192KHz 24bit DAC No oversampling and No digital filter

[SoNic_real_one]

HDCD sounds better than CD on any decent decoder - regardless if it is "hardware" or "firmware".

 

[ThorstenL]

Hi,

It fits the M$ business model so well that they bought the company!

M$ bought Pacific Microsonic because they could add a completely invisible code to a CD, one which did not alter the music in any way but was reliably detectable. It is now IIRC part of M$ DRM Platform.

The actual HDCD application is something M$ never particulary cared about. If they had all CD-Players, DVD Players and DAC's sold worldwide would now be HDCD and all CD's made would be HDCD.

Ciao T

 

[regal]

Hi,



M$ bought Pacific Microsonic because they could add a completely invisible code to a CD, one which did not alter the music in any way but was reliably detectable. It is now IIRC part of M$ DRM Platform.

The actual HDCD application is something M$ never particulary cared about. If they had all CD-Players, DVD Players and DAC's sold worldwide would now be HDCD and all CD's made would be HDCD.

Ciao T

Actually it had nothing to do with music. At the time it was bought as a potential anti-piracy technique where you would need a chip in your computer to use M$ software. Somehow it was replaced with a better idea and audiophiles are held hostage. The biggest fustration for me is that a well mastered HDCD sounds much better played thru a PMD100 DAC than one without. And if you do any DSP (like add a little treble) the HDCD no longer works and you lose SQ. Note that I am not saying that had the CD been well mastered without HDCD in the first place it would sound any better than with HDCD ( this is impossible to prove.)

Compound that with the fact that PM had a genuis of an engineer who developed what most mastering engineers consider the best ADC ever made (PM2). This ADC is still used by some of the top studios who care about SQ. Yet the company that owns the HDCD code (M$) has no interest in audiophilia and it is just a complete mess that won't be settled until the patent expires and 352.8khz computer transports are developed (computer can function as a 8x oversampler with the full hdcd code.)

 

[tritosine]

On the other hand, pro audio guys dont miss hdcd, all they want is the next "virtual analog plugin" with its inherent nonlinearity, 80's hardware reverbation unit in plugin form ( say its 15 bits 48khz... , but 14bits 26 khz sounds good too) , clipping, phase shift, tape saturation plugin! , uh oh.
The thing with HDCD is,that even if it was true ( it isnt) , it would be way overrated , as it is now...

 

[audiodesign]

The HI-Face USB -> I2S module have got 102 person interested to buy it.

The Digital section with shift register probably will be created and sell by Eric Juaneda (pcb and kit).

The power supply sections and the output stage will be created and sell by Quang Hao (pcb and kit).

 

[EUVL]

Well, not quite 102 persons, but 102 units, as some are ordering more than 1, like yourself.
Minor details.

Patrick

 

[tritosine]

Microsoft Audio Watermarking Tool - Microsoft Research

whew, some sulfuric odor there

 

[Nazar_lv]

Complementary filtering is not neccesarily the better choice, in fact often severely non-complementary filtering (apodising e.g.) seems the better choice, if filtering must absolutely be used.

for low sampling rates (such as 44.1) complementary filtering (HDCD) is the best choice

I know of no customer or reviewer who ever preferred using any of the digital filters consistently (most use them very rarely).

I know only 10% people prefered NOS

That's only if applied in the mastering. HDCD.exe can tell you (detect) if the filters have been used. The tracks I tried did not use the filtering

Filtering is used always

Apparently it has a 20bit 96khz multibit converter

20/88.2

I think its built around a residual ADC

SAR ADC with proper dithering

I don't think that filter selection thingie is all that important once you record at 96khz

importan, but for downsample to 44.1 it is very important

How about the pmd100 coefficients? Its an old chip, I think it must be around 24 bits , no?

27

Basically IMO there are only 3 ever made that sound right: PMD100, SM5842, SM5847

not 3, from best-to worst SM5845,CXD1244S(CXD2567), PMD100, CXD1144BP, SM5842, SM5847, partially SM5846, YSF210

 

[ThorstenL]

Hi,

for low sampling rates (such as 44.1) complementary filtering (HDCD) is the best choice

This is not my experience.

I know only 10% people prefered NOS

You know that among our customers with our products only 10% of them prefer Non-OS? How do you know that? And what about the reviewers (who all and sundry did not like any of the over/up-sampling modes better than correctly implemented Non-OS. Even where it does not affect products I had a hand in but ones that implement Non-OS with great care the reviewers pretty much universally preferred the results!

So where does your statistic come from?

The usual? ("99% of all statics are made up on the spot, including this one!")

Ciao T

 

[regal]

Hi,



This is not my experience.



You know that among our customers with our products only 10% of them prefer Non-OS? How do you know that? And what about the reviewers (who all and sundry did not like any of the over/up-sampling modes better than correctly implemented Non-OS. Even where it does not affect products I had a hand in but ones that implement Non-OS with great care the reviewers pretty much universally preferred the results!

So where does your statistic come from?

The usual? ("99% of all statics are made up on the spot, including this one!")

Ciao T

He said he only know 10% of people who prefer NOS, thats not a statistic that personal experience.

Basically your tests with your customers are invalid because you weren't comparing OS with a good OS digital filter. So I challenge us to design this new DAC with the ability to run NOS filterless with a switch to run thru a PMD100. I think the results will surprise you and if anything at least we will be able to listen to tue HDCD mastered CD's without SQ penalty.

 

[ThorstenL]

Hi,

He said he only know 10% of people who prefer NOS, thats not a statistic that personal experience.

Sorry but I wrote:

I know of no customer or reviewer who ever preferred using any of the digital filters consistently (most use them very rarely).

So I am saying "among our customers non prefer OS"

I know only 10% people prefered NOS

In effect he is directly contradicting my statements and says HE KNOWS [that of our customers] ONLY 10% PEOPLE PREFERED NOS.

So either all our customers (or at least 90%) are lying as are all reviewers, or I am lying, or Nazar_Iv simple made a statement in error, which he can of course retract any time he pleases.

Basically your tests with your customers are invalid because you weren't comparing OS with a good OS digital filter.

How do you know that? Do you even know what hardware and software is involved?

All you know is that it was not HDCD PDM100/200.

So, unless we accept that the only good OS digital filters are PDM100/200 (in which we may as well quit using digital filters as either is basically depreciated),quite frankly you do not know!

So I challenge us to design this new DAC with the ability to run NOS filterless with a switch to run thru a PMD100.

Sure. Are you going to supply a few 100 OMD100's?

I think the results will surprise you and if anything at least we will be able to listen to tue HDCD mastered CD's without SQ penalty.

Well, I think I already do...

Ciao T

 

[regal]

Hi,



Sorry but I wrote:



So I am saying "among our customers non prefer OS"



In effect he is directly contradicting my statements and says HE KNOWS [that of our customers] ONLY 10% PEOPLE PREFERED NOS.

So either all our customers (or at least 90%) are lying as are all reviewers, or I am lying, or Nazar_Iv simple made a statement in error, which he can of course retract any time he pleases.



How do you know that? Do you even know what hardware and software is involved?

All you know is that it was not HDCD PDM100/200.

So, unless we accept that the only good OS digital filters are PDM100/200 (in which we may as well quit using digital filters as either is basically depreciated),quite frankly you do not know!



Sure. Are you going to supply a few 100 OMD100's?



Well, I think I already do...

Ciao T

Only a few would buy and install the PMD100 'cause as you say not many people like the sound of OS DAC right??

First this starts out as an updsampling NOS DAC (if thats not an oxymoron I don't know what is.) I think you are close minded and don't have the technical skills to implement a DAC with a good SIC digital filter switchable to NOS. And you are afraid that folks will prefer the filter if there was a switch to compare the two. JMHO.

 

[Nazar_lv]

Hi Thorsten, first sorry, my mistake and bad english, i told about audiophiles, melomans, diyers etc who i know.
Second NOS and OS designs are much different, and if you do universal (n-os) dac than one of it(NOS or OS) always lose, and almost always OS because you still need to pay more attention to the NOS otherwise it will sound very badly but it is not optimal for OS design and vice versa(OS-NOS)

 

[ThorstenL]

Hi,

Hi Thorsten, first sorry, my mistake and bad english, i told about audiophiles, melomans, diyers etc who i know.

Okay.

I still doubt that the numbers are valid. I think 90% of all Non-Os DAC's out there are based on the same three or four schematics and they are all universally quite bad. And most use TDA1543, which is even worse.

The fact that Non-Os still manages to get any following given what is generally on offer is perhaps more than proof that it does something right all OS systems (excepting possibly PDM100/200 when playing HDCD recordings) do wrong.

Second NOS and OS designs are much different, and if you do universal (n-os) dac than one of it(NOS or OS) always lose, and almost always OS because you still need to pay more attention to the NOS otherwise it will sound very badly but it is not optimal for OS design and vice versa(OS-NOS)

Not sure I agree. If I use a simple "flat response" analog stage after the DAC for OS and compensate the Sinc Rolloff for Non-Os and then use the same DAC with optimum powersupplies that keep impedances low enough at all sample rates (including OS) I cannot see this happening.

Ciao T

 

[regal]

Not sure I agree. If I use a simple "flat response" analog stage after the DAC for OS and compensate the Sinc Rolloff for Non-Os and then use the same DAC with optimum powersupplies that keep impedances low enough at all sample rates (including OS) I cannot see this happening.

Ciao T

But for the NOS Sinc Rolloff design will have an issue now that we could play possibly 44.1khz or 96khz ? I assumed you would be switching the analog stage based on sample rate regardless

Also please be warned that the PCM1704 sound pretty bad with an I/V resistor above 25 ohms, it sound wonderful with an I/V of 10 ohms, so the tube stage should be around 100x gain. Its not like the AD1865 that can handle a big I/V resistor.

 

[Marek]

Regal,

about PCM1704. Please tell me this 10ohm you mentioned is related to output voltage swing? so if I want to piggy-back 2 or even 4 PCM1704 load resistance should be 2 or 4 times lower then with 1 chip? (5 and 2,5ohm?)

 

[ThorstenL]

Hi,

But for the NOS Sinc Rolloff design will have an issue now that we could play possibly 44.1khz or 96khz ? I assumed you would be switching the analog stage based on sample rate regardless

Up to now all my stuff is set up for 44.1K.

TBH, I think the SINC rolloff compensation is only needed for 44.1/48KHz. As it is not straightforward to get a perfect fit for the compensation you are likely leaving a little droop at the HF end at 44.1K, so you can use the same circuit for 48KHz (in practice, I am aware of the theoretical issues, but those who excessively hold with theory do not use NOS DAC's anyway).

With 88.2KHz sample rate the Sinc Rolloff is only 0.2dB @ 10KHz and 0.75dB @ 20KHz, I will live with that without loosing sleep. On the other hand the -0.75dB at 10KHz for 44.1KHz I find very audible (less so I suspect than the -3.2dB at 20KHz).

Also please be warned that the PCM1704 sound pretty bad with an I/V resistor above 25 ohms, it sound wonderful with an I/V of 10 ohms, so the tube stage should be around 100x gain.

I would not use the 1704 if you paid me... ;-)

I still have a few K-Grade PCM63 stashed. I found the PCM63 the close second after the TDA1541. I did not like the PCM1702 and 1704 anywhere nearly as much. Why, oh why did Philips never make a 20 Bit DAC on the TDA1541 Technology?

Ciao T

 

[regal]

Hi,

I would not use the 1704 if you paid me... ;-)

I still have a few K-Grade PCM63 stashed. I found the PCM63 the close second after the TDA1541. I did not like the PCM1702 and 1704 anywhere nearly as much. Why, oh why did Philips never make a 20 Bit DAC on the TDA1541 Technology?

Ciao T

Couldn't agree more.


Marek,
As far as doubling up on the PCM1704s to use a lower I/V resistor, that is something I have always wanted to try but have been told it probably won't work, don't have the answer. There is a long thread at the other site about I/V for PCM1704.

 

[EUVL]

For the solution PCM1704 -> passive R -> high gain preamplifier (100x to 250x) :

nx PCM1704 in parallel => divide passive R by n.
Gain of the following circuit stays the same.


Patrick

 

[Telstar]

I would not use the 1704 if you paid me... ;-)

I still have a few K-Grade PCM63 stashed. I found the PCM63 the close second after the TDA1541. I did not like the PCM1702 and 1704 anywhere nearly as much. Why, oh why did Philips never make a 20 Bit DAC on the TDA1541 Technology?

Ciao T

You may be right on the sound, but less than 24 bits may NOT be enough if people wants to play highres material without too much resolution loss.

About the superiority of the pcm68 and 58 vs the 1704, i believe that the output stage for the latter shouldnt be the same.
Anyway, I almost always liked the pcm1704 when used right.