Multi-bit DSD versus PCM

[esldude]

These discussions are no different than discussions over how many angels can fit on the head of a pin.

Please, please, be a bit more learned.

The question was how many Angels can dance on the head of a pin. Fit is a static thing. Dancing implies dynamic action which might even have involved music. The number on the pin and how many can dance are two very different situations. Dancing implies interaction with the pin over time.

So, in modern terms which digital sampling representation can most accurately track the true essence of Angelic dancing?

 

[Blizzard]

So, in modern terms which digital sampling representation can most accurately track the true essence of Angelic dancing?

Not the one that uses the crappiest SDM/SRC algorithms that's for sure.

 

[Bruce B]

There is no reason to rename anything. PCM allows any sample rate or bit depth. It is of no need of new name just because someone uses it at non-standard bit depths like 5 bits.

I thought it was called PDM (Pulse Density Modulation) vs. PWD or Pulse Width Modulation

DSD Wide or PCM Narrow

 

[Blizzard]

I wished I were more convinced all of this sounds better. Or even perceptibly different.

I have used HQplayer as a trial. It does do excellent filtering. Even using it with a PCM DAC that tops out at 192khz, his filtering gave measurably superior results at the analog out when playing redbook files. The areas of superiority were small, but real. I don't have a DSD option, but I don't doubt it would perform excellently as he says.

Despite all of that I didn't hear a difference. Even old plain PCM can be superbly good. Convince me we aren't gilding the lily here.

Upsampling redbook to 24/192 is better, but not near as good as when you upsample to DSD. Especially quad DSD. This is because, as I explained in my last post, you bypass all of the inferior modulators/filters in the DAC chip. So these DAC manufacturers can argue till their blue in the face about PCM being better, but if they use a modern SDM chip, like 90% do, they are out to lunch. And the R2R resurgence fad will soon be over with the latest crop of SDM chips unveiled this year like the AKM AK4497 and ESS 9038pro. That outdated tech can finally be put to rest.

 

[c1ferrari]

And the R2R resurgence fad will soon be over with the latest crop of SDM chips unveiled this year like the AKM AK4497 and ESS 9038pro. That outdated tech can finally be put to rest.

That would be so cool.
I'd like right of first refusal on all the obsolete master tape protection copies

 

[Blizzard]

That would be so cool.
I'd like right of first refusal on all the obsolete master tape protection copies

Ha ha, no I was talking about R2R DAC's. Not R2R analog tape

Into the grinder they go



Time to make way for the new kings of digital

 

[LarsS]

Slightly off-topic.

Been sponsoring a crowd funding project now for three years with hopefully a delivery of the "God Machine" (DAC, ADC, Preamp, DSP, Phono) during this year.

Designer is a hard core advocate of Native/Pure DSD (no PCM resampling whatsoever). The God machine will be able to record from vinyl to Pure DSD as well as playing Native DSD (no PCM resampling in chain) & PCM. Specs in chart below.

 

[Purite Audio]

Slightly off-topic.

Been sponsoring a crowd funding project now for three years with hopefully a delivery of the "God Machine" (DAC, ADC, Preamp, DSP, Phono) during this year.

Designer is a hard core advocate of Native/Pure DSD (no PCM resampling whatsoever). The God machine will be able to record from vinyl to Pure DSD as well as playing Native DSD (no PCM resampling in chain) & PCM. Specs in chart below.


View attachment 25475

I wonder whether the designer will actually deliver any product, I sincerely he hope he does for all the investors.
I hope it isn't just a giant 'Ponzi ' scheme.
Keith.

 

[Fitzcaraldo215]

Okay, please explain how it's possible for DSD to be compatible with Meridian digital systems?

You are explaining more misconceptions as well. Did you even consider that the hardware/software they used at the time may of had an influence on their impressions? All of the ADC's and mastering packages sound different. And this has nothing to do with the formats.

Have you tried resampling all of your PCM with HQplayer yet? This might change your mind if you try.

And lastly I already explained a few posts back that HQplayer can process all of this DSP on a DSD stream.

As I said, Meridian products could be or could have been designed around a DSD-capable DAC chip. That they are not and were were not was a design choice made by Stuart. If he thought it were important, he could always release new models that did so, as well as supporting PCM, as now. There is nothing really radically new as to those formats. He is not forever locked in to what he did in the past. He consciously chose to be for his own reasons, and he still chooses that approach. But, time moves on and new equipment models come and go. Many others agree with him, many do not.

Of course, specific implementations may or may not achieve the ultimate from their respective DSD or PCM recorded formats. Some recordists swear by one or another format based on their experience. Their choice might have some sonic implications in the quality of the recorded result. But, there are a host of other choices and techniques in engineering a recording beyond just recorded format. I think most of those other engineering factors swamp the DSD vs. PCM choice in determining the sonic quality of the result.

I have many excellent recordings done in each format. I hear no pervasive pattern consistently favoring one format over the other in playing them in hi res in their native format/sampling rate. I am not anti-DSD. I am for the best sounding recordings. Some are DSD, some are hi rez PCM.

No, Mr. Fan Boy, I have not tried HQPlayer to DSD. I rechecked their features list, and I still do not see bass management, in addition to many others I absolutely need in my preamp/processor-less system setup, with all control in PC software. It may indeed sound as great as you keep saying over and over, but it is still quite short of my requirements.

Virtually all the glowing endorsements of HQP I have seen are about upsampling and DSD conversion of RBCD recordings. HQP may indeed shine in that role. But, well over 99% of my listening is purely to native hi rez recordings, in Mch, BTW. Yeah, HQP handles Mch up to a point. But, I am not really excited about trying it.

 

[Blizzard]

As I said, Meridian products could be or could have been designed around a DSD-capable DAC chip. That they are not and were were not was a design choice made by Stuart. If he thought it were important, he could always release new models that did so, as well as supporting PCM, as now. There is nothing really radically new as to those formats. He is not forever locked in to what he did in the past. He consciously chose to be for his own reasons, and he still chooses that approach. But, time moves on and new equipment models come and go. Many others agree with him, many do not.

Of course, specific implementations may or may not achieve the ultimate from their respective DSD or PCM recorded formats. Some recordists swear by one or another format based on their experience. Their choice might have some sonic implications in the quality of the recorded result. But, there are a host of other choices and techniques in engineering a recording beyond just recorded format. I think most of those other engineering factors swamp the DSD vs. PCM choice in determining the sonic quality of the result.

I have many excellent recordings done in each format. I hear no pervasive pattern consistently favoring one format over the other in playing them in hi res in their native format/sampling rate. I am not anti-DSD. I am for the best sounding recordings. Some are DSD, some are hi rez PCM.

No, Mr. Fan Boy, I have not tried HQPlayer to DSD. I rechecked their features list, and I still do not see bass management, in addition to many others I absolutely need in my preamp/processor-less system setup, with all control in PC software. It may indeed sound as great as you keep saying over and over, but it is still quite short of my requirements.

Virtually all the glowing endorsements of HQP I have seen are about upsampling and DSD conversion of RBCD recordings. HQP may indeed shine in that role. But, well over 99% of my listening is purely to native hi rez recordings, in Mch, BTW. Yeah, HQP handles Mch up to a point. But, I am not really excited about trying it.

The problem isn't the DAC chip being DSD capable. The problem is their DSP chips, and networking system isn't DSD compatible. It's impossible to use DSD with the way they do things. It's simply not an option. This has nothing to do with DSD being better or worse, or his feelings about DSD. It has everything to do with his DSP chips and networking system being 100% DSD incompatible with no hope anywhere in sight with anything changing in the near future.

As for the rest of what you said, you must have skipped through the rest of the content that was said on this thread, or you wouldn't have said it. If you did read it, you didn't absorb it that's for sure.

 

[Fitzcaraldo215]

The problem isn't the DAC chip being DSD capable. The problem is their DSP chips, and networking system isn't DSD compatible. It's impossible to use DSD with the way they do things. It's simply not an option. This has nothing to do with DSD being better or worse, or his feelings about DSD. It has everything to do with his DSP chips and networking system being 100% DSD incompatible with no hope anywhere in sight with anything changing in the near future.

As for the rest of what you said, you must have skipped through the rest of the content that was said on this thread, or you wouldn't have said it. If you did read it, you didn't absorb it that's for sure.

Yes, well, Meridian is quite far from being alone in applying DSP using PCM. The DSP chip capabilities just are not there for DSD, and prospects for their emergence are quite poor, given the small, niche market status of DSD.

Nope, I have read every word of the discourse here. Who is not absorbing what is said? I hear what you have said, oft repeatedly. I just do not agree. Does absorbing mean I must agree with you or anyone else?

 

[amirm]

Here's a great example to clarify things. This is the block diagram of the new AKM AK4497:

View attachment 25471

You can see on the left where the PCM data interface and the DSD data interface is.

Let's follow the data path of the PCM first:

First it goes into a "soft mute" this is to eliminate pops and clicks when the internal registers are accessed. From there it goes into the oversampling interpolator where the resource constrained filters are applied. Then to the modulator. Inside the modulator, the data is converted into multibit SDM. It uses the best modulators they could come up with based on the limited power of such a tiny little chip. This is where the digital volume control can be applied. then from there to a filter, and then the analog outs.

Now lets look at what happens when you send it DSD. With DSD there's 2 different options. You can send it through the DSD filter, soft mute, multibit SDM modulator if you want. Or you can bypass all of that stuff. Option 2 completely bypasses all of the resourced constrained SRC/SDM inside the chip altogether, as well as the multibit conversion. It's by far the purest path in the chip. So if you use HQplayer to handle the SDM/SRC with it's far superior SDM/SRC algorithms, you can send all of the data through option 2 of the chip. There's no better way to do it.

This is not a proper comparison. This DAC is designed to handle dual formats: DSD and PCM. So its output stage by definition is sigma-delta and that is how it outputs the PCM with the extra modulator.

Even here, there is no problem with PCM since it achieves noise floors better than 120 dbFS from what I recall. The best dynamic range of the ear is about 116 dbFS.

The other consideration is what comes *into* the DAC. Your analysis assumes they are of equal fidelity. THat is not the case. We can demonstrate conclusively that PCM is mathematically correct format to represent the input signal. Same math shows that we cannot do that with 1-bit DSD. Due to use of feedback there and lack of ability to dither a single bit to eliminate quantization noise, that format comes with less assurance of fidelity than PCM. Any problems there remain regardless of what your DAC does.

 

[amirm]

The problem isn't the DAC chip being DSD capable. The problem is their DSP chips, and networking system isn't DSD compatible. It's impossible to use DSD with the way they do things. It's simply not an option. This has nothing to do with DSD being better or worse, or his feelings about DSD. It has everything to do with his DSP chips and networking system being 100% DSD incompatible with no hope anywhere in sight with anything changing in the near future.

It has nothing to do with "hope." As mentioned, you could wake up tomorrow and see new products from them that support DSD. The networking stack by the way doesn't care what the bits are. It is moving data whether it is metadata, PCM samples or in this discussion, DSD.

As for the rest of what you said, you must have skipped through the rest of the content that was said on this thread, or you wouldn't have said it. If you did read it, you didn't absorb it that's for sure.

Please be more cordial Mike. I have read the thread and see nothing that he has missed. He saying Meridian has chosen not to support DSD not out of some inherent difficulty in implementing it.

That said, their active speaker filters would not work in DSD domain. They would have to convert the bits to PCM and while at it, why go back to DSD??? Same point I raised in my first post regarding HQ having to do the same for any bass or EQ processing.

 

[Blizzard]

It has nothing to do with "hope." As mentioned, you could wake up tomorrow and see new products from them that support DSD. The networking stack by the way doesn't care what the bits are. It is moving data whether it is metadata, PCM samples or in this discussion, DSD.


Please be more cordial Mike. I have read the thread and see nothing that he has missed. He saying Meridian has chosen not to support DSD not out of some inherent difficulty in implementing it.

That said, their active speaker filters would not work in DSD domain. They would have to convert the bits to PCM and while at it, why go back to DSD??? Same point I raised in my first post regarding HQ having to do the same for any bass or EQ processing.

Meridian's DSP and and networking system is simply not compatible. They would have to be able to do the multibit DSD conversion for DSP like the DAC chips and HQplayer does for this to be possible. Even with HQplayer, you need like a 12 core Intel Xeon processor to be able to handle 8+ channels of DSD DSP simultaneously. With a DSP chip, not a chance in hell. And even if they could, their networking protocol maxes out at 24/192 PCM. It's just not possible to do DSD with their system. I have been researching this for years now. I have talked to engineers working on Linn's next gen DSP networked based system that the public won't see for a while, and the same deal with them. They use a very similar approach.

And I did explain why DSD is superior if you are using an modern SDM chip in your DAC. If his DAC is a PCM R2R type, then PCM is superior simply due to it not being DSD compatible.

 

[Blizzard]

Yes, well, Meridian is quite far from being alone in applying DSP using PCM. The DSP chip capabilities just are not there for DSD, and prospects for their emergence are quite poor, given the small, niche market status of DSD.

Nope, I have read every word of the discourse here. Who is not absorbing what is said? I hear what you have said, oft repeatedly. I just do not agree. Does absorbing mean I must agree with you or anyone else?

You say you understand how modern SDM DAC chips work, and you never even tried Hqplayer, but you don't agree that it's a better way to do things.

Do you not see a problem of why it would be hard to put much weight on your opinion of this matter?

 

[Blizzard]

This is not a proper comparison. This DAC is designed to handle dual formats: DSD and PCM. So its output stage by definition is sigma-delta and that is how it outputs the PCM with the extra modulator.

Even here, there is no problem with PCM since it achieves noise floors better than 120 dbFS from what I recall. The best dynamic range of the ear is about 116 dbFS.

The other consideration is what comes *into* the DAC. Your analysis assumes they are of equal fidelity. THat is not the case. We can demonstrate conclusively that PCM is mathematically correct format to represent the input signal. Same math shows that we cannot do that with 1-bit DSD. Due to use of feedback there and lack of ability to dither a single bit to eliminate quantization noise, that format comes with less assurance of fidelity than PCM. Any problems there remain regardless of what your DAC does.

Okay which other type of DAC chip configurations are there out there? Are you saying the circa 1998 BB1704 R2R chip is superior technology? It was the last of the R2R chips.

It has nothing to do with the noise floor/dynamic range. It's the algorithms for the modulators and filters that are the bottleneck.

About 98% on the market today use SDM, and are very similar to how this block diagram is. Only some don't have the DSDdirect option.

And as far as the format before we get to the modulation/filtering stages of the DAC, that's irrelevant to this issue. What's relevant is how the DAC handles the information once it receives it. This is the point that most people miss. You are arguing about what doesn't matter in the big picture. I'm talking about what does matter.

 

[Blizzard]

Okay which other type of DAC chip configurations are there out there? Are you saying the circa 1998 BB1704 R2R chip is superior technology? It was the last of the R2R chips.

It has nothing to do with the noise floor/dynamic range. It's the algorithms for the modulators and filters that are the bottleneck.

About 98% on the market today use SDM, and are very similar to how this block diagram is. Only some don't have the DSDdirect option.

And as far as the format before we get to the modulation/filtering stages of the DAC, that's irrelevant to this issue. What's relevant is how the DAC handles the information once it receives it. This is the point that most people miss. You are arguing about what doesn't matter in the big picture. I'm talking about what does matter.

The link Bob shared earlier about the Ayre PCM vs DSD comparison explains the difference that matters clearly. This isn't anything up for debate here among anyone who knows what's going on in the industry.

"If you are fortunate enough to own either an Ayre DX-5-DSD or Ayre QB-9-DSD, then the same is true for the playback also. There are no differences in the signal path whatsoever except for the algorithms used in the ESS ES9016 Digital-to-Analog converter chip. There is no way to make a more fair comparison."


http://www.ayre.com/insights_dsdvspcm.htm

When you listen to the DSD version, you are listening to the algorithms that the ADC used for the DSD. When you listen to the PCM version, you are listening to the algorithms the DAC chip uses to apply the SDM/SRC to the PCM. The winner will be whoever used the superior algorithms. The mastering house at the studio, or the resources constrained DAC chip. Chances are a guy like Bruce B with his Horus ADC and Pyramix software has better algorithms to use at his disposal than are built into the $10 resource constrained DAC chip.

 

[amirm]

And as far as the format before we get to the modulation/filtering stages of the DAC, that's irrelevant to this issue. What's relevant is how the DAC handles the information once it receives it. This is the point that most people miss. You are arguing about what doesn't matter in the big picture.

Not at all. This is a system from A/D to D/A. You cannot just look at half of it. And that is what matters, not a simplistic view of just looking at half the reproduction chain.

I'm talking about what does matter.

I have not seen that analysis either. What matters is what we hear, right? Where is the analysis that shows the audibility of any of this?

 

[amirm]

It has nothing to do with the noise floor/dynamic range. It's the algorithms for the modulators and filters that are the bottleneck.

Have not seen any analysis demonstrating this. The only point is made that the PC version of the same is better. Even that has not been quantified let alone shown to be an audible bottleneck.

The analysis again also ignores inbuilt distortion in DSD conversion at the start of the capture.

 

[Blizzard]

Not at all. This is a system from A/D to D/A. You cannot just look at half of it. And that is what matters, not a simplistic view of just looking at half the reproduction chain.


I have not seen that analysis either. What matters is what we hear, right? Where is the analysis that shows the audibility of any of this?

Do you think guys like Ayre who make both SOTA ADC's and DAC's don't understand how things work?