attain full attenuation at Fs/2, and should do so with a transition band wide enough so as not to make its own ringing audible (to newly-borns, that is). Given such a signal, the DAC filter will not ring itself, it will only pass on the ADC-side ringing. That is mathematical fact.

 

Do you have any sort of idea what portion of the RedBook and higher resolution music we listen to was treated in this way at the ADC end of things?

[manisandher]

If the rates that come out are higher than the rates that go in, it's interpolation (which avoids any of the nonsensical "upsampling" vs. "oversampling" marketing terminology arguments).

 

PeterSt referred to it as 'up-scaling' IIRC - in the sense that it's similar to what's done in video. But 'interpolation' works for me

 

Mani.

[mansr]

PeterSt referred to it as 'up-scaling' IIRC - in the sense that it's similar to what's done in video. But 'interpolation' works for me

 

Mani.

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

[smartin]

Yes, but the PCM version does use an FPGA. Many running the DSD only variants are still likely converting PCM>DSD or DSD64/128>DSD256, and I do not see that as NOS either.

 

For PCM, Lampi's use either a canned DSM DAC chip (not sure what, they don't say if it's AK, ESS, BB, or what) or an R2R (the GG uses an R2R array), none use an FPGA. For DSD playback Lampi uses a chipless module that passes the signal straight to the analog stage, no oversampling. If people with the DSD only version are converting PCM>DSD, or DSD64/?>DSD256/512, they are doing so in their computer, before it gets to the Lampi. The Lampi will play whatever DSD format it is given without oversampling.

[4est]

For PCM, Lampi's use either a canned DSM DAC chip (not sure what, they don't say if it's AK, ESS, BB, or what) or an R2R (the GG uses an R2R array), none use an FPGA. For DSD playback Lampi uses a chipless module that passes the signal straight to the analog stage, no oversampling. If people with the DSD only version are converting PCM>DSD, or DSD64/?>DSD256/512, they are doing so in their computer, before it gets to the Lampi. The Lampi will play whatever DSD format it is given without oversampling.

 

OK, and so the GG uses the Soekris that does not use an FGPA. My fault, I was thinking of the DIY modules.

 

The people such as myself that are doing the upsampling/conversions in the computer are still upsampling. I think it is misleading and confusing to present such as NOS even if the DAC is not doing the upsampling/interpolation/"upscaling". This was my only point and I do not want to derail this conversation further.

[VenturaRV]

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

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]

True. But these synthetic test signals are not band-limited, and viewed from the point of view of the entire signal chain (ADC-DAC) they are illegal.

 

What Yamamoto rightly said is that a filter rings only when it is hit by signal at the filter's cut-off frequency. For a DAC there should be no energy at this frequency, if the recording-side anti-aliasing filter has done its job properly.

 

Proper ADC-side anti-aliasing should attain full attenuation at Fs/2, and should do so with a transition band wide enough so as not to make its own ringing audible (to newly-borns, that is). Given such a signal, the DAC filter will not ring itself, it will only pass on the ADC-side ringing. That is mathematical fact.

 

And a NOS DAC also passes on the ADC-side ringing.

For the same reasons, we must assume as "illegal" the pure sine waves shown in the oscylloscope by oversampling DACs, since they are synthetic generated too.

From this approach, we only can admit as a proper result at the end of the ADC > DAC chain sine, square waves and pulses generated by a loudspeaker in front of a measurement calibrated mike, as terms of comparison.

Lampi chipless DSD does not use an FPGA and it does not change the native rate of the bitstream input.

 

 

VenturaRV

[Miska]

Do you have any sort of idea what portion of the RedBook and higher resolution music we listen to was treated in this way at the ADC end of things?

 

If you have RedBook content that was recorded at 44.1k, then practically all that content has gone through oversampling ADC and brick-wall filter. The rest that hasn't, has been recorded most typically at 96k sampling rate using oversampling ADC and then converted to 44.1k at mastering stage. For checking out those there's good database here SRC Comparisons . And the brickwall is needed because 44.1k sampling rate just doesn't leave space for doing antialiasing in any other way.

 

If you do recording and playback in DSD, the whole ringing stuff is not an issue, because you don't have decimation filters on the path and the Nyquist frequency of delta-sigma converters is so high, that first order analog antialiasing filter at ADC side is enough to avoid aliases. First order filters don't "ring"... And even if the editing has been done in DXD, the Nyquist of that is still high enough at 176.4k that there is very unlikely content in the source that would trigger the ringing.

 

So the whole ringing thing is only an issue for low rate PCM, and in practically all cases it is built into the content at ADC/mastering stage. It can be later changed with use of apodizing upsampling filters at playback time. But that's it. This is very straightforward and logical thing.

 

When such low rate PCM content is played through a NOS PCM DAC, it just plays the ringing as it is in the source content and produce lot of distortion as a side effect. So the point of avoiding ringing by using a NOS DAC is completely moot (it's a myth)...

[mansr]

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.

 

I suggest you expand your knowledge.

[Jud]

I suggest you expand your knowledge.

 

Why, it is a true statement - as far as he knows....

[Jud]

So the whole ringing thing is only an issue for low rate PCM, and in practically all cases it is built into the content at ADC/mastering stage. It can be later changed with use of apodizing upsampling filters at playback time. But that's it. This is very straightforward and logical thing.

 

When such low rate PCM content is played through a NOS PCM DAC, it just plays the ringing as it is in the source content and produce lot of distortion as a side effect. So the point of avoiding ringing by using a NOS DAC is completely moot (it's a myth)...

 

Which is how you can have "too much of a good(?) thing" where a gentler filtering slope is concerned - it must be gentle enough not to ring much itself, but have enough cut to remove the ringing that was there before.

[Jud]

Proper ADC-side anti-aliasing should attain full attenuation at Fs/2, and should do so with a transition band wide enough so as not to make its own ringing audible (to newly-borns, that is). Given such a signal, the DAC filter will not ring itself, it will only pass on the ADC-side ringing. That is mathematical fact.

 

 

If you have RedBook content that was recorded at 44.1k, then practically all that content has gone through oversampling ADC and brick-wall filter. The rest that hasn't, has been recorded most typically at 96k sampling rate using oversampling ADC and then converted to 44.1k at mastering stage. For checking out those there's good database here SRC Comparisons . And the brickwall is needed because 44.1k sampling rate just doesn't leave space for doing antialiasing in any other way.

 

So the whole ringing thing is only an issue for low rate PCM, and in practically all cases it is built into the content at ADC/mastering stage. It can be later changed with use of apodizing upsampling filters at playback time. But that's it. This is very straightforward and logical thing.

 

 

Does anyone have any notion what proportion of CDs might be considered "well recorded" in Fokus' sense, for classical and/or popular music?

[VenturaRV]

I suggest you expand your knowledge.

About Shannon theorem or about digital treatment of serial images (and it upscaling to 4k resolution)?

 

VenturaRV

[mansr]

About Shannon theorem or about digital treatment of serial images (and it upscaling to 4k resolution)?

 

VenturaRV

Both.

[mansr]

Why, it is a true statement - as far as he knows....

Indeed, but the statement is in fact false, so expanding the reach of his knowledge would be beneficial.

[VenturaRV]

Both.

I was sure that was gonna be your answer.

Thanks a lot. :-)

 

Sent from my G620S-L01 using Computer Audiophile mobile app

[VenturaRV]

Indeed, but the statement is in fact false, so expanding the reach of his knowledge would be beneficial.

The "fact" is you (as many) do not understand properly sampling theorem.

Sampling refers ONLY to samples of a signal that can take ANY REAL value before quantification. Quantification is a process that comes after sampling (and its consequences) has taken place. Now, the samples ONLY can take integer or rational values (the case of floating point), and always contains quantification noise, which is not a consequence of sampling process.

If you doubt it, please refer to this paper in Google and see if you can read in it the word "quantification" at any place (the paper contains the math explanation of the theorem):

[PDF] Sampling Theorem and its Importance.

You're welcome. :-)

 

 

Enviado desde mi G620S-L01 mediante Tapatalk

[yamamoto2002]

For the same reasons, we must assume as "illegal" the pure sine waves shown in the oscylloscope by oversampling DACs, since they are synthetic generated too.

 

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)?

 

[George Hincapie]

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.

[mansr]

The "fact" is you (as many) do not understand properly sampling theorem.

Sampling refers ONLY to samples of a signal that can take ANY REAL value before quantification. Quantification is a process that comes after sampling (and its consequences) has taken place. Now, the samples ONLY can take integer or rational values (the case of floating point), and always contains quantification noise, which is not a consequence of sampling process.

If you doubt it, please refer to this paper in Google and see if you can read in it the word "quantification" at any place (the paper contains the math explanation of the theorem):

[PDF] Sampling Theorem and its Importance.

You're welcome. :-)

 

 

Enviado desde mi G620S-L01 mediante Tapatalk

What I said has nothing to do with quantisation noise.