How can a non-oversampling DAC sound good?

[Miska]

I've said ANY model. This includes those converting PCM up to 24/192k, manufactured until 2010 or so.

 

All the Luxman ones I found measurements for are oversampling type with analog reconstruction filters... That goes at least back to 2005.

[Miska]

I did not find measurements, but Sakura speaks of the DA-07 as having "three taps." Does this not imply some type of filter?

 

Yes, it functions as a filter as well and of course such is not really a "NOS" DAC.

 

Although not even necessarily have "taps" like for example the polynomial interpolators in HQPlayer. They are not filters as such, but different spline-style interpolators. Those are still characterized by similar trade-offs as NOS, but to lesser extent.

 

Ayre DACs also have similar low-tap (although traditional FIR) slow-roll-off filters in "listen" mode. Some DAC chips also offer that kind of filter options, like Cirrus Logic in the CS4398 and AKM on some of their DAC chips. ESS also has slow-roll-off filter option.

 

I personally rather go for minimum-phase filters.

[VenturaRV]

Are you old enough to have tuned in television channels with a rotary antenna on a mast? With several channels in a nearby town, did you adjust the antenna to do the best job getting the weakest signal? That's what we always did. Why would we just focus on *one* channel of *many*?

 

 

Because it provided the thing we were looking for. If you adjusted the antenna using a channel that came in strong over a wide directional range, when you changed to a weaker channel the signal might not be so good. Tune in the weakest channel in a city and all channels in that city were as good or better.

 

 

When Miska wants to listen for particular aspects of the sound, he naturally listens to the instrument that best allows him to tell if that aspect is correct. I am supposing Miska does not always and forever listen only to one aspect of the sound, but would like all aspects to be correct, so he listens for other aspects at other times in other ways.

Still I don't see Miska's point.

I can, of course, focus my attention in a particular instrument while listening to music, provided such instrument has been clearly recorded, but this fact, as far as I can understand, does not have anything to do with the global SQ of the recording, less to say with its "naturalness".

Three points:

1) What's the point to focussing on a particular instrument in order to evaluate the quality of a WHOLE recording?

Second: Absolutely most recordings of "traditional classic music" from the early '60s to present day (I'm discounting "audiophile" ones, wich are a very small niche) are made using multi mike close up techniques (hence one of the reasons of its lack of naturalnes).

Third: Since when classical music has not enough transients content? What type of "traditional classic music" have you listen to? When you hear or play a piano, there are not transients in the attack of the keys? The percussion section of a symphonic orchestra, does not produces transients at all? Transients are present in ANY genre of music, not only in those that have a constant rythmic scheme. (By the way; well recorded jazz is a good genre to find or not naturalness in sound, because normally all musicians play together during the sessions, and with minimal if any post-prod manipulation)

 

 

VenturaRV

 

 

VenturaRV

[VenturaRV]

You cannot evaluate aything without previously knowing the quality of the signal.

 

VenturaRV

[VenturaRV]

I said: "(I'm discounting "audiophile" ones, wich are a very small niche)

 

VenturaRV

[VenturaRV]

I have just to go to the symphonic Auditorium, better to look at spectrum plots.

 

VenturaRV

[VenturaRV]

Well that depends on what piece is being performed and WHO is the performer.

 

VenturaRV

[VenturaRV]

Again, better to go to the Auditorium and forget dubious recordings.

 

VenturaRV

[VenturaRV]

All the Luxman ones I found measurements for are oversampling type with analog reconstruction filters... That goes at least back to 2005.

I posted, wrong, until 2010 or so. Last Luxman models (they called them."universal" because they could converse, in a normal way, DSD too, and PCM up to 24/192k in Fluency and Shannon modes) date from the beggining of the '00s.

 

VenturaRV

[Miska]

I posted, wrong, until 2010 or so. Last Luxman models (they called them."universal" because they could converse, in a normal way, DSD too, and PCM up to 24/192k in Fluency and Shannon modes) date from the beggining of the '00s.

 

The ones I've found and play DSD use TI/BB oversampling delta-sigma DAC chips. And AFAIK, both modes are certainly oversampling but just use different filters.

[Miska]

Still I don't see Miska's point.

I can, of course, focus my attention in a particular instrument while listening to music, provided such instrument has been clearly recorded, but this fact, as far as I can understand, does not have anything to do with the global SQ of the recording, less to say with its "naturalness".

 

If you want to evaluate quality of the hardware, you only need limited, carefully selected well known set of test material.

[Jud]

I'm refferring all the time to non-oversampling AND filterless converters. All in one.

 

 

 

VenturaRV

Please don't ever quote me out of context to make it seem as if I am agreeing with you again.

[VenturaRV]

Please don't ever quote me out of context to make it seem as if I am agreeing with you again.

Excuse me: if you think I have the less interest in quoting you or in giving the impression to someone thay you agree with me, you're certainly very wrong and assumes that I give you far more importance than I do.

Another thing: don't ever say me what I can or must do. Mostly because you'll waste your time and efforts.

 

VenturaRV

[lmitche]

Excuse me: if you think I have the less interest in quoting you or in giving the impression to someone thay you agree with me, you're certainly very wrong and assumes that I give you far more importance than I do.

Another thing: don't ever say me what I can or must do. Mostly because you'll waste your time and efforts.

 

VenturaRV

 

Ventura RV,

 

Jud could not have been more gracious in his simple request. Your response is rude and completely over the top. Please stop!

[Jud]

Excuse me: if you think I have the less interest in quoting you or in giving the impression to someone thay you agree with me, you're certainly very wrong

 

 

You must be very forgetful indeed. Post #123 in this thread five hours ago:

 

I'm refferring all the time to non-oversampling AND filterless converters. All in one.

Originally Posted by Jud

Yes, of course you are right. Thanks.

 

 

Sent from my iPhone using Computer Audiophile

 

 

You quoted me out of context to make it seem as if I was agreeing with you, when actually it was me agreeing with something else Miska said upthread. I have no idea why you would do something misleading like this, nor why you would deny it when it is so easily confirmed. Do you not know how to behave with courtesy?

 

 

Another thing: don't ever say me what I can or must do. Mostly because you'll waste your time and efforts.

 

VenturaRV

 

 

I did not tell. I asked, putting it in the form of a request and courteously saying "Please," though of course I was very unhappy with what you had tried to do. Are you not capable of responding to a polite request for correct behavior toward someone else? What you wish to say about equipment is your business, but when you try to make it appear as if I​ said something I did not, that is my business, and I once again politely request that you stop.

[VenturaRV]

Ventura RV,

 

Jud could not have been more gracious in his simple request. Your response is rude and completely over the top. Please stop!

Excuse me you, too.

I'm still wondering the cause of Jud's post you find so gracious.

If I feel (and read) someone is attacking me without any motive, that one cannot expect from me flowers and honey, that's for sure.

 

VenturaRV

[audiventory]

That's why the output spectrum of a 0 - 22.05 kHz sweep with a filterless NOS looks like this:

[ATTACH=CONFIG]32969[/ATTACH]

 

This is without any intermodulation products from the electronics in play yet. So the high frequency roll-off and images are there.

 

All for killing of dynamic range. Especially for a wideband system.

 

Let suggest, we have no ringing by digital filter more.

 

But, interestingly, what about own ringing of speakers?

 

I suppose, electromechanical oscillatory processes exists there.

[VenturaRV]

You must be very forgetful indeed. Post #123 in this thread five hours ago:

 

 

 

You quoted me out of context to make it seem as if I was agreeing with you, when actually it was me agreeing with something else Miska said upthread. I have no idea why you would do something misleading like this, nor why you would deny it when it is so easily confirmed. Do you not know how to behave with courtesy?

 

 

 

 

 

I did not tell. I asked, putting it in the form of a request and courteously saying "Please," though of course I was very unhappy with what you had tried to do. Are you not capable of responding to a polite request for correct behavior toward someone else? What you wish to say about equipment is your business, but when you try to make it appear as if I​ said something I did not, that is my business, and I once again politely request that you stop.

I told you: don't waste your time and efforts.

And that's all.

 

VenturaRV

[Superdad]

It is different thing if NOS DAC is used only at higher sampling rates like at least 352.8 kHz.

 

Of course that's all I feed my NOS DAC any more (since iZotope in A+, and for the past 2 years with HQ Player).

 

My favorite PCM SRC filter is Poly-sinc-short (not that much of a minimum phase fan).

 

Can you remind us of which HQP filters you consider to be "apodizing?" Are those just the ones that start cutting earlier? I always forget.

 

 

Sent from my iPad using Computer Audiophile

[VenturaRV]

There is also very common misconception that with a NOS DAC there wouldn't be ringing due to filters. However, the ringing is there already in the source data, because any ADC running at 44.1 kHz output needs to have steep anti-alias filters. And practically all ADC's used since late 80's are oversampling ones, employing fairly steep digital anti-alias filters to down-convert the sampling rate.

 

Also lot of source content these days is recorded, edited and mixed in hires, typically at least 96/24 and then the 44.1/16 conversion is done at final mastering stages. This is where similar anti-alias filtering happens in software.

 

This ringing can be reduced/modified using apodizing upsampling filters, which is what I prefer to use. Or it can be upsampled with a filter that preserves the original digital filter properties and doesn't touch the ringing.

 

Thus I completely fail to see point of a NOS DAC playing at such low sampling rate. It is different thing if NOS DAC is used only at higher sampling rates like at least 352.8 kHz.

You're probably right in your very last statement.

In a general sense, that will be the case for any type of PCM converter. The higher the sampling rate (and the larger the number of bits per sample), the lesser the artifacts will be found at the output.

But this approach leads us, forcing it to its last consequence, to old and well known analogue sound representation: infinite (theoretically) sampling rate and infinitely small (down to quantic scale) bits size and infinite number of them.

And that will be, again, a solution in search of a problem (the limits of analogue that digital tryed to avoid).

 

VenturaRV

[YashN]

For full picture need use sweep sine and learn time-spectrum diagramm in expanding different level ranges.

 

Otherwise we may miss distortions.

 

For interesting areas of the time-spectrum diagramm we can use 2-D spectrum.

 

REW definitely does the sine sweep. There's also a waterfall view and a spectrogram.

 

Most of my sub 20-30Hz issues were caused by... sub-optimal layout/referencing issues pre-regulator, can you believe it?

 

So these sub-bass rumbles were robbing a lot of energy too, once I fixed the layout, the chart changed drastically and the sound gained in 'solidity' across the board:

 

 

That means, I not only heard what a sub-optimal layout pre-regulator (there's a lot of noise going on there) does to SQ, but I also saw on an FFT chart how the issues look.

 

The layout issue was inadvertent. Hopefully this means I don't need to clone my custom connector and LRP but can work on advancing the LRP itself several levels above its current state and performance.

[YashN]

I am thinking of the paper linked to earlier in the thread. Is there a way to simulate the ear's response/filtering in software like LTSpice?

[Miska]

But this approach leads us, forcing it to its last consequence, to old and well known analogue sound representation: infinite (theoretically) sampling rate and infinitely small (down to quantic scale) bits size and infinite number of them.

And that will be, again, a solution in search of a problem (the limits of analogue that digital tryed to avoid).

 

That's why there are delta-sigma DACs which allow you to reduce number of bits as sampling rate increases, because you don't need to encode vast empty silence at frequencies far in the 100+ kHz range...

 

Even without such or other fancy stuff, you gain about one bit worth of dynamic range at every doubling of sampling rate. So you could as well remove one bit out of the DAC every time you double the sampling rate.

[Miska]

But, interestingly, what about own ringing of speakers?

 

I suppose, electromechanical oscillatory processes exists there.

 

Yes, sure. Of course coincidentally that looks like ringing of a minimum-phase filter.

 

That can be measured both acoustically and also with LDV...

[audiventory]

Yes, sure. Of course coincidentally that looks like ringing of a minimum-phase filter.

 

That can be measured both acoustically and also with LDV...

 

I suppose, laser is more precise for this kind of measurements, because there no air as intermediate medium.