DSD playback via computer better than from SACD

[MylesBAstor]

How many commercial recordings are available in DSD format? And from those, what percentage is pure DSD, never having been converted to PCM and back to DSD at some point?

I think you already know the answer to that question

 

[microstrip]

I think you already know the answer to that question

As I am including SACDs I do not have the faintest idea. And I am really interested in an exact answer, because Devialet's do not accept DSD format and as far as I know do not consider it a priority ...

 

[docvale]

As I am including SACDs I do not have the faintest idea. And I am really interested in an exact answer, because Devialet's do not accept DSD format and as far as I know do not consider it a priority ...

Are you considering to shrink your system down to a Devialet? It has to be impressively good
I wish I may listen to one, sometimes...

 

[Bruce B]

How many commercial recordings are available in DSD format? And from those, what percentage is pure DSD, never having been converted to PCM and back to DSD at some point?

The answer to that can be found over at SA-CD.net

 

[jkeny]

Delivery where? I do not know if the processing DSD bits see is the same that SACD uses by the time it gets to the DAC;

Ok, you can help me out in my lack of knowledge here. Is DSD & SACD not one & the same?

I would be surprised if the ECR and digital filtering is the same but do not know.

If DSD & SACD are the same then this is a mute point, no?

On the output side, delivery of the correct bits does not mean the analog output is "good", but it should be the same if the input bits are the same.

Don, you already put in the proviso on this statement & a very big proviso it is "all else equal" & in this "all else" you mentioned timing, filters. I think that you might be misunderstanding what was done here & I should have picked it up earlier - both the SACD stream & the DSD stream from HDD are both being processed by the same DAC. AFAIK, so your citing of different PSRR, CMRR, etc is mute.

So I'm not sure why you have now I think I am not following so by definition must be "blinkered".

By your statements earlier, I thought you understood that there is more to it than bit-identical files so I'm now confused with your latest post?

I don't think I can contribute to this discussion but it is interesting to watch.

Of course you can contribute & I appreciate your expertise & would appreciate your involvement!

 

[DonH50]

Yes, I am confused, and we are crossing wires here, my bad. I do not know the difference, if any, between DSD and SACD, especially at the DAC inputs. Sorry I did not make that clear. My background is not audio, this is just a hobby. I did not look up DSD and SACD formats; one reason I need to stay out of this discussion.

"All else equal", yes, a thorny concept. However, if the bits at the DAC's inputs are identical for the two sources shouldn't the output be the same? What is non-deterministic?

I was thinking there must be some difference in signal processing, whether due to input transceivers, clock generation/recovery, filters applied, whatever between the DSD input and SACD input to the system to cause the bits at the DAC to be different. Is that what you are thinking? If so, we agree, and "bit identical" from the source means little because the signal path to the DAC is different. If that assumption is wrong, and identical bits at the DAC cause different outputs, I am at a loss to explain it save for external effects on the DAC and analog chain.

Sorry, I feel I jumped into the middle of this without really knowing where it's been and where it's going. - Don

 

[jkeny]

Yes, I am confused, and we are crossing wires here, my bad. I do not know the difference, if any, between DSD and SACD, especially at the DAC inputs. Sorry I did not make that clear. My background is not audio, this is just a hobby. I did not look up DSD and SACD formats; one reason I need to stay out of this discussion.

I understand. I believe your expertise is in the field of computers, forgive me if I'm wrong or defining it as something lesser.

"All else equal", yes, a thorny concept. However, if the bits at the DAC's inputs are identical for the two sources shouldn't the output be the same? What is non-deterministic?

Ah, I thought we were already addressing this earlier - timing (jitter), CM noise at the inputs. Both of these can vary & yet the streams remain bit identical. Both of these differences will result in different analogue outputs. That is why I was so interested in Bruce's null test & any difference file he might have!

I was thinking there must be some difference in signal processing, whether due to input transceivers, clock generation/recovery, filters applied, whatever between the DSD input and SACD input to the system to cause the bits at the DAC to be different. Is that what you are thinking? If so, we agree, and "bit identical" from the source means little because the signal path to the DAC is different. If that assumption is wrong, and identical bits at the DAC cause different outputs, I am at a loss to explain it save for external effects on the DAC and analog chain.

My thinking is as above - other differences in the digital stream besides differences in bits.

Sorry, I feel I jumped into the middle of this without really knowing where it's been and where it's going. - Don

No problem.

 

[Bruce B]

To shine a little light on this, here is what I did.

1. I ripped an SACD. Just as .wav is on a CD, DSD is on a SACD.

2. I loaded the ripped DSD file into Sonoma. Sonoma workstation feeds the Playback Designs MPS-5 via AT&T ST-glass fibre.

3. I played the SACD from the Playback Designs MPS-5.

4. I captured the analog signal one at a time into a Pyramix workstation at 32/352.8kHz. I lined up both files from a known bit and then hit play. I reversed the phase of one track. I got signals as high as -34dB!

 

[jkeny]

Thanks, Bruce
So the Sonoma can play DSD files & Playback designs handle the DSD files, natively - no need for USB packaging!
The one proviso in your null test is that you are introducing an analogue to digital conversion stage & this might mask some differences or cause variations in the results.
Have you had much success with null tests in other comparisons where the difference signal was effectively null i.e you conducted a control to test that your null tests work as expected?
Have you done the same null test with CD & the same file ripped to .WAV on disk?

 

[RBFC]

Does the Playback Designs use an apodizing filter or any other processing of DSD/CD for playback?

Lee

 

[Bruce B]

Thanks, Bruce
So the Sonoma can play DSD files & Playback designs handle the DSD files, natively - no need for USB packaging!
The one proviso in your null test is that you are introducing an analogue to digital conversion stage & this might mask some differences or cause variations in the results.
Have you had much success with null tests in other comparisons where the difference signal was effectively null i.e you conducted a control to test that your null tests work as expected?
Have you done the same null test with CD & the same file ripped to .WAV on disk?

We've done null tests here for years with CD's burned at different speeds with different media. We've done the same with CD and the files did null down to around -80dB or so. Figured it was just noise/dither.

 

[DonH50]

jkeny: My profile is on LinkedIn. My expertise is high-speed integrated circuit design, transistor-level, focused primarily on ADCs and DACs used in RF/mW systems. Virtually all analog bipolar, examples include 16-bit 100+ MS/s DAC (switched at better than 1 GS/s) and 8-bit 5-GS/s ADC (>10 GHz input bandwidth). I am now working for customer support in a computer storage company, but still mostly HW/analog stuff... I have done a fair amount of systems work, and in the primordial past played in the audio biz, tech/sales and also recording/mastering/live sound and as a gopher (mainly) doing sound installs (commercial and private).

Bruce: Wow, (as you know) -34 dB means something is pretty dramatically different in the chain...

 

[microstrip]

Are you considering to shrink your system down to a Devialet? It has to be impressively good
I wish I may listen to one, sometimes...

I am not considering to shrink, but could not resist a kind offer to try the Devialet again.

The Devialet seems to be impressive when used with the Sonus Faber Amati Futura or the Aida.

I hope to listen to two Devialet's in mono with the Aida's soon - although I loved some aspects such as detail, bass and making the speaker disappear, the sound is sometimes constrained using just one unit. Our distributor told me that he has once assembled a great system with the TheSonusFaber using six Devialet's!

 

[jkeny]

Ah, sorry, Don, I suspected I was doing you a disfavour with my description of your expertise. Apologies & I feel you have a lot to contribute to this thread so please stick around. Let's avoid it getting into trench warfare & try to start with an analysis of why the exact same SACD & DSD file sound different when played back from different media as testified by Cookie & Bruce!

 

[microstrip]

The answer to that can be found over at SA-CD.net

Bruce,

Thanks - but I could only find the number of releases in SACD - no general statistics about percentage of pure DSD recordings.

 

[Bruce B]

Bruce,

Thanks - but I could only find the number of releases in SACD - no general statistics about percentage of pure DSD recordings.

If you go over to the right at the top of the page under the search, there is a drop down box that says "DSD Recordings". I see about 1523 so far.

 

[DonH50]

jkeny: No problem, no way to know my background from Internet posts, and I am sure many here have far more expertise and credibility in the audio arena.

I am very curious as to what exactly is causing the discrepancy; clearly something in the chain is very different. If we cannot determine the exact signal path, it would be interesting to do some simple tests (single and two-tone sine waves, (im)pulse tests) to see what frequency and time-domain analysis might show. With such a large difference in the null tests, it must go beyond error correction (since I would expect low BER for either stream). I am guesing now the two paths go through different digital filters on the way to the DAC.

One thing I did not ask/do not know is if it is a delta-sigma DAC and if the modulator changes for the two inputs.

Are the DSD and SACD bit streams the same resolution and sample rate in both cases? No sample rate or bit depth convversions? I need to go back a little and read about the test case.

Bruce, is there a link to a technical description of the DAC unit you used?

 

[Bruce B]

Bruce, is there a link to a technical description of the DAC unit you used?

Playback Designs MPS-5

Stereophile Review

 

[DonH50]

Interesting. The measurements show clear differences between DSD and SACD signals. I strongly suspect different filters are being used in the processing of the two signals.

 

[jkeny]

We've done null tests here for years with CD's burned at different speeds with different media. We've done the same with CD and the files did null down to around -80dB or so. Figured it was just noise/dither.

Thanks, Bruce, I missed this post.
So null tests of CDs burned at different speeds showed no marked difference in null tests (& I presume no difference in listening)?

I'm trying to remember who it was that is also involved in the recording area, posting that he could clearly hear the difference between the CD master & the copies done in different CD repro facilities (& the repros sounded different from one another also). He tested the digital output from each of these CDs & found them to be bit-perfect & identical. I don't think he did any null tests, however. I can't remember who it was or I would link to the posts!