Debate of DAC design regarding DSD vs PCM among 5 VIPs

[louisxiawei]

I read one very interesting article from a Taiwan HIFI magazine a while ago.

 

http://www.pubu.com.tw/magazine/音響論壇電子雜誌-第339期-12月號-83923

 

The editor interviewed five famous designers for their opinions regarding DAC design in terms of DSD and PCM 

 

1. Larry Guliman from MSB
2. Andreas Koch from Playback
3. Florian Cossy from CH precision
4. Ted Smith from PS audio
5. Robert Watts from Chord

 

 

 

 

 

 

 

 

 

Till now, I realize that it might be a good idea to share it on CA and let everyone discuss further since I somehow disagree some of their argument points against DSD.

 

However, I feel that people on CA always tend to be honest and offer sensible reasons. So let the debate to be continued.

 

Since it's going to be a discussion thread, I try to avoid the first post to be lengthy. So I won't do the full translation of the article but list the bullet point of each interview's questions and the main general opinion of each designer.

 

I will do the translation bit by bit everyday, hopefully. Alternatively, I can prioritize some part of the translation based on most people's interest. Please tell me the subtitle number of these interview questions.

(Some questions are quite a common sense rather than meaningful ones)

 

English is not my native language, but I will try my best to do the proper translation and if anything confuses you, please let me know.

 

Here we go:

 

Interview with Larry Guliman from MSB

 

1. General opinion from Larry: multi-bit ladder DAC is the purist passive DAC, in theory, it does not have any upper limit in terms of DAC speed.

 

1.1 What is ladder DAC?

1.2 How does MSB find the high-precision resistor?

1.3 What's the difference between Sign Magnitude ladder R2R dac developped by MSB and other normal dac?

1.4 What's the advantage of MSB ladder dac?

1.5 Nowadays, many DACs accentuate on the ability of 32-bit processing ability, what's the processing ability of your MSB select dac (ladder dac)?

1.6 The jitter of your FemtoSecond Galaxy Clock is very at 33 fs. What's the impact of it on the real music playback?

1.7 How can the FemotoSecond Galaxy clock be controlled to this low?

1.8 Is the internal FemotoSecond Galaxy clock better than external atomic clocks?

1.9 Select DAC uses hybrid dac to support DSD, how does it work?

1.10 What's the improvement of your new USB module?

1.11 Why does MSB dac use Analog volume control?

 

Interview with Andreas Koch from playback

 

2. General opinion from Andreas: Rethink the nature of the DAC and how to completely solve the jitter problem 

 

2.1 Your playback DAC convert files no matter PCM or DSD into DSD first and then do the D-A conversion afterwards, do you think it is a better way to do this even for the PCM signal?

2.2 Many people think only 1 bit DSD D-A conversion is the genuine and correct way to work, what do you think?

2.3 How does your developed 2D DAC work?

2.4 Do you really solve the jitter problem completely?

2.5 Do you think the external ultra-precision atomic clock controller is very effecitve?

2.6 If you claim that you solved the jitter completely, does it mean that the signal will not be affected anymore by any digital cables and digital interfaces? 

 

Interview with Florian Cossy from CH Precision

 

3. General opinion from Florian: The advantage of multi-bit R2R DAC will be fully revealed during music signal playback

 

3.1 Why does C1 use muti-bit R2R modulation instead of normal Delta-sigma?

3.2 Why does C1 still use pcm 1704 DAC chip? That chip has been released for quite a long time.

3.3 Why do you use 4 PCM 1704 chips on each channel?

3.4 Do you use your own DSP to do the DSD to PCM conversion?

3.5 Does C1's volume control also use DSP digital control?

3.6 What's special of your C1's clock?

3.7 Which is a better interface on your C1? USB or Ethernet?

3.8 Could you please explain your three-level analogue output on C1 in detail?

 

Interview with Ted Smith from PS audio

 

4. General opinion from Ted: Only 1-bit DSD DAC can do the most linear, closest to the analogue D-A conversion

 

4.1 Before doing the D-A conversion on DirectStream DAC, no matter PCM or DSD signal, they are both being converted to the DSD first. Are you suggesting that DSD is superior to PCM?

4.2 Does it mean that the chips of some Delta sigma DAC are not 1 bit?

4.3 But many people think DSD dac has high-frequency problems, what's your opinion?

4.4 Are PCM signal being upsampled 10 times before D-A conversion?

4.5 It's quite difficult to do the DSD signal processing. However, DirectStream DAC uses digital volume control, can you explain how it works?

4.6 Don't you think the conversion between DSD and PCM will cause any distortion?

 

Interview with Robert Watts from Chord

 

5. General opinion from Robert: DSD format has its native flaw while the PCM's limit is on the D-A conversion part, which can be solved by proper design.

 

5.1 Many people think by using the way of interpolation to do the up-sampling will damage the original signal, but the Chord always uses very high up-sampling rate, what is the intention by doing this?

5.2 What's the advantage of  high-rate up-sampling?

5.3 What's so special about your Pulse Arrays DAC? Is it similar to Delta sigma modulation?

5.4 From Hugo to DAVE, your DACs all equipped with digital volume control, will it reduce the resolution at low volume?

5.5 Since Pulse Arrays DAC are multi-bit PCM DAC, are you suggesting that  PCM is superior to DSD?

5.6 How does Pulse Array process the DSD signal?

 

 

That's it for now.

[Speed Racer]

Not all that interesting unless you read whatever form of Chinese that is.....

[bibo01]

@louisxiawei, we wait for you to translate further

[louisxiawei]

22 minutes ago, bibo01 said:

@louisxiawei, we wait for you to translate further

Will do once I have enough spare time. It's going to take a while for the full translation. Guess I will start with Robert's interview. 

 

Or you can tell me the questions you are most interested so that I can do that bit translation first. 

 

 

[bibo01]

1 minute ago, louisxiawei said:

Will do once I have enough spare time. It's going to take a while for the full translation. Guess I will start with Robert's interview. 

 

Or you can tell me the questions you are most interested so that I can do that translation bit first. 

 

 

Fine.

Sorry, I forgot to thank you.

[Ralf11]

Thanks for taking the time on this.

 

I have a question about the multi-bit R2R DACs - seems like they were not in fashion for a while, but then mounted a come-back.  Has there been a recent improvement in getting the resistors in the ladder more accurate?

[louisxiawei]

14 minutes ago, Ralf11 said:

Thanks for taking the time on this.

 

I have a question about the multi-bit R2R DACs - seems like they were not in fashion for a while, but then mounted a come-back.  Has there been a recent improvement in getting the resistors in the ladder more accurate?

 

No idea about the improved resistors. Which R2R dac you are referring for a mounted come-back? 

 

I believe Aqua, Meturm and Rockna are quite popular in the R2R circle because they are relatively affordable compared to some R2R dacs. 

[Ralf11]

just in general; not specific brands

[jabbr]

21 minutes ago, Ralf11 said:

Thanks for taking the time on this.

 

I have a question about the multi-bit R2R DACs - seems like they were not in fashion for a while, but then mounted a come-back.  Has there been a recent improvement in getting the resistors in the ladder more accurate?

 

Yes, the Heisenberg principle is felt to be outdated and no longer applies...

 

I like question 4.5. Answer is trivial once you realize that 1 bit SDM math is the same as analog math, just in the frequency domain ... get it?  

[Nikhil]

Louis,

 

Thank you very much for sharing this discussion.  

 

Quote

Interview with Robert Watts from Chord
 5.3 What's so special about your Pulse Arrays DAC? Is it similar to Delta sigma modulation?

 

Would you please translate this section for me.  Thanks!

 

Regards.

[louisxiawei]

I will translate one question at once in one single post, begin with the question me/others think the most interesting.

 

Here we go: @Nikhil

 

5.3 What's so special about your Pulse Arrays DAC? Is it similar to Delta sigma modulation?

 

Pulse arrays DAC,  this split D-A conversion is different from delta sigma and multi-bit R2R modulation.

Two main different features:

 

1. Its characteristic of distortion is very "analogy". The smaller the signal is going to be converted, the distortion will be reduced more significantly close to 0. This is also close to our ears' characteristic and can more accurately present the sound stage and depth of the music.

 

(I'm quite confused about this ear part, no idea whether its the problem of editor's description or Rob's)

 

2. The DA conversion rate of Pulse arrays is constant. This feature make it immune to the effect of signal jitter, and naturally, will not lead to any harmonic distortion and background distortion caused by the jitter.

 

The number of Pulse Arrays DAC block has been increased to 20 compare to the Hugo which has only 4.

This makes Dave has higher resolution, more natural, smooth sound. The separation, image position focus are also much more accurate and  clearer.

 

These feature of Pulse Array DAC make our products immune to jitter while R2R and Delta Sigma modulations themselves are very sensitive to the jitter, which however might be solved by using high-precision atomic clock and improve the music playback performance.

 

 

 

P.S. I missed one interview question with Robert Watts. 

 

5.7 The word "tap" is mentioned quite often by Chord when talking about D-A conversion, what on earth is tap?

 

 

[audio.bill]

An excerpt from dspGuru describing FIR Filter Basics includes a basic definition of a 'TAP':

 

"Tap – A FIR “tap” is simply a coefficient/delay pair. The number of FIR taps, (often designated as “N”) is an indication of 1) the amount of memory required to implement the filter, 2) the number of calculations required, and 3) the amount of “filtering” the filter can do; in effect, more taps means more stopband attenuation, less ripple, narrower filters, etc."

 

[louisxiawei]

5.7 The word "tap" is mentioned quite often by Chord when talking about D-A conversion, what on earth is tap?

 

In general, the longer tap length will lead to the result like: digital filter will be more precised, time error of each digital sampling will be smaller. From the point of music playback, the start and the end time of each sound will be more accurate. In theory, the time error from listening experience will be totally eliminated

only if the tap has the length of 1,000,000, which cannot be be done according to current technology. The tap of normal digital signal has the length of 100, the top-class device is only 256 maximum, which is far away from the ideal tap length (1,000,000).

 

To solve this problem, I developed WTA (Watts Transient Aligned Filter) filtering technology, which is quite different from some conventional digital filtering. WTA's 256 Taps can considerably outperform Kaiser digital fitler's 2045 Taps.

 

In 1995, DAC 64 was released, the Tap had already reached 1024. Chord Dave uses LX75 version Spartan 6 FPGA chips, which improve the performance significantly and improve the Tap length to 164,000. Although it is still quite away from the ideal tap length in theory, it does significantly narrow the gap.

[Nikhil]

7 hours ago, louisxiawei said:

 

5.7 The word "tap" is mentioned quite often by Chord when talking about D-A conversion, what on earth is tap?

 

 

Thanks for adding this.  I have also wondered what the term "taps" meant.
I don't think I still understand it but it's a starting point. Need to read up on this some more.

 

Quote

In general, the longer tap length will lead to the result like: digital filter will be more precised, time error of each digital sampling will be smaller. From the point of music playback, the start and the end time of each sound will be more accurate. In theory, the time error from listening experience will be totally eliminated only if the tap has the length of 1,000,000, which cannot be be done according to current technology. The tap of normal digital signal has the length of 100, the top-class device is only 256 maximum, which is far away from the ideal tap length (1,000,000).

 

 

I'm not sure if this is a translation issue but I think that the correct word in English would be "number" instead of "length".  Chord always talks about the number of taps increasing the accuracy etc.  audio.bill's post also refers to number of (FIR) taps etc.  Can someone confirm this?

 

Regards.  

 

[louisxiawei]

18 minutes ago, Nikhil said:

 

Thanks for adding this.  I have also wondered what these taps meant.
I don't think I still understand it but it's a starting point. Need to read up on this some more.

 

Regards.  

 

Me neither. Either the editor over simplified Rob's words or Rob didn't explain in detail. 

 

I hope DSP experts like Miska will make some comment after I finish all the translation.

 

Quote

I'm not sure if this is a translation issue but it seems to me that the correct word in English would be "number" instead of "length".  Chord always talks about the number of taps increasing the accuracy etc.  Can someone confirm this?

 

I think you are right, I did a lousy translation. number is the correct word for tap. length for filter.

 

So tap number and filter length. Thanks for the correction. Please do point out if any translation you feel is somehow wrong. 

 

 

 

 

 

[rredline]

To be honest I'm not a fan of them. These people are selling DACs at exorbitant price where most people won't be able to access. They seldom reveal technical know how in depth to the public.

 

I'm more interested to hear something from someone like Mr Ed Meitner or Dustin Forman.

[asdf1000]

Hello friend and greatly appreciate your time and effort with this :-)

 

If you can help with the following that would be appreciated. I've owned both Ted Smith and Rob Watt's Dacs and always found how amazing the sound coming out of their analogue outputs sound, even with such different design philosophies! Much respect to all these designers.

 

Interview with Ted Smith from PS audio

4.1 Before doing the D-A conversion on DirectStream DAC, no matter PCM or DSD signal, they are both being converted to the DSD first. Are you suggesting that DSD is superior to PCM?

4.3 But many people think DSD dac has high-frequency problems, what's your opinion?

4.6 Don't you think the conversion between DSD and PCM will cause any distortion?

 

Interview with Robert Watts from Chord

5.3 What's so special about your Pulse Arrays DAC? Is it similar to Delta sigma modulation?

5.5 Since Pulse Arrays DAC are multi-bit PCM DAC, are you suggesting that  PCM is superior to DSD?

5.6 How does Pulse Array process the DSD signal?

[louisxiawei]

11 minutes ago, Em2016 said:

Hello friend and greatly appreciate your time and effort with this :-)

 

If you can help with the following that would be appreciated. I've owned both Ted Smith and Rob Watt's Dacs and always found how amazing the sound coming out of their analogue outputs sound, even with such different design philosophies! Much respect to all these designers.

 

 

Will do. You are good at picking questions. 

 

BTW: 5.3 has been translated 

 

 

[asdf1000]

2 hours ago, louisxiawei said:

Will do. You are good at picking questions. 

 

BTW: 5.3 has been translated 

 

 

 

:-) Yes I go straight to the important stuff ! :-)

 

Sorry, yes I missed 5.3.

 

Thanks again for this thread.

[louisxiawei]

5.5 Since Pulse Arrays DAC are multi-bit PCM DAC, are you suggesting that PCM is superior to DSD?

 

In fact, the dac chip I designed in early time was for the architecture of DSD. At that time, DSD technology has the advantage of musicality that some other D-A approaches lack.

 

However, during my research through time, I begin to realise that the DSD format has its inborn flaw, which must reply on noise shaping to eliminate the noise. This will cause the loss of tiny signal.

 

Furthermore, the jitter of DSD is also very sensitive. From the point of listening experience, this will give you a very flat soundstage. For PCM, although the D-A part architecture is more complex, it can re-build the intact D-A signal only if with the proper design.

 

The key point is that DSD will cause non-linear time error that already existed during DSD recording and it cannot be solved by any DSP. From the perspective of listening experience, DSD is difficult to reproduce instantaneous transient of the music.

 

In conclusion, DSD has its nature flaw while PCM’s limit is on the D-A part. The former (DSD) has no solution to tackle with while the latter (PCM) can be solved with proper design.

[louisxiawei]

5.6 How does Pulse Array process the DSD signal?

Dave has its newly added feature called DSD plus, which converts DSD to PCM. However, it does not pass through the Decimation for the re-sampling so that it can contain the DSD information as much as possible. Dave also supports transmitting Native DSD signal to meet the DSD fans’ need.

[audiventory]

On 11.04.2017 at 11:07 PM, louisxiawei said:

1. Its characteristic of distortion is very "analogy". The smaller the signal is going to be converted, the distortion will be reduced more significantly close to 0. This is also close to our ears' characteristic and can more accurately present the sound stage and depth of the music.

 

(I'm quite confused about this ear part, no idea whether its the problem of editor's description or Rob's)

 

Hi Louis,

Thank you for translation.

 

I also don't sure that exists clear link between "stage and depth of the music" and distortions.

 

Because I don't know what does "stage and depth of the music" mean in technical terms.

In music production "wider stage" or "more space" may be achieved with reverberation as example.

 

Also I don't know, what does terms "analog distortions" and "digital distortions" mean exactly.

Though digital distortions have exact feature - aliases (mirrored base audio spectrum thru frequency range).

[semente]

On 11/04/2017 at 9:07 PM, louisxiawei said:

I will translate one question at once in one single post, begin with the question me/others think the most interesting.

 

Here we go: @Nikhil

 

5.3 What's so special about your Pulse Arrays DAC? Is it similar to Delta sigma modulation?

 

Pulse arrays DAC,  this split D-A conversion is different from delta sigma and multi-bit R2R modulation.

Two main different features:

 

1. Its characteristic of distortion is very "analogy". The smaller the signal is going to be converted, the distortion will be reduced more significantly close to 0. This is also close to our ears' characteristic and can more accurately present the sound stage and depth of the music.

 

(I'm quite confused about this ear part, no idea whether its the problem of editor's description or Rob's)

 

2. The DA conversion rate of Pulse arrays is constant. This feature make it immune to the effect of signal jitter, and naturally, will not lead to any harmonic distortion and background distortion caused by the jitter.

 

The number of Pulse Arrays DAC block has been increased to 20 compare to the Hugo which has only 4.

This makes Dave has higher resolution, more natural, smooth sound. The separation, image position focus are also much more accurate and  clearer.

 

These feature of Pulse Array DAC make our products immune to jitter while R2R and Delta Sigma modulations themselves are very sensitive to the jitter, which however might be solved by using high-precision atomic clock and improve the music playback performance.

 

 

 

P.S. I missed one interview question with Robert Watts. 

 

5.7 The word "tap" is mentioned quite often by Chord when talking about D-A conversion, what on earth is tap?

 

 

 

Watts talk s a bit about taps here:

 

http://www.the-ear.net/how-to/rob-watts-chord-mojo-tech

 

R

[semente]

 

12 minutes ago, audiventory said:

 

Hi Louis,

Thank you for translation.

 

I also don't sure that exists clear link between "stage and depth of the music" and distortions.

 

Because I don't know what does "stage and depth of the music" mean in technical terms.

In music production "wider stage" or "more space" may be achieved with reverberation as example.

 

Also I don't know, what does terms "analog distortions" and "digital distortions" mean exactly.

Though digital distortions have exact feature - aliases (mirrored base audio spectrum thru frequency range).

 

I think he probably means forms of distortion that affect the recreation of the spatial effects you find in a recording, e.g. higher noise floor will reduce or mask the decay of instruments.

 

R

[rredline]

On 4/13/2017 at 7:10 AM, louisxiawei said:

The key point is that DSD will cause non-linear time error that already existed during DSD recording...

 

 

Disagree

If DSD recording causes non-linear time error, PCM will too, because that PCM is came from DSD after decimation filter applied.