MDAC First Listen (part 00110101)

[JohnW]

Mqa, Ill take my digital files vanilla with no drm or watermarkING. Frankly I have no interest in it at all.

So far they've proven no Sq increase from it, so I'll pass.

Well I'd like to see some End to End impulse response of the MQA system if they have optimised the Time domain with No pre and post ringing then it solves the major problem with PCM digital.

Trouble is - even as a designer, I've seen no such impulse results so I HATE working in the blind like this - basically driven / forced by "market demands" of those who also have no idea!

 

[mattgbell]

Matt,

Each sponsor of the FDAC / MDAC2 development does so on his own accord - they understand the risk and agree to the terms. What they then decide to do with there position is there own business - but I hope they will respect the community. [...]

John,
Thanks for the clarification. I had assumed that one reason for keeping the list of sellers was to prevent this thread from becoming a marketplace. On the contrary, you've now effectively acknowledged that positions can be bought and sold on this thread, as well as via the list.
Matt

 

[JohnW]

Matt,

Well I hope it will not become an open market place here - with the production orders opening in a couple of months it be better to wait until demand naturally picks up.

As I say we will endeavour to transfer all the "unwanted" production slots before offering units for sale at market rates (as opposed to the discounted sponsored "At Cost" pricing).

Financially its better resell a physical unit at market rates then dropping out and trying to sell the production slot - you then also get a chance to try the DAC

The number of MDAC2's will be limited as we dont intend to consider continuous Mass Production... I'm happy to finally have my own reference DAC and for Friends.... I don't really want to get involved with supporting multiple production batches - I need to move onto VFET amp then finally FDAC with all the feedback and lessons learnt from MDAC2.

 

[BigDog]

I agree - I dont feel comfortable about the whole MQA business, but I also have to keep an open mind – atleast to be able to support MQA if there is a market demand.

MQA could really better explain what MQA does – my overriding feeling is that’s there was some honest intent to improve the sound quality by concentrating on the Time Domain aspects of the digital reproduction (VERY VERY IMPORTANT) – but this has been used to front a commercial operation packaged with “DRM” to convince the record company’s to release “Studio Master” quality recordings.

Sad its turned out like this – but is this my battle? IF and it’s a big IF we gain access to higher quality recordings maybe it’s the price we need to pay and accept we have sold our soul to a devil and move on… I only wish Native DSD was more feasible…

Here are 2 posts by Archimago on the blogspot.ca website discussing MQA. I find his Musings on various audio topics to be based on unbiased measurement, a characteristic relatively rare in our hobby. Definitely, IMHO, worth a look-see.

<http://archimago.blogspot.ca/2017/02/comparison-hardware-decoded-mqa-using.html>

And a follow up...

http://archimago.blogspot.ca/2017/02/musings-discussion-on-mqa-filter-and.html

 

[timola]

Oh

Here are 2 posts by Archimago on the blogspot.ca website discussing MQA. I find his Musings on various audio topics to be based on unbiased measurement, a characteristic relatively rare in our hobby. Definitely, IMHO, worth a look-see.

<http://archimago.blogspot.ca/2017/02/comparison-hardware-decoded-mqa-using.html>

And a follow up...

http://archimago.blogspot.ca/2017/02/musings-discussion-on-mqa-filter-and.html

Interesting.... Thanks, I took the bottom line of the first article to be MQA is a neat solution, but SQ wise its no big shakes on high res file playback.....and on the second article about time domain fix in MQA...its nice but insignificant compared to other factors.....did I get the gist about right?

 

[JohnW]

BigDog,

I read some of "archimago" postings and as a result dont give him much time as he blankly discounts certain aspects of Digital Audio that I do know have a significant impact on audio quality.

Its one thing to "repeat" theory without practical experience on the impact to sound quality - its another to actually have some working practical experience where you become forced to question your own preconceived ideas and biases – I guess this happens in any field of research though practical application.

Heres a decent article on whats good about MQA (concentrating on Time domain performance).

http://www.soundonsound.com/techniques/mqa-time-domain-accuracy-digital-audio-quality

When I returned to Europe just after the Hong Kong handover FPGA started to become powerful and affordable enough to allow experimentation with Digital filters and modulators – the Optimal transient filter is a descendant of this early work – where we emphasis time domain performance above all over the frequency domain – this is why despite my “discomfort / disgust” of the DRM aspects of MQA I know that there underlying focus on PCM’s time domain is absolutely the correct thing to be doing.

Heres the standard “Fast” Dac filter typical in 99% of DAC’s:-

https://dl.dropboxusercontent.com/u/86116171/Fast FIR.jpg

Heres standard Minimum Phase:-

https://dl.dropboxusercontent.com/u/86116171/Minimum Phase Fast.jpg

Our custom Minimum Phase:-

https://dl.dropboxusercontent.com/u/86116171/Minimum Phase Custom.jpg

And my one and only listening filter (Optimal Transient):-

https://dl.dropboxusercontent.com/u/86116171/Optimal Transiant Custom.jpg

The beauty of MQA is that they can optimise the end to end time domain performance of the whole reproduction system upto the DAC's output (starting from the ADC) where we can only optimise the DAC’s performance.

 

[BigDog]

Oh

Interesting.... Thanks, I took the bottom line of the first article to be MQA is a neat solution, but SQ wise its no big shakes on high res file playback.....and on the second article about time domain fix in MQA...its nice but insignificant compared to other factors.....did I get the gist about right?

I had a similar takeaway, timola...that Archimago is not excited about MQA as a format.

My personal opinion is that IF MQA offers some perceived value to those who stream audio via Tidal, etc., then they should go for it. I for one will not be investing in MQA replacements for anything in my existing library nor will I be buying new releases which require an MQA-enabled DAC for playback. In fact I am trying to decide if I will upgrade my Mytec Manhattan I DAC to the Manhattan II when one of the sale points is the incorporation of MQA decoding. The touted improvement in SQ attributed to the DAC chip upgrade will have to be significant for me to justify the expenditure.

I HAVE gotten on the HiRes train only where provenance warrants but I am not replacing any significant portion of my existing CD or SACD/DSD library. I will have to say that 4 tracks I have of 2 acoustic guitarists/vocalists that were recorded live at the Rocky Mountain Audio Fest a couple of years ago on a Sony portable digital recorder in DSD64 are pretty amazing. The tracks were obtained unmassaged...no mixing, no mastering, no messing, just straight up.

 

[Julf]

I read some of "archimago" postings and as a result dont give him much time as he blankly discounts certain aspects of Digital Audio that I do know have a significant impact on audio quality.

He does measurements. He does listening. He reports on the results.

 

[DANOFDANGER]

BigDog,






When I returned to Europe just after the Hong Kong handover FPGA started to become powerful and affordable enough to allow experimentation with Digital filters and modulators – the Optimal transient filter is a descendant of this early work – where we emphasis time domain performance above all over the frequency domain – this is why despite my “discomfort / disgust” of the DRM aspects of MQA I know that there underlying focus on PCM’s time domain is absolutely the correct thing to be doing.

Heres the standard “Fast” Dac filter typical in 99% of DAC’s:-

https://dl.dropboxusercontent.com/u/86116171/Fast FIR.jpg

Heres standard Minimum Phase:-

https://dl.dropboxusercontent.com/u/86116171/Minimum Phase Fast.jpg

Our custom Minimum Phase:-

https://dl.dropboxusercontent.com/u/86116171/Minimum Phase Custom.jpg

And my one and only listening filter (Optimal Transient):-

https://dl.dropboxusercontent.com/u/86116171/Optimal Transiant Custom.jpg

The beauty of MQA is that they can optimise the end to end time domain performance of the whole reproduction system upto the DAC's output (starting from the ADC) where we can only optimise the DAC’s performance.

How is it that ringing artifacts actually affect sound ? As its been made aware to me, ringing is not impacted in our band of interest. Do you have any theorys or experience how this appears to affect the output.

An experience with minimum phase has put me off any non linear filter, something about it just seems off. It does however give you a good example of how pre ringing affects the output.

 

[ChrisPa]

The beauty of MQA is that they can optimise the end to end time domain performance of the whole reproduction system upto the DAC's output (starting from the ADC) where we can only optimise the DAC’s performance.

Sounds like the ideal compromise to me.
They provide the information on and compensation for the recording chain.

As I understand it (but will happily be corrected, and I don't want to divert this into an MQA thread) MQA core provides high resolution data, compensates for the ADC chain, is "preconditioned for generic DACs", and access is allowed to the digital output stream

3 of those are ideal - the decoded stream can then be processed (DSP) for filtering, digital crossover, room correction etc.

The unknown/unwanted is "preconditioned for generic DACs" and question is how to "uncompensate" for this to get access to the "ADC corrected" stream

 

[JohnW]

How is it that ringing artifacts actually affect sound ? As its been made aware to me, ringing is not impacted in our band of interest. Do you have any theorys or experience how this appears to affect the output.

Changes in Time domain are clearly audible, but wither this can be put down to differences in Pre or post ringing I cannot say as they mutually effect each other and you cannot say "which " effect your hearing...

Its maybe easier to start with a simpler "analogue" system - varying the value of the classic Cdom compensation capacitor in any Gain stage - lower phase margin results in under damping (post ringing) and too much phase margin results in an under damped response. Whats REALLY surprising is how VERY VERY sensitive our auditory system is to minute changes in Phase Margin = Step Response.

So with this in mind, its understandable to be likewise concerned about the rather complex and extended time domain ringing measured with digital filtering.

What I can say is that when design steps are taken to reduce pre / post ringing the system sounds far more natural / less digital so maybe I'm jumping too fast to a conclusion, but I put the sonic deficiencies of PCM digital down to these time domain artefacts.. A fact that is further borne out when the SAME ADC frontend and DAC backend is used in DSD mode and the system now sounds "transparent.

An experience with minimum phase has put me off any non linear filter, something about it just seems off. It does however give you a good example of how pre ringing affects the output.

Well all MDAC filters sound pretty much the same to me (small "audiophile" differences) APART from the Optimal Transient types... I too don't like the minimum phase filters.

 

[JohnW]

The unknown/unwanted is "preconditioned for generic DACs" and question is how to "uncompensate" for this to get access to the "ADC corrected" stream

Its the job of the MQA "render" block after the MQA decoder to then optimise the DAC / analogue stage... MQA need to tune the renderer block for each DAC type and this can also be extended to include the analogue domain.

I'd not say that the MQA data is Pre-Coded for any particular DAC type - this would not make sense, but what does seem sensible is to tune the post decoder Render block to achieve an overall system time domain response - or as close as possible

 

[delange]

He does measurements. He does listening. He reports on the results.

Totaly agree with that.
John's response is one of the many reasons why I lost interest in the MDAC / FDAC project. While I admire his quest for perfection, he is chasing ghosts. I believe he is trying to improve things that are well beyond the capability of human hearing.
Archimago is a realist, he looks at things with an open mind, does measurements and listening tests. And as Julf says: his results are based on those facts.

 

[DANOFDANGER]

Changes in Time domain are clearly audible, but wether this can be put down to differences in Pre or post ringing I cannot say as they mutually effect each other and you cannot say "which " effect your hearing...

It sounds like that our equipment and techniques today cannot "directly" measure what we are hearing, not that there is nothing to find but that we cannot currently examine what we are interpreting.
Have you ever found any direct evidence in the fourier domain that shows the impact of time domain accuracy. I can only assume not due to how how small the energy is. But it must be there, unseen.

Its maybe easier to start with a simpler "analogue" system - varying the value of the classic Cdom compensation capacitor in any Gain stage - lower phase margin results in under damping (post ringing) and too much phase margin results in an under damped response. Whats REALLY surprising is how VERY VERY sensitive our auditory system is to minute changes in Phase Margin = Step Response.

Yes the indeed accuracy of the hearing system is amazing. It shows it in some ways how more sensitive they are compared to modern analyzers. You almost have to invision the ear as a fourier analyzer itself.

What I can say is that when design steps are taken to reduce pre / post ringing the system sounds far more natural / less digital so maybe I'm jumping too fast to a conclusion, but I put the sonic deficiencies of PCM digital down to these time domain artefacts.. A fact that is further borne out when the SAME ADC frontend and DAC backend is used in DSD mode and the system now sounds "transparent.

You are very held back with definitive conclusions which is the best outlook in this feild tbh. Its very much like biological evolution, its evidence is everywhere but we cannot create a test to prove its existence.

With a wide interest in time domain developement all over the board, there shows a narrowed suspicion with the time domain in pcm.

Anyway thanks, keeping me intrigued.

 

[radamel]

Totaly agree with that.
John's response is one of the many reasons why I lost interest in the MDAC / FDAC project. While I admire his quest for perfection, he is chasing ghosts. I believe he is trying to improve things that are well beyond the capability of human hearing.
.

Chasing ghosts, then?
How do you know that the time domain issues JohnW is trying to improve are "well beyond the capability of human hearing"?

 

[dtd]

Totaly agree with that.
John's response is one of the many reasons why I lost interest in the MDAC / FDAC project. While I admire his quest for perfection, he is chasing ghosts. I believe he is trying to improve things that are well beyond the capability of human hearing.
Archimago is a realist, he looks at things with an open mind, does measurements and listening tests. And as Julf says: his results are based on those facts.

Archimago didn't even cover anything about time domain, whatsoever. Incredibly simplisitic view.

I suggest if possible you have a listen to speakers that really put an onus on time domain performance. Eye opener to say the least.

 

[sq225917]

There's a an awful lot of cart before horse recently. Someone needs to step back and prove there are audible differences between level matched pcm and DSD under blind conditions before jumping to conclusions as to why one sounds better than the other.

Similarly, mqa is just a marketing vehicle to deliver drm by the back door for a generation who don't own their own music, they just rent it. Like hdcd it'll be a feint footnote in the history of digital audio in a year or two

 

[radamel]

Similarly, mqa is just a marketing vehicle to deliver drm by the back door for a generation who don't own their own music, they just rent it.

How is MQA's DRM implemented exactly?

 

[JohnW]

There's a an awful lot of cart before horse recently. Someone needs to step back and prove there are audible differences between level matched pcm and DSD under blind conditions before jumping to conclusions as to why one sounds better than the other.

Oh WE HAVE done this with a Delay line I designed for a Pro Audio application.

The board sampled at PCM 234KHz or could be slected for DSD @ 6MHz.

Everything was identical, expect the PCB could select DSD or PCM data path, so same playback levels, same ADC front end - DAC analogue stage etc. When we first tested the board we had a toggle switch flying off the PCB so you could toggle between PCM or DSD mode - so in effect the mode was "Blind" to he user (don't forget the toggle switch was loose so you would have to confirm which mode the board was in by checking if the mode select line was High or Low with a multimeter).

We ask everyone who walked past the lab (including the accountant) to try and say which mode they felt sounded best - everyone but everyone preferred DSD and the difference was not small.

Whats interesting is that our ears are very sensitive to this "Time Domain" distortion, the accountant was a lady in her 50's and after she tried the listening test, she went back to her office on the other side of the lab (think of a large barn type structure with her small office in a corner room) and from her office she would yell out "This is DSD mode" and she would be right everytime!

If you hear NativeDSD verses PCM under the same conditions its so immediately obvious what is "natural" and transparent so you would not need to ask such a question.

If 100% Native DSD is never going to see traction with the major record labels - then I can only hope the MQA offers some hope as PCM is so poor verses a good analogue source (think Master Tape or a high end Vinyl setup).

 

[Julf]

If 100% Native DSD is never going to see traction with the major record labels...

It won't, simply because native DSD can't be mixed or processed without being converted to an intermediate PCM format such as DXD.