DSD vs PCM resolution

[Miska]

No spectrum plots or spectrograph plots, please. I am interested only in looking at the time domain performance for this exercise. That is the raw data.

 

 

One other problem I have with these plots. There is no way to calibrate the depth of null without knowing the FFT gain. So, for example, at what level does 16 bit dither noise show up in these plots?

 

What you see is the 16-bit dither noise. The original 44.1/16 is dithered with TPDF dither and the conversion result is also dithered with TPDF dither.

[chrille]

My standard was, and still is, a live microphone feed vs. something recorded and then played back. I never "got used to" any kind of distortion, be it tape hiss, tape saturation, vinyl distortion due to geometry errors or mis tracking, FM radio transmitter/receiver distortion (including with a Marantz 10B), not to mention the myriad of complex distortions created by the CD process, especially early versions thereof.

 

I don't listen to amplified music, except for jazz vocals and then I do not judge the sound quality based on the vocals, just the instruments. My belief is that people who do not listen to acoustic music are not qualified to judge music the way that I do. They can enjoy their music if they like it. I will enjoy mine. It would be better if there were separate forums for discussing audio reproduction, segregated according to acoustic vs. non-acoustic. Many forum discussions have gone off the rails, until it turns out that the factions are listening to completely different music and judging what they here by completely different standards.

 

It would be really good if reviewers too actually used acoustic music only because in a true HI FI sense there can only be one reference, live acoustic in the hall compared to mic feed and how close that sounds via any product in a reproducing chain actually sounds to the acoustic reference.

Yes let's sort the wheat from the chaff!

I sometimes laugh out loud when I read comments or reviews here and elsewhere which don't realize this simple and fundamental fact.

[audiventory]

No spectrum plots or spectrograph plots, please. I am interested only in looking at the time domain performance for this exercise.

 

Time domain is not sensitive to distortions as spectrogram.

 

Sine with ideal (for eyes) oscillogram can have -60 dB distortions. -100 dB distortions that is not allowable we cant see on oscillogram.

[Tony Lauck]

I don't look at oscillograms. I look at the actual numbers in the files. I compare them by taking their differences and look at the numbers. This will provide a proof of identity between the two files if there is no distortion, but this will fail if there is any distortion. Audio editors will do this. First you have to line up the two files so that samples are matched if for some reason they are out of sync. Then you mix the two files together out of phase, a function all editors have. Then you look at the difference file, look at maximum difference, RMS difference, even listen to it. This requires that the two files be synchronized, so if the sample rate conversion is not done synchronously then there won't be a suitable null. Also, it requires that any sample rate conversions or filters don't add any delay, i.e. they must be linear phase with identical delay between the two files, down to a minute fraction of a sample other wise the null will fail. And the gain must be perfectly matched.... If this is not done then timing changes and gain changes that would be completely inaudible may give false differences that will hide any real differences in the file. All of this is relatively easy to do in the digital domain and very difficult to do in an analog system.

[Miska]

I don't look at oscillograms. I look at the actual numbers in the files. I compare them by taking their differences and look at the numbers.

 

That is not very useful in the 44.1 -> 96 -> 44.1 case, because the samples are/may not be in the same phase position they originally were... And same goes for DSD too, depending a bit on a case. So you cannot look straight at the sample values, you'd need to realign the sample positioning in sub-sample precision...

 

Spectrogram tells exactly any possible errors, as we know FFT is completely reversible back to the original sample values.

 

But with PCM conversion in a NOS mode you begin to get timing errors you can detect from DAC output using scope in DPO mode, you can see that edge of the square becomes unstable because samples are not synchronous to the sampling rate:

 

You can also see it with sine waves:

 

But this is all old stuff we've gone through many times before. So I'm getting a bit tired, if we are not proceeding to something new.

[audiventory]

samples are not synchronous to the sampling rate

 

What does you mean? How sync troubles relate to NOS?

[Miska]

What does you mean? How sync troubles relate to NOS?

 

Let's say you have 44.1k sampling rate and 7 kHz sine wave, the samples are not placed always at the same position of the waveform. If you run a NOS DAC at 44.1k it is really difficult to have perfect analog reconstruction filter and thus edginess of the waveform leaks through and it can be seen with the phenomenon I've shown above. In above case DAC is running at 176.4k in "NOS" mode thus quivalent of 4x oversampling ratio used for example with old TDA1541A.

 

Higher the oversampling ratio, more the waveform converges with the real waveform. You can also see this in the spectra results as long as you use high enough analysis bandwidth.

 

Main challenge with ladder PCM DACs is you cannot keep increasing the oversampling ratio much, because settling time becomes an issue. There's no advantage of using high oversampling ratio if the conversion section doesn't manage to settle properly to ½ LSB accuracy within short fraction (about 0.1%) of the sample time. PCM1704 supports max 768 kHz rate. This is one of the reasons why SDM DACs became popular, because if you use noise shaping, you can keep reducing conversion accuracy as the oversampling ratio becomes higher and thus the boundaries for ½ LSB become more and more relaxed.

 

24-bit ladder DAC would need to settle to 0.15 µV while DSD DAC needs to settle to 2.5 V.

[audiventory]

If you run a NOS DAC at 44.1k it is really difficult to have perfect analog reconstruction filter and thus edginess of the waveform leaks through and it can be seen with the phenomenon I've shown above. In above case DAC is running at 176.4k in "NOS" mode thus quivalent of 4x oversampling ratio used for example with old TDA1541A.

 

Me seems, I understand what you want say. You said about "tremor of phase» due noise (unfortunately, I don’t know right English term). However for it noise

 

Let's say you have 44.1k sampling rate and 7 kHz sine wave, the samples are not placed always at the same position of the waveform.

 

I suppose explanation don’t depend on samples placed on sine wave (if don’t account ADC's instability).

 

[ATTACH=CONFIG]20200[/ATTACH]

 

If sample rate of ADC and DAC is absolute stable (zero sample rate deviation) no difference between input and output oscillogram as showed above.

 

If will some instability of sample rate clock of DAC/ADC/both we can see such difference between input and output oscillogram as showed above.

 

And you said that non-filtered noise above [sample rate/2] can give same effect, isn’t it?

 

 

Higher the oversampling ratio, more the waveform converges with the real waveform. You can also see this in the spectra results as long as you use high enough analysis bandwidth.

 

Yes. We can consider oversampling like restoring via analog filter.

 

 

Main challenge with ladder PCM DACs is you cannot keep increasing the oversampling ratio much, because settling time becomes an issue. There's no advantage of using high oversampling ratio if the conversion section doesn't manage to settle properly to ½ LSB accuracy within short fraction (about 0.1%) of the sample time. PCM1704 supports max 768 kHz rate. This is one of the reasons why SDM DACs became popular, because if you use noise shaping, you can keep reducing conversion accuracy as the oversampling ratio becomes higher and thus the boundaries for ½ LSB become more and more relaxed.

 

24-bit ladder DAC would need to settle to 0.15 µV while DSD DAC needs to settle to 2.5 V.

 

Of course, simpler provide 1 level instead 16777216 (24 steps).

 

Why 0.15 µV?

 

Linear output is 250 mV. Therefore must be 250 mV/16777216 = 0.015 µV

[mordante]

It would be really good if reviewers too actually used acoustic music only because in a true HI FI sense there can only be one reference, live acoustic in the hall compared to mic feed and how close that sounds via any product in a reproducing chain actually sounds to the acoustic reference.

Yes let's sort the wheat from the chaff!

I sometimes laugh out loud when I read comments or reviews here and elsewhere which don't realize this simple and fundamental fact.

 

I strongly disagree.

 

You have no way of knowing how the sound was altered during mastering, what mics were used. How the mics were setup. What sound the conductor (if there was one) was aiming for. Maybe (most likely) the reviewer is unfamiliar with the hall/room where the recording took place. There are many more unknown variables.

[chrille]

I strongly disagree.

 

You have no way of knowing how the sound was altered during mastering, what mics were used. How the mics were setup. What sound the conductor (if there was one) was aiming for. Maybe (most likely) the reviewer is unfamiliar with the hall/room where the recording took place. There are many more unknown variables.

 

What if I actually have acess to quite a few masterfiles both DSD and PCM up to DXD where I have actually both been to the sessions for sometimes up to a week 8 hours a day, and heard the orchestra both live and mic feed and via various both pro and consumer products?

 

You are really barking up the wrong tree this time.

 

And to make it really clear once more: nothing else than acoustic music can ever be a proper reference in any true HI FI comparison between products.

By all means enjoy your pop rock or whatever, but it has no real significance in HIFI.

Like the post I responded to said we really need to establish a valid reference before embarking on any valid discussion of SQ in this FORUM.

Let's sort the wheat from the chaff.

A lot of very opionionated posters here don't even have the foggiest idea what real reference SQ is, simply because they don't listen to music or recordings that can be of any real reference value.

Cheers Chrille

[Jud]

What if I actually have acess to quite a few masterfiles both DSD and PCM up to DXD where I have actually both been to the sessions for sometimes up to a week 8 hours a day, and heard the orchestra both live and mic feed and via various both pro and consumer products?

 

You are really barking up the wrong tree this time.

 

And to make it really clear once more: nothing else than acoustic music can ever be a proper reference in any true HI FI comparison between products.

By all means enjoy your pop rock or whatever, but it has no real significance in HIFI.

Like the post I responded to said we really need to establish a valid reference before embarking on any valid discussion of SQ in this FORUM.

Let's sort the wheat from the chaff.

A lot of very opionionated posters here don't even have the foggiest idea what real reference SQ is, simply because they don't listen to music or recordings that can be of any real reference value.

Cheers Chrille

 

Without doubt being present at the session itself and hearing the mic feeds is the best reference. For the vast majority of us who do not have that opportunity, all is fortunately not lost.

 

What we must listen for is *variety* in the presentation of the music. Different tracks, even on the same album, should sound different from each other. To the extent things are the same from track to track and album to album, you are hearing the sound of the equipment, and when equipment has a sound of its own that is by definition distortion. Conversely, to the extent different tracks sound different from each other, the equipment is letting the music come through, which is the definition of "fidelity," or accuracy to the input.

 

I know this comes off a little vague, but it is not uncertain at all in practice. I have a recording with a deliberately squashed soundstage. (It's Tom Waits' version of "Heigh Ho" from the Snow White soundtrack. In Wait's rendition it is authentically something that would have come from dwarves working in a coal mine.) Any equipment that renders this with a big, glorious soundstage is not accurate. I have another recording of jazz bass where the player is extremely, erm, "enthusiastic." (It's the opening track of Brian Bromberg's "Wood" album, entitled "The Saga of Harrison Crabfeathers.") He is playing a 300-year-old German bass as if it's a Fender electric in a death metal band. If it sounds like just another jazz bass and you don't absolutely fear for the continued physical integrity of that 300-year-old German instrument, your system is rounding off transients and muting dynamics.

 

So yes, even if you haven't been privileged to be present at the session, there is still hope.

[Miska]

If sample rate of ADC and DAC is absolute stable (zero sample rate deviation) no difference between input and output oscillogram as showed above.

 

This doesn't depend on stability of sample rate and you don't need to consider ADC. It depends on incomplete reconstruction. Perfect reconstruction of RedBook would require analog lowpass that has -96 dB attenuation by fs-bw frequency, where fs is sampling rate and bw is the passband. So a DAC playing RedBook that has 4x oversampling would require analog lowpass filter that has -96 dB attenuation by 176400-22050 = 154350 Hz, assuming perfect brickwall digital filter for the 4x oversampling. So sixth order filter with fc=25 kHz would be close, however the phase response in audio band wouldn't be nice.

 

For example 1333 Hz square at 44.1k sampling rate you can notice how sample positions on the corners run, because the sampling rate is not in sync with the signal. Through incomplete reconstruction this can be seen also in the DAC output using a DPO scope like I explained.

 

However, the same upsampled to 352.8k sampling rate helps to stabilize the waveform and the analog filter can be relaxed too.

 

And you said that non-filtered noise above [sample rate/2] can give same effect, isn’t it?

 

Essentially it is about reconstruction which is about removing noise above fs/2 and in case of PCM specifically about removing correlated noise (causing of course deterministic variations).

 

Yes. We can consider oversampling like restoring via analog filter.

 

Analog filter is always needed, because no matter how high the sampling rate, the images above fs/2 must be removed for proper reconstruction. Just the analog filter specs become more relaxed because distance between audio band and first image band increase.

 

Why 0.15 µV?

 

Linear output is 250 mV. Therefore must be 250 mV/16777216 = 0.015 µV

 

I was using the same 5 Vpp spec for both. Normal line output is supposed to be about 2 Vrms, but quite many devices get closer to 2.54 Vrms. (and settling needs to be within ½LSB so half of the smallest possible step)

[YashN]

And to make it really clear once more: nothing else than acoustic music can ever be a proper reference in any true HI FI comparison between products.

 

I agree, technically acoustic is the only way to even have a proper reference and try to reproduce it.

 

We can't easily do the same thing with electronic or amplified music, since it's too equipment-dependent and one would need the exact same equipment to do a real Hi-Fi reproduction - not easy.

[Miska]

And to make it really clear once more: nothing else than acoustic music can ever be a proper reference in any true HI FI comparison between products.

By all means enjoy your pop rock or whatever, but it has no real significance in HIFI.

 

I think I know for example how close miked drum kit should sound like. And I know how different kinds of digital artifacts sound like and if I think I don't remember anymore I know how to generate those as examples.

 

But what I've also been doing is collecting instruments that I know sounding tricky (including keys jangling) and recording those with different equipment as well as playing the recordings back on different equipment. Plus doing the normal objective analysis of those things.

 

Stuff like soprano glockenspiel, metallic and wooden claves, wooden maracas, wood tone block and good old wooden Spanish castanets.

 

Recording distance is same as my hand-to-ear distance so I can compare how I heard things IRL to how the recording sounds like. I also try to match the playback volume so that the SPLs match (on Sennheiser HD800 headphones).

 

In the past I've been recording, playing and hearing quite a bit of piano + violin combination in home environment so I have pretty good idea of those too.

[Jud]

Stuff like soprano glockenspiel, metallic and wooden claves, wooden maracas, wood tone block and good old wooden Spanish castanets.

 

 

/Gets image of Miska prancing about with castanets, wants it to go away!

[audiventory]

Yes. We can consider oversampling like restoring via analog filter.

 

Analog filter is always needed, because no matter how high the sampling rate, the images above fs/2 must be removed for proper reconstruction. Just the analog filter specs become more relaxed because distance between audio band and first image band increase.

 

I said «like». Not «instead»

 

 

 

What about test music stuff I also preffer acoustic ones: guitars, cymbals, trumpets.

 

However for testing bass for me better synthetic signals - notes with special adjusted:

1. length,

2. time position,

3. pitch sequence of notes,

4. pause's times between notes,

5. spectrum.

 

Some such combinations allow to light weak places of bass range of system including subwoofer.

 

I first time pay attention to it after listening on test disk for bass range.

 

One of song was most "hard" for playback and tuning of system with subwoofer. There was unusual rhythm also.

 

I made experement: recorded some [only bass music] test tracks with different combinations described above.

 

During listening I operative change parameters of listened samples for achieving of effect "hard playback" like listening the test disk.

 

For better result for different systems need have several such tracks (in different pitch).

[Hazard]

Bonsoir,

 

 

 

I did try a very similar "no-chip" DAC (DIYINHK USB Board with isolator and external clock) after it was talked up so much in some babillards. I used 3rd order passive filter at 100kHz.

 

Using Foobar2k's DSD converter I tried this at the different supported DSD rates. Even at DSD256 I felt the original Pass converter with PCM63 DAC Chip's did a much better job of sounding "analogue" or "like the best of vinyl minus clicks, pops, noise and tracking distortion". At slower speed the advantage of the Pass converter increased, I felt CD->DSD64 by comparison produced a hilarious dummy sound, caricature de la musique de merde.

 

But the killer was listening to actual DSD downloads. I found that many DSD files I played via this kind of "chipless DSD DAC" had very audible distortion and background sounds I would call "birdies" (based on my Ham radio days). Merde de merde.

 

Funny thing, DSD converted to 88.2kHz PCM and played via Pass D-1 sounds okay, but not as a good CD.

 

Salud M.I.

MI, sorry ti hear that your experience was not positive. I would add a couple of things:

- my prototype was picking up lots of noise. I know what you mean about birdies. But now that I have settled down on the circuit layout, I have re-done all the wiring and got rid of a lot of 'antennas" that were the result of an untidy build. At megahertz, layout is critical. The noise is gone with a proper layout.

- I use a flip flop with complementary outputs for a balanced signal. Perhaps this helps with noise. My motivation was simply that I have a balanced system, and I wanted to keep this new source balanced as well. But I think that noise rejection may be another benefit.

- Forget Foobar. I used foobar for years, but I found HQPlayer a few months ago and there is no going back.

- I have a DIYINHK board, and as far as I can tell it only plays DSD as DoP. Therefore it is limited to DSD128 AFAIK. How did you get DSD256?

- I am now using a JLSounds USB card. With Linux (Ubuntu Studio 14) and HQ Player I can play DSD256. The only LPF is the capacitance in the Lundahl transformer.

 

Regards, Hazard

[Paul R]

What if I actually have acess to quite a few masterfiles both DSD and PCM up to DXD where I have actually both been to the sessions for sometimes up to a week 8 hours a day, and heard the orchestra both live and mic feed and via various both pro and consumer products?

 

You are really barking up the wrong tree this time.

 

And to make it really clear once more: nothing else than acoustic music can ever be a proper reference in any true HI FI comparison between products.

By all means enjoy your pop rock or whatever, but it has no real significance in HIFI.

Like the post I responded to said we really need to establish a valid reference before embarking on any valid discussion of SQ in this FORUM.

Let's sort the wheat from the chaff.

A lot of very opionionated posters here don't even have the foggiest idea what real reference SQ is, simply because they don't listen to music or recordings that can be of any real reference value.

Cheers Chrille

 

I disagree of course. Why am I beginning to feel that people are bound and determined to tell other people what they shoukd hear and believe in this hobby? Must be an internet phenomena or something...

 

A great deal of music has been made over the past 65-70 years that depends upon instruments like electric guitar, electric organ/piano, studio recording and so forth. If one enjoys that type of music, it is sheer insanity to ignore it when designing one's system.

 

This whole idea of some elitist absolute sound is not onky wrong, but accepted by far too many peoole as fact and not just opinion. There are some people who don't like to listen to mush orchestral recordings, and digital has the nasty habit of bringing to light every posdible flaw in those recordings as well. Especially since a middle of the road digital system is way more revealing than a similar priced vinyl system.

 

If *anything* - what the engineer hear through the mic feeds is the absolute sound to reproduce, and nobody but the engineer has any reference to that. I see some folks buying stuff that makes sounds they do not really enjoy because it is "the best at reproducing the sound of a flute" - insanity!!

 

So- as has been pointed out, the variety of sound that your system is capable of reproducing is one measure that anyone can use, and come to accurate judgements about, and is not very dependent upon the type of music one listens to. If everything sounds the same, you got some work to do...

 

Now, if you want to only use the sound of acoustic instruments- go ahead. But please, stop putting on superior airs. (grin- pun intended of course...)

[magnum innominandum]

Bonjur,

 

At megahertz, layout is critical. The noise is gone with a proper layout.

- I use a flip flop with complementary outputs for a balanced signal. Perhaps this helps with noise. My motivation was simply that I have a balanced system, and I wanted to keep this new source balanced as well. But I think that noise rejection may be another benefit.

 

I build this fully balanced using Potato Semi PO74G74A Flip Flop with complementary outputs and fully balanced signal path (my whole system is) and with a super-reg kit for the supply. PCB made from solid copper plane FR4 for groundplane and SMD Cap's etc. in very tight "dead bug" style. You can prototype GHz stuff like that, we did that back when I worked as EE for that Texan Semi maker...

 

Birdies did not happen for any PCM material converted with Foobar2k DSD upsampler even to DSD64. They only happened with real DSD recordings (SACD Rips more precisely).

 

As said, compared to a good multibit DAC I found the sound quality lacking, though compared to some contemporary DAC's I would say this little "Flip-Flop" DAC did quite well, if the sample rate was high enough... But I already own a better DAC.

 

- Forget Foobar. I used foobar for years, but I found HQPlayer a few months ago and there is no going back.

 

I use J-River. I only use the PCM-DSD conversion part of Foobar in this experiment.

 

I downloaded the Trial of HQ Player. Sorry, I have never encountered such an infuriatingly unusable piece of software before. Usability scores on one out of ten are minus infinity. I pass, thank you.

 

- I have a DIYINHK board, and as far as I can tell it only plays DSD as DoP. Therefore it is limited to DSD128 AFAIK. How did you get DSD256?

 

The problem is the XMOS USB Driver, not the hardware/xmos firmware which has support build into the code. Some unscrupulous people in a part of the world where copyright is mostly ignored have been modifying drivers to allow many cheap chinese devices to use ASIO DSD.

 

http://www.pchifi.cn/forum.php?mod=viewthread&tid=104816&extra=&page=1

 

Salud M.I.

[audiventory]

If *anything* - what the engineer hear through the mic feeds is the absolute sound to reproduce, and nobody but the engineer has any reference to that. I see some folks buying stuff that makes sounds they do not really enjoy because it is "the best at reproducing the sound of a flute" - insanity!!

 

Current audio systems (that I know) can't made full reproducing of concert hall. Due there need re-create full wave picture with correction to listening room. It is not amplitude or phase response matter.

 

In my opinion, only records like by Barry Diament's 2 mics head model, is the most approached to the reproducing now. For full effect, me seems, need headphones playback. It allow avoid sound re-reflections in listening room, that distort recorded sound picture.

[Paul R]

Current audio systems (that I know) can't made full reproducing of concert hall. Due there need re-create full wave picture with correction to listening room. It is not amplitude or phase response matter.

 

In my opinion, only records like by Barry Diament's 2 mics head model, is the most approached to the reproducing now. For full effect, me seems, need headphones playback. It allow avoid sound re-reflections in listening room, that distort recorded sound picture.

 

I agree, and it is notable - that model can deal with acoustic instruments, electric instruments, vocals, and anything else that one cares to use it for.

 

-Paul

[Jud]

Current audio systems (that I know) can't made full reproducing of concert hall. Due there need re-create full wave picture with correction to listening room. It is not amplitude or phase response matter.

 

In my opinion, only records like by Barry Diament's 2 mics head model, is the most approached to the reproducing now. For full effect, me seems, need headphones playback. It allow avoid sound re-reflections in listening room, that distort recorded sound picture.

 

https://members.nativedsd.com/albums/just-listen-1-compilation

 

See the binaural versions of Beethoven's "Heiliger Dankgesang." It's a beautiful piece, well recorded, and might make good test material.

[audiventory]

Hi Jud. Thank you for link.

[YashN]

- my prototype was picking up lots of noise. I know what you mean about birdies. But now that I have settled down on the circuit layout, I have re-done all the wiring and got rid of a lot of 'antennas" that were the result of an untidy build. At megahertz, layout is critical. The noise is gone with a proper layout.

- I use a flip flop with complementary outputs for a balanced signal. Perhaps this helps with noise. My motivation was simply that I have a balanced system, and I wanted to keep this new source balanced as well. But I think that noise rejection may be another benefit.

- Forget Foobar. I used foobar for years, but I found HQPlayer a few months ago and there is no going back.

- I am now using a JLSounds USB card. With Linux (Ubuntu Studio 14) and HQ Player I can play DSD256. The only LPF is the capacitance in the Lundahl transformer.

 

Yep to all the above, Hazard, and great thread over at diyaudio which became much better than my older one.

 

The so-called 'birdies' most probably had to do with the implementation since the down-conversion to 88 PCM was OK, but some people conclude the format isn't good...

[YashN]

Current audio systems (that I know) can't made full reproducing of concert hall. Due there need re-create full wave picture with correction to listening room. It is not amplitude or phase response matter.

 

I think we can approach it with DSP in the PCM domain. However, nowadays I am not too fond of any DSP treatment for my music, I'd rather use high-rate DSD with gentle filtering.

 

We're thinking of going to see Kent Nagano to calibrate our ears for a fuller orchestra than what we heard in Toronto with Peter Gabriel's New Blood Tour Live.

 

I am especially interesting in noticing where the system fails in reproduction in this case, and finding organic and acoustic ways of making things better.