World’s First Valid Comparison of PCM versus DSD?

[Miska]

I'm just feeding some information about the approach of SONY ¨-)

 

AFAIK, DXD is not approach of Sony, but more approach of Philips.

 

This is the Sony approach:

http://www.superaudiocenter.com/images/Sonoma32.pdf

 

And based on what Michal said in RMAF '13 panel, big studios multitrack to a digital recorder, then play it back and mix it on an analog table and record the mix back to digital recorder. Seemed to be the case for Mark Knopfler's recent albums too, the effects, compressors and such used are analog ones.

 

So from work flow perspective it doesn't matter if the recorder is PCM or DSD.

 

So this kind of stuff is used for top mixes:

http://www.ams-neve.com/products/music/88-rs

[Miska]

By the way, it is of course also trivial to record the analog mix in parallel to both PCM and DSD...

[alfe]

AFAIK, DXD is not approach of Sony, but more approach of Philips.

 

This is the Sony approach:

http://www.superaudiocenter.com/images/Sonoma32.pdf

 

And based on what Michal said in RMAF '13 panel, big studios multitrack to a digital recorder, then play it back and mix it on an analog table and record the mix back to digital recorder. Seemed to be the case for Mark Knopfler's recent albums too, the effects, compressors and such used are analog ones.

 

So from work flow perspective it doesn't matter if the recorder is PCM or DSD.

 

So this kind of stuff is used for top mixes:

AMS Neve | 88RS

 

Miska,

 

My remarks are about SACD physical format and hardware SACD players not DSD

[Hiro]

I'm just explaining why they use PCM for low pass filtering in the first generation of SACD players.

 

I'm sorry but what relevance do the first generation SACD players have to the present-day SACD players and DSD DACs? New SACD players like the Marantz SA8005 do not convert DSD to PCM, whether they receive DSD data from an optical disc or via USB interface.

[alfe]

I'm sorry but what relevance do the first generation SACD players have to the present-day SACD players and DSD DACs?

 

Just to prove that Charles Hansen wasn't wrong for his statement about SONY when they released SACD and he don't deserve to be attacked permanently in this thread.

Except that I agree with you for all the rest ¨-)

[Miska]

My remarks are about SACD physical format and hardware SACD players not DSD

 

All the Sony SACD players I've had have DSD processors and have only DSD pins of the center/lfe/rear channel DAC chips connected... Thus physically unable to play multichannel PCM. So I don't know what you are talking about.

 

It is practically the same DSD processor chip that is used on the Sonoma DAW processing card.

[alfe]

All the Sony SACD players I've had have DSD processors and have only DSD pins of the center/lfe/rear channel DAC chips connected... Thus physically unable to play multichannel PCM. So I don't know what you are talking about.

 

It is practically the same DSD processor chip that is used on the Sonoma DAW processing card.

"And as far as it goes, that sounds all well and good. But then I came across an interesting paragraph in a 10-year old technical paper from Philips in Holland (who, together with Sony, were the driving force behind SACD). Here they talk about the typical “DAC” configuration used in a SACD player, and I was rather surprised to read it. According to this paper, the low-pass analog filters that are required to convert pure DSD to analog do not possess the impulse response characteristics they consider to be necessary for high-end audio performance. However, digital low-pass filters are more than up to the task. Therefore, the first thing the DAC does is to use a digital low-pass filter to convert the DSD to a 2.822MHz multi-bit PCM signal. This PCM signal is then fed into a SDM to generate a multi-bit (typically between 3 and 5 bits) SDM signal at 5.6MHz or even 11.3MHz. This multi-bit SDM can finally be passed through a low-pass analog filter without having to sacrifice the impulse response characteristic."

[Hiro]

"And as far as it goes, that sounds all well and good. But then I came across an interesting paragraph in a 10-year old technical paper from Philips in Holland (who, together with Sony, were the driving force behind SACD). Here they talk about the typical “DAC” configuration used in a SACD player, and I was rather surprised to read it. According to this paper, the low-pass analog filters that are required to convert pure DSD to analog do not possess the impulse response characteristics they consider to be necessary for high-end audio performance. However, digital low-pass filters are more than up to the task. Therefore, the first thing the DAC does is to use a digital low-pass filter to convert the DSD to a 2.822MHz multi-bit PCM signal. This PCM signal is then fed into a SDM to generate a multi-bit (typically between 3 and 5 bits) SDM signal at 5.6MHz or even 11.3MHz. This multi-bit SDM can finally be passed through a low-pass analog filter without having to sacrifice the impulse response characteristic."

 

Not sure if their explanation is correct. Most SACD players and DACs I see use analog filters, including the new SONY DSD128 HDD players.

 

From Sony's website:

 

"Analog FIR Filters

This digital-to-analog converter effectively reduces high-frequency noise in DSD signals. It features independent right and left channels, with four analog FIR (Finite Impulse Response) filters per channel producing a combined one-clock delay. FIR filters are ideal for D/A conversion of DSD signals."

[Hiro]

By the way, it is of course also trivial to record the analog mix in parallel to both PCM and DSD...

 

That's the very approach of Steve Hoffman for example. He does a split-feed A/D transfer from his analog console to PCM and DSD for CD and SACD layer respectively.

[alfe]

Not sure if their explanation is correct. Most SACD players and DACs I see use analog filters, including the new SONY DSD128 HDD players.

 

From Sony's website:

 

"Analog FIR Filters

This digital-to-analog converter effectively reduces high-frequency noise in DSD signals. It features independent right and left channels, with four analog FIR (Finite Impulse Response) filters per channel producing a combined one-clock delay. FIR filters are ideal for D/A conversion of DSD signals."

 

This white paper from Philips is from 2000 things has changed since.

As I said that was the early days of SACD today I agree with you it's irrelevant.

[Hiro]

And based on what Michal said in RMAF '13 panel, big studios multitrack to a digital recorder, then play it back and mix it on an analog table and record the mix back to digital recorder. Seemed to be the case for Mark Knopfler's recent albums too, the effects, compressors and such used are analog ones.

 

So from work flow perspective it doesn't matter if the recorder is PCM or DSD.

 

So this kind of stuff is used for top mixes:

AMS Neve | 88RS

 

If someone records to DSD or double-DSD and then mixes the recording in analog on this SOTA analog console I am all for it

 

BTW, Jan-Eric Persson at Opus3 uses a slightly different approach:

 

"The mic feeds goes through the famous Opus3 4-channel valve mixer where mild Equalizing and high-pass filters are applied. Then straight into the Korg MR2000S and captured in DSD128. "

 

http://shop.dsdfile.com/2013/10/04/yamina-love-letters-ep/

[magnum innominandum]

Hiro,

 

"The mic feeds goes through the famous Opus3 4-channel valve mixer where mild Equalizing and high-pass filters are applied. Then straight into the Korg MR2000S and captured in DSD128. "

 

dsdfile.com | Yamina – Love Letters EP

 

Funny thing is, the ADC in the Korg uses and ADC chip with a modulator that is in effect DSD Wide at 256FS (6 Bit), so it cannot record at all in DSD, UNLESS the signal is converted (on chip by BB's digital decimation filter) to DSD. The same of course is true for PCM.

 

So clearly any DSD recording from this machine is not pure DSD. On PCM recordings it is not pure PCM. What is it? A Chimaera, a three headed calf or a GM modified Super AD?

 

In fact, I may be mistaken, but only a few Meitner machines, the old Philips & Sony units and the Grimm AD1 offer pure DSD. And among the AD's that offer better than red book capabilities only the Pacific Microsonic ones are pure PCM.

 

Anything else is some kind of mongrel that is neither DSD nor PCM, neither fish nor fowl...

 

Looking at the various 'scope traces Miska posted was certainly very educational.

 

What is worse, massive levels of steady state ultrasonic noise or signal dependent ultrasonic noise?

 

I remember Pioneer once had a DA converter that they claimed re-created the ultrasonic components of the music lost when converting to CD. Didn't Wadia do something similar? In measured performance these were filters that released a large amount of ultrasonic images of the highest audio frequencies, but little of the lower ones. I quite liked Wadia's gear but I liked the DAC I have now better.

 

It seems that at least some people seem to think that signal dependent ultrasonic noise (images) are a "good thing".

 

Seems we can only get so far with physical measurements and listening to those who "listen" it seems many don't care if DSD or PCM, others favour PCM outrageously and declare DSD to be garbage while still others refuse to even to listen to PCM lest it has been turned into DSD first. So where is the truth?

 

Could it be that all of these people are right (even though they do not agree with each other) and are simply listening to different recordings, systems and with different preferences in sound?

[Hiro]

Funny thing is, the ADC in the Korg uses and ADC chip with a modulator that is in effect DSD Wide at 256FS (6 Bit), so it cannot record at all in DSD, UNLESS the signal is converted (on chip by BB's digital decimation filter) to DSD. The same of course is true for PCM.

 

So clearly any DSD recording from this machine is not pure DSD. On PCM recordings it is not pure PCM. What is it? A Chimaera, a three headed calf or a GM modified Super AD?

 

In fact, I may be mistaken, but only a few Meitner machines, the old Philips & Sony units and the Grimm AD1 offer pure DSD.

 

Korg uses PCM4202 ADCs which employ 1-bit delta-sigma modulators.

 

http://www.ti.com/lit/ds/sbas290b/sbas290b.pdf

 

As for delta-sigma modulators using a few bits more, it's still Delta Sigma Modulation we all know and love. There's no need for creating new names for Delta Sigma Modulation. We can either refer to such ADCs as multibit SDM or multibit DSD.

 

Looking at the various 'scope traces Miska posted was certainly very educational.

 

Food for thought for those who believed that PCM at 44.1kHz or low multiple was the best there is.

 

Seems we can only get so far with physical measurements and listening to those who "listen" it seems many don't care if DSD or PCM, others favour PCM outrageously and declare DSD to be garbage while still others refuse to even to listen to PCM lest it has been turned into DSD first. So where is the truth?

 

Could it be that all of these people are right (even though they do not agree with each other) and are simply listening to different recordings, systems and with different preferences in sound?

 

Such terribly polarized opinions should always be taken with a grain of salt. Many people love the sound quality of DSD recordings, even John Atkinson writes about being gobsmacked by the "sheer beauty of double-DSD recordings", and even Barry Diament uses 128fs SDM front-end to create all his 24/192 PCM recordings.

[Miska]

Korg uses PCM4202 ADCs which employ 1-bit delta-sigma modulators.

 

AFAIK, TASCAM also uses the same chip. And that's what I've used on my own ADC that can record both DSD128 and PCM 192/24 simultaneously from the same common analog front-end stage.

[Hiro]

AFAIK, TASCAM also uses the same chip. And that's what I've used on my own ADC that can record both DSD128 and PCM 192/24 simultaneously from the same common analog front-end stage.

 

You're right.

Product: DA-3000 | TASCAM

 

BTW, they also suggest using analogue consoles for their 5.6MHz DSD masters...

 

[manisandher]

AFAIK, TASCAM also uses the same chip.

 

My Korg MR1000 uses the PCM4202 1-bit ADC. My Tascam DV RA1000 uses the PCM4222 multi-bit ADC. For some reason, the Korg can record at 2.8MHz and 5.6MHz, but the Tascam only 2.8MHz. Don't know why that's the case, because I'm sure the PCM4222 is DSD128 capable.

 

FWIW, I spent quite a while comparing PCM to DSD and decided to go with PCM for my needle drops. But I don't think this was a reflection of PCM bettering DSD. More a reflection of my PCM DAC being far superior to the DSD DAC I owned at the time. For me, my needle drops recorded on my 'true' PCM ADC, and played back on my 'true' PCM DAC sound close enough to the original vinyl for me not to bother pursuing the DSD route. I've sold my DSD DAC, but have kept my two DSD recorders... in case I'd like to revisit DSD in the future.

Mani.

[Miska]

My Korg MR1000 uses the PCM4202 1-bit ADC. My Tascam DV RA1000 uses the PCM4222 multi-bit ADC. For some reason, the Korg can record at 2.8MHz and 5.6MHz, but the Tascam only 2.8MHz. Don't know why that's the case, because I'm sure the PCM4222 is DSD128 capable.

 

Yes, that's the older TASCAM, I was referring to the new DA-3000 model... (see Hiro's link)

 

I should have been more precise about the model.

[manisandher]

So, the newer Tascam uses the older 1-bit ADC? Surely the newer PCM4222 ADC measures better than the older PCM4202 ADC, no? There must have been a reason why Tascam went from multi-bit SDM back to 1-bit SDM. Maybe something for the 'The Multibit DSD debate' thread?

 

Mani.

[Hiro]

So, the newer Tascam uses the older 1-bit ADC? Surely the newer PCM4222 ADC measures better than the older PCM4202 ADC, no? There must have been a reason why Tascam went from multi-bit SDM back to 1-bit SDM. Maybe something for the 'The Multibit DSD debate' thread?

 

Mani.

 

If it sounds equally well, or better than its predecessor (DV RA1000), and this must have been the aim of TASCAM, then it's food for thought indeed.

[Miska]

So, the newer Tascam uses the older 1-bit ADC? Surely the newer PCM4222 ADC measures better than the older PCM4202 ADC, no? There must have been a reason why Tascam went from multi-bit SDM back to 1-bit SDM.

 

Could be just price, DA-3000 costs less than half of DV-RA1000. Although DV-RA1000 also includes DVD writer and HDD which has much higher price impact than the $10 for the ADC chips.

 

Who knows...

[Hiro]

Could be just price, DA-3000 costs less than half of DV-RA1000. Although DV-RA1000 also includes DVD writer and HDD which has much higher price impact than the $10 for the ADC chips.

 

Who knows...

 

It would be great if someone could compare the two units. Too bad the earlier one didn't support DSD128.

[magnum innominandum]

Hiro,

 

Korg uses PCM4202 ADCs which employ 1-bit delta-sigma modulators.

 

My bad, you are right. I read somewhere it used PCM4222 and did not check to be sure.

 

So, Korg users are exonerated from making mongrel DSD.

 

I would like to see a comparison between DSD using a Grimm AD1 and a pure DSD DAC (CS4398 in bypass mode? Lampizator?) chain vs. a Pacific Microsonic Model 2. That would be a fair comparison, I think and may actually reveal something...

 

Actually, I routinely do part of this, by playing HDCD via PDM100 HDCD Filter and PCM63 plus non feedback analogue stage with 2 X OS, or by playing Reference Recordings HRx.

 

Playing those as my personal reference (and originating with Pacific Microsonic Converters) I have not heard anything from either DS equipped Players/DAC or indeed DSD Playback that has given me the feeling I need something better or that anything heard so far was better, different sure, but different not in a way I would click

 

on.

 

So, anyone care to arrange a fair comparison?

 

Would a Reference Recordings HRx at 176.4KHz and a DSD File derived from this count (if so - get downloading the HRx and buy the SACD)?

 

Reference Recordings SACD Recording

 

Reference Recordings in the News

[Hiro]

So, anyone care to arrange a fair comparison?

 

Would a Reference Recordings HRx at 176.4KHz and a DSD File derived from this count (if so - get downloading the HRx and buy the SACD)?

 

Considering that the main aim of DSD is to get PCM out of the way, the proposed comparison wouldn't make any sense whatsoever.

 

In order to make a fair comparison you would have to use the Grimm AD1, or even the new TASCAM recorder DA-3000 that can do 5.6MHz DSD (but in the case of the latter we should compare it against a PCM recorder in the similar price range of course).

 

[alfe]

Considering that the main aim of DSD is to get PCM out of the way, the proposed comparison wouldn't make any sense whatsoever.

 

Wish fool thinking ? ¨-)

[Miska]

Would a Reference Recordings HRx at 176.4KHz and a DSD File derived from this count (if so - get downloading the HRx and buy the SACD)?

 

You don't need to buy SACD for that, you can use any of the players that can do PCM-to-DSD conversion and then you can run the conversion to DSD128 or DSD256 too...

 

Likewise you could take a DSD download and play it with those players to a PCM DAC of your choice with conversion going the other way around.

 

Of course, what you test in addition to the format itself here is sound of the particular ADC, conversion algorithms and sound of the particular DAC. So it's not any more fair that many of the other comparisons...