DSD comparison to PCM.

[zenelectro]

The jury is still out, but I still think the Grimm sounds better. Just wish it would do DSD128/256fs.

The Horus, I believe uses Arda AT1201 ADC and looking at datasheet it has some pretty impressive specs - until you look at it's DSD performance.
AFAICS they definitely were thinking PCM on this one. It may well do DSD256 but the performance looks a bit compromised.


cheers

Terry

 

[tailspn]

AFAICS they definitely were thinking PCM on this one. It may well do DSD256 but the performance looks a bit compromised.

http://www.ardatech.com/pdf/AT1201 DS.pdf

Yes, the Horus, along with the Ayer QA-9 do use the Arda AT 1201. What's unknown is whether Merging used the DSD outputs(s), or Multi-bit output(s) with their own Delta-Sigma Modulator.

 

[Bruce B]

http://www.ardatech.com/pdf/AT1201 DS.pdf

Yes, the Horus, along with the Ayer QA-9 do use the Arda AT 1201. What's unknown is whether Merging used the DSD outputs(s), or Multi-bit output(s) with their own Delta-Sigma Modulator.

Testing this one out now... QA-9

 

[MylesBAstor]

I personally prefer originally recorded DSD, not processed with DXD, over extremely high resolution PCM, and I would claim that DXD/PCM is not transparent with multi-channel content.

The reduction in image resolution and spaciousness is easily recognizable when comparing a multi-channel analog mixed DSD edit master, with the same content converted to DXD with Pyramix. The same may not be recognizable in stereo, due to the significant reduction in information content, and resulting aural cues.

I do agree with Morton that "I find that the placement of microphones has an infinite more important role in the final experience of music, than the difference between HiRes PCM and DSD."

Tom

Or possibly the recording venue?

 

[tailspn]

Or possibly the recording venue?

Absolutely! Given that all other elements of a recording being equal, it's a large, or largest contributor to a great sounding recording. And that extends to how the venue is populated. It was very instructive listening to the Levine Boston Symphony DSD archival done by SoundMirror over a seven year period, both the stereo and multi-channel, between the empty hall rehearsals, and the filled hall concert takes. It was hard to believe that all that changed was the absorptive presence of 2,000+ people. It was a very different sound.

 

[zenelectro]

http://www.ardatech.com/pdf/AT1201 DS.pdf

Yes, the Horus, along with the Ayer QA-9 do use the Arda AT 1201. What's unknown is whether Merging used the DSD outputs(s), or Multi-bit output(s) with their own Delta-Sigma Modulator.

I didn't notice any DSD specs on the Horus data sheet / tech info. There may be a good reason why they don't publish them

 

[tailspn]

I didn't notice any DSD specs on the Horus data sheet / tech info. There may be a good reason why they don't publish them

Does this help?

http://www.filedropper.com/horusdxddsdspecsv81b2

Sorry for the gymnastics, I can't find a simple way to post a link to this 10 page Word doc.

The explanation on page 5 of figure 3 of the above document answers my question of whether Merging used the AT1201 on-board Delta-Sigma Modulator to derive DSD from the multi-bit, or used their own modulator:

"The above FFT, using Merging’s own DSD 256 re-modulator shows a further significant improvement over the DSD256 signal as generated directly by the AT1201 chip itself. Using our own proprietary design to extract the best possible 1-bit signal out of the chip’s internal 6-bit 256fs sigma-delta quantizer, it is possible to achieve a flat noise response to beyond 50 kHz with a measured dynamic range of over 118 dB in the 20 Hz-20kHz range and still close to 112 dB in the 20 Hz-30 kHz range."

 

[zenelectro]

Does this help?

http://www.filedropper.com/horusdxddsdspecsv81b2

Sorry for the gymnastics, I can't find a simple way to post a link to this 10 page Word doc.

The explanation on page 5 of figure 3 of the above document answers my question of whether Merging used the AT1201 on-board Delta-Sigma Modulator to derive DSD from the multi-bit, or used their own modulator:

"The above FFT, using Merging’s own DSD 256 re-modulator shows a further significant improvement over the DSD256 signal as generated directly by the AT1201 chip itself. Using our own proprietary design to extract the best possible 1-bit signal out of the chip’s internal 6-bit 256fs sigma-delta quantizer, it is possible to achieve a flat noise response to beyond 50 kHz with a measured dynamic range of over 118 dB in the 20 Hz-20kHz range and still close to 112 dB in the 20 Hz-30 kHz range."

Thanks for that. Yes looks like their own additional modulator.

The results are mixed and TBH a bit perplexing. At DSD256, for a 22k bandwidth it states 118dB DR. That's same as the old PCM4202 (Korg MR1000) in DSD128. They do state complete round trip I believe (AD + DA) so
that skews results a bit. The FFT graphs are not consistent with this result. An FFT with the 'grass' at -110dB with 20k bandwidth has a DR of nowhere near 118dB. The 'grass' (noise floor) should be down around
-140dB or better - however this will depend on FFT settings.

I'm looking at the FFT of PCM4222 at DSD128 posted here by Bunpei (sorry for referring to another forum) http://www.computeraudiophile.com/a...hest-resolution-recording-pcm4222evm-spec.jpg and it looks staggeringly good + closer to these type of specs. It is just ADC and it is also at -20dB sig level.

The Grimm DSD AD is even better, save OOB noise due to DSD64.

Not sure what all this means if anything. It will be interesting to see what Bruce thinks of the Ayre. Obviously it will be another kettle of fish with its discrete, open loop IP stages and different clocking.

Z

 

[tailspn]

Yes, the FFT plots between the Horus and PCM4222 are not really comparable, one being a system, and the other a component. Witness the difference between the Red (no signal) and Blue (0dB 1KHz) traces on figure 4, along with the explanation below it.

It also needs to be recognized, for better or worse, that Merging has been unique in fielding a professional complete integrated system with Horus, as opposed to yet another translation component. The measurements, and FFT plots are for the complete system, through the mic preamps, A/D, D/A, and analog output stage(s). Unlike their previous Mykerinos systems, for DSD, there's no ability for external Mic pres, A/D converters, D/A converters etc. It's kind of a take it or leave it proposition.

 

[MLGrado]

I know it is digging up and old thread, but this subject is something in which I have great interest, DSD processing.

DSD-Wide was/is Sony's answer to editing DSD. It is advertised as maintaining the sample rate, that is, non-decimating.

I have had conversations with several knowledgeable people. Some who seem to think it is no big deal to create a non-decimated conversion of 1 bit to multibit. Others tell me that they know of no way possible to convert a 1 bit DSD signal to multi-bit without decimating.

So decimated or non-decimated, that is the question. All I can go by are the claims made by Sony about the e-chip and assume that it is non-decimating.

So now, lets assume that it is indeed non decimating. Now we have another problem. I have the DSD-wide white paper right here in front of me, and it is very clear that the FIR filter used in summing the 1-bit signal into multi bit has an impulse response of around 96khz. Why is that a problem? Because filtering with that kind of time smear means DSD-wide is really no better than PCM, regardless of the sample rate. And it is very conceivable that DXD could sound as good or better than DSD-wide, depending on the quality of the filtering.

That said, I know Michael Bishop, an excellent engineer formerly of Telarc fame, is on record saying he prefers the sound of DSD-wide to DXD. I wonder what others who have had the same opportunity to compare think?

 

[CKKeung]

I know it is digging up and old thread, but this subject is something in which I have great interest, DSD processing.
DSD-Wide was/is Sony's answer to editing DSD. It is advertised as maintaining the sample rate, that is, non-decimating.
I have had conversations with several knowledgeable people. Some who seem to think it is no big deal to create a non-decimated conversion of 1 bit to multibit. Others tell me that they know of no way possible to convert a 1 bit DSD signal to multi-bit without decimating.
So decimated or non-decimated, that is the question. All I can go by are the claims made by Sony about the e-chip and assume that it is non-decimating.
So now, lets assume that it is indeed non decimating. Now we have another problem. I have the DSD-wide white paper right here in front of me, and it is very clear that the FIR filter used in summing the 1-bit signal into multi bit has an impulse response of around 96khz. Why is that a problem? Because filtering with that kind of time smear means DSD-wide is really no better than PCM, regardless of the sample rate. And it is very conceivable that DXD could sound as good or better than DSD-wide, depending on the quality of the filtering.
That said, I know Michael Bishop, an excellent engineer formerly of Telarc fame, is on record saying he prefers the sound of DSD-wide to DXD. I wonder what others who have had the same opportunity to compare think?

Hi MLGrado,

It so happens that Audiostream.com interviewed Michal Jurewicz, the designer/boss of Mytek Digital and some of your questions have answer there :
http://www.audiostream.com/content/qa-michal-jurewicz-mytek-digital

 

[bmoura]

Hi MLGrado,

It so happens that Audiostream.com interviewed Michal Jurewicz, the designer/boss of Mytek Digital and some of your questions have answer there :
http://www.audiostream.com/content/qa-michal-jurewicz-mytek-digital

Speaking of Mytek, here's a video of a listener using the Mytek Stereo 192 DSD DAC in a high quality music room auditioning the Vanish album on Stereo DSD Download from Yarlung Records.
https://www.youtube.com/watch?v=m64vGFex72g

 

[Yuri Korzunov]

I know it is digging up and old thread, but this subject is something in which I have great interest, DSD processing.

DSD-Wide was/is Sony's answer to editing DSD. It is advertised as maintaining the sample rate, that is, non-decimating.

I have had conversations with several knowledgeable people. Some who seem to think it is no big deal to create a non-decimated conversion of 1 bit to multibit. Others tell me that they know of no way possible to convert a 1 bit DSD signal to multi-bit without decimating.

So decimated or non-decimated, that is the question. All I can go by are the claims made by Sony about the e-chip and assume that it is non-decimating.

So now, lets assume that it is indeed non decimating. Now we have another problem. I have the DSD-wide white paper right here in front of me, and it is very clear that the FIR filter used in summing the 1-bit signal into multi bit has an impulse response of around 96khz. Why is that a problem? Because filtering with that kind of time smear means DSD-wide is really no better than PCM, regardless of the sample rate. And it is very conceivable that DXD could sound as good or better than DSD-wide, depending on the quality of the filtering.

That said, I know Michael Bishop, an excellent engineer formerly of Telarc fame, is on record saying he prefers the sound of DSD-wide to DXD. I wonder what others who have had the same opportunity to compare think?

MLGrado,

Decimating (with filtration, of course) need for more fast processing DSD and possibilities non-linear processings. It is not depend on 1 bit or multibit DSD can be used.

Correctly done decimating, we can consider, as not damaging audio content (0 ... 20 kHz), comparing further processing (mixing, postproduction).

However using integer math in the processing progect can some decrease quality due accumulation quantization errors by big number of processing.

Better way do all processings in floating point format. It also allow avoid accumulated overload.

Alternative way, do mix from DSD/wide-DSD-sourse without decimation.

But it allowable for strongly linear processing only.

If DSD-source will processed via any non-linear plugin (as example, guitar plugin like "distortion", "overdrive", ...) all non-filtered strong technological DSD noise from ultrasound will shifted to audible range and seriously damage audio content.
Any overload during mix/postproduction lead to same effect.

Example of effect damaging of audio content possibly see here

.

Best regards,
Yuri Korzunov

 

[John57]

My first post

Oh, really? Because you say so? I am not impressed, sorry.

It was a long time ago in this wonderful thread, sorry I was triggered a little by this posting.

I thought there are strong rules here about courtesy? I find MarinJim's reply unhelpful and unnecessary as well as being rude. MarinJim did not present any alternative evidence simply made made a personal remark. On his signature he says he is 'Addicted to the Best', then why not present his best behaviour?

As a newbie here I was really hoping this forum is controlled for behaviour such as this example. It seems not yet...we do aim for self-regulated behaviour don't we rather than 'I can say anything I want to' - this means managing yourself to be the best you can be with and around others. I trust. Still burps like this do escape.

Back to DSD, I just read an interesting post on the Trinity DAC thread over at AudioExotics by Dietmar - http://audioexotics.vanillaforums.com/discussion/10845/trinity-supreme/p33

- that points to many issues including recording engineering and post recording processing, abuse of DSD and PCM, variations in different recordings that interfere with such comarisons. This is why HiFI News began separately testing HD Tracks and other source materials for 'authenticity'.

As we know well, in quite a few cases the recordings were not as claimed...

Its a real pleasure to meet you all here.

 

[wisnon]

John57, who are you talking to/about?

 

[John57]

John57, who are you talking to/about?

Thanks Wisnon, I reposted to clarify...

 

[MarinJim]

John, sorry if you were offended, but I did not attack you personally. I just stated my opinion. Just because someone "likes" a particular gear , sample rate, etc., does not make it "better". And in this throw away and what is new is better culture, do not always believe what you read. And FTW, I am addicted to the best I can afford.

 

[John57]

John, sorry if you were offended, but I did not attack you personally. I just stated my opinion. Just because someone "likes" a particular gear , sample rate, etc., does not make it "better". And in this throw away and what is new is better culture, do not always believe what you read. And FTW, I am addicted to the best I can afford.

Sure Marin, thank you for your kind apology really. I look forward to hearing more from you about the reasons for your feelings, I am sure they have a solid bass line!

Bring on the audio...it moves me.

 

[Lee]

DSD simply sounds better than PCM. Next.

Agreed.

 

[MarinJim]

Agreed.

Do not agree. I am with Anthony Cordesman on this issue. Read his article on the Burmester MC151 in The Absolute Sound.