DSD comparison to PCM.

[opus111]

I don't see how any engineer would hold their tongue about sample rates for fear it would hurt their business.

Dan said it hurt his business initially from his saying 'no' to 192k. I can't see any reason for Dan to lie here and it makes perfect sense that when there are customers asking for 192k that it will hurt his business if the engineer says 'no, that's stupid, use 96k'.

 

[bbb]

Here's Lynn's review of the Invicta http://www.positive-feedback.com/Issue65/invicta.htm in the same issue as his other article. He seems to be very impressed by it and from what I gather, he thinks well recorded DSD64 is better than 24/172 PCM. He credits the ESS 9018 chip and the care that Resonessence puts into the Invicta. He also says the difference between the Invicta and his Monarchy on PCM is small with the advantage to the Monarchy.

<edit> Reading the 2 articles in conjunction, I think he's convinced that ESS has made substantial headway in algorithm on the 9018 and that with enough care to the other parts, you could have a new king. <edit>

Over to you, Opus.

 

[PeterSt]

Dan said it hurt his business initially from his saying 'no' to 192k. I can't see any reason for Dan to lie here and it makes perfect sense that when there are customers asking for 192k that it will hurt his business if the engineer says 'no, that's stupid, use 96k'.

If you read that "white paper" from Grimm from 2005 (referred to a couple of posts back) then you see a nice resemblance with Dan's papers telling about the "why not". For me it reads as "the more arguments for the lower rates, the more from it droops that you just couldn't do it boy". But that's me. And if you read deeper into everything coughed up, someone like DL is able to tell you that 16/44.1 really really is enough. But of course it is not and all "we" do is finding excuses for something we couldn't achieve. Btw, by no means this is meant to disrespect either man and the sheer fact that way good products leave their shelves tells me to show respect either way. And why would we not respect everybody for that matter.

Still, and again when all what's produced for "white papers" over time is combined, we can learn from inconsistencies how things really are or how those "guys" really think about what should be done, but what could not be achieved anyway. I am not different although I tend to talk in so much "vagueness" that nobody will know what my real merits are, which at such time I don't know myself yet. This is different from "stating" non-facts which I may have to come back on later and which at some stage will be read by a couple of smart-asses.

Regarding DSD vs PCM my "facts" are that I don't know. Facts are also that in July 2011 I announced a user upgradable DSD module for the NOS1. It still is not there. Why not ? because I still can't tell whether it's a good thing. Of course I could have made up a 100 reasons why PCM has to be better, and I really can and in a way people would dig it. Next I'd have to go along with the stream and provide it anyway (or otherwise loose business). And then what ? And so it has to come down to really do it, which in my case is a sort hell of a job because I'd have to implement it differently than others in order to *know* it is the good thing technically. And wasn't the sheer reason I started my own playback software because all sounded different (bit perfect and all) but nothing sounded good ? This is the same and here it starts. And only when finished I can tell but still only from the software side of matters. Next comes the hardware if the software side tells me to proceed.

I better start making up that story why PCM is better. Haha.
Peter

 

[opus111]

Over to you, Opus.

Haven't read the review yet but your remarks make sense to me. Lynn's not a fan of the NOS DACs he's heard, his preference for PCM1704 with oversampling tells me he's partial to a touch of noise modulation. If he's playing back 24/176 through the ESS then probably it won't sound as good as DSD owing to the internal processing of the ESS. To get 24/176 sounding really good (meaning with very low noise modulation, i.e. maximum dynamics) in multi-bit requires an extremely low glitch DAC, probably lower glitch even than TDA1541A. I agree that ESS has done a fine job with S-D in the Sabre32 but there's still the dynamics issue of the following analog stage that's too great a challenge, SQ-wise. Nobody to my knowledge has implemented it yet with passive LC filtering which ISTM is really demanded to preserve anything close to the DS's 130dB+ dynamic range.

I shall read the linked article now and if I have anything further to add, I will add it

 

[Orb]

Well Peter,
some engineers at Chord Electronics while liking and preferring DSD think properly done high rez is more accurate from a DAC perspective

Problem IMO is the amount of high rez PCM masters/DL that are not correct, HiFi-News identifies roughly 20% to 30% in their high rez review section with some kind of flaw every month, rarely they find only one with a flaw - usually review around 6-8 hi-rez releases a month so up to 30% of that with an issue raises a lot of questions in the industry.
That is depressing, and then we may have the additional headache of transcoding in the studio of DSD/DXD to 192/24 PCM downloads, if there is already problems dealing with up/down sampling for some studios-labels how the heck they going to manage adding DSD into the processing chain.

And yeah the Grimm whitepaper is only that in the loosest sense, apologies after mentioning it a few times forgot that caveat last time.

Cheers
Orb

 

[PeterSt]

That is depressing, and then we may have the additional headache of transcoding in the studio of DSD/DXD to 192/24 PCM downloads, if there is already problems dealing with up/down sampling for some studios-labels how the heck they going to manage adding DSD into the processing chain.

The problem exists with DSD just the same or maybe worse;
Theoretically (I think) more DSD material should be available than genuine PCM Hires. So, while most often the PCM Hires had to be created from something because it wasn't kept as a master or never existed in the first place (DVD-A multichannel yes, but what to do with that), DSD will be inherently Hires (say 80KHz) but was either

- mixed in analogue (could be the best form);
- mixed in digital through more or less PCM (which makes it sort of moot when compared to real Hires PCM);
- was created from analog and never saw an A/D during the original recording.

... or was recorded A/D live and never mixed.

Only this latter should be optimal and without doubts the best version of DSD. This is where I said that a relative hand full ot it would exist *and* it to be found by us and proven to be just that. Never mind the 500 or 10fold or whatever people came up with earlier in this thread, because first it is not allowed to be compromised with the issues from my above little list. And, as long as I am correct that it won't exist or can't be judged for it anyway, it leaves us with ...

the PCM Hires sh*t which is also nothing for all the sort of inverse reasons from that little list.
And so, what to do this all for ?
I sure *am* working on it, but I really don't know what for and to satisfy who.
And almost the worst of it all is : Redbook is so damn good (ok, as I perceive it through my own stuff) and there is sooo much more of it (all actually) that one must be really crazy to ever spend one minute more on "Hires". Not even the newly produced stuff is better by any small margin and we could say this is already so because now it 100% depends on the recording, the equipment used, the engineers, the mixing and foremost the artist. And I said it before : when I had to like the artists produced by e.g. Barry D. then I'd rather be send to hell first. Just an example and nothing to do with Barry.

My personal opinion is that this whole functional and technical discussion is completely moot. If we could only see new artists popping up in the amounts of what all is there, then I could have the idea of surviving into perhaps my 90's if my ears would ever last - and enjoy it then. But I don't see that happening of course as I actually don't see a single new artist at the horizon that I like. And the Led Zeps etc. I do like, are too damn good on Redbook to begin with. And you know what ? again Barry's fault. And that for a second stage (hence not taken in digital back then).

And so the discussion comes down to just this IMHO. As long as mastertapes have deteriorated or just have been lost it is just so ...
Peter

 

[opus111]

Here's Lynn's review of the Invicta http://www.positive-feedback.com/Issue65/invicta.htm in the same issue as his other article.

Now I've had more time to reflect on what Lynn said and his experience with the Invicta's rendering of DSD, I have a few questions. I accept his point (this also chimes in with DSD fans here) that merely listening to 'single box' SACD players does not deliver the full DSD experience. He says that DSD and PCM approach each other's sound more closely then, when a decent outboard DSD DAC is used. So what's wrong with the DSD mastered RBCDs that I have? Anyone have any idea for why (apart from one single exception which I put down to mis-labelling at present) they clearly have noise modulation, sapping the dynamics? If its down to the conversion to RBCD is there something restricting the quality of that process and if so what? Any ideas?

 

[rbbert]

Now I've had more time to reflect on what Lynn said and his experience with the Invicta's rendering of DSD, I have a few questions. I accept his point (this also chimes in with DSD fans here) that merely listening to 'single box' SACD players does not deliver the full DSD experience. He says that DSD and PCM approach each other's sound more closely then, when a decent outboard DSD DAC is used. So what's wrong with the DSD mastered RBCDs that I have? Anyone have any idea for why (apart from one single exception which I put down to mis-labelling at present) they clearly have noise modulation, sapping the dynamics? If its down to the conversion to RBCD is there something restricting the quality of that process and if so what? Any ideas?

The biggest problem I have with those reviews is the idea that the sonic ideal (to strive for) is embodied by SET amps. They may be euphonic, but few think they are anything even approaching a "straight wire with gain". Dismissing all solid state equipment (as he does in the longer article on DACs) is a poor way to start.

 

[opus111]

I'm not sure you've heard Lynn's amp designs - they're ones of his own design and construction, not conventional SETs (not that I know much at all about SETs). I haven't heard them so can't remark on whether they're euphonic or not, but I don't hear him dismissing all SS equipment. I've read his writings at length over on DIYA - this thread : http://www.diyaudio.com/forums/multi-way/100392-beyond-ariel.html (be warned its very long, took me a couple of weeks to get through it) and he does say that he's heard SS amps he likes the sound of, just not commercially available ones in the main.

What I hear him saying, putting his DAC article into the context of all his other writings I've digested is that its the low level linearity of his triode designs that he values, probably giving him the tonality and spaciousness he's looking for which helps him to overlook the other colourations that come alongside those qualities. I value those very same things that Lynn does but I'm seeking to realize them in SS designs which hopefully will free us from SET-type colourations.

Where I find myself differing from Lynn I found when I read this review he wrote of the Monarchy DAC: http://www.positive-feedback.com/Issue25/monarchy_m24.htm. Here's an excerpt :

On a subjective level, I've never liked the sound of the TDA 154X—I'm sorry, it's just too crude and old-fashioned for my tastes.

 

[bbb]

So what's wrong with the DSD mastered RBCDs that I have? Anyone have any idea for the sapping of (sic)the dynamics? If its down to the conversion to RBCD is there something restricting the quality of that process and if so what? Any ideas?

My guess is that it has less to do with the recordings and more to Lynn's choice of equipment. His Monachy oversamples and probably has some digital filter built in. This would, in comparison with a NOS design, already results in less dynamics and more noise modulation artifacts and phase issues. So, the difference with a SD dac would not be as apparent as a NOS one. Further, the Monarchy has a proper tube gain stage which will claw back some of the dynamics and contrasts lost with oversampling and filters which presumably led him to prefer multibits over the prevailing SD chips prior to the Invicta. Over our build of several dac platforms, we find that the digital section is not the ultimate arbiter of the final sound. If the implementation is so so, most times the ESS dac will win in a casual listening, it is simply more impressive upfront (like a good new world vs old world wine kinda way). Only when the implementation is good enough, then the difference is obvious. It also takes some training to discern the difference as most people would prefer the ESS sound as it ticks most audiophile checkboxes ie. more spacious, airy, smooth, detailed etc etc. But do they hear the phrasing and inflection from the throat box, the sticks on drum skins, the hammers and bow on strings and all the glorious resonance created that color each performances? Can they hear each individual notes, esp. piano starts and ebb without blurring into one? It took me sometime to get there and to me, this is the biggest difference between high end and the rest and no, you can't get there just with a super DAC chip.

Btw, Resonessence is closely related to ESS as the founder is the brother of the CTO of ESS so maybe they have a better insight or secret sauce into how to make the ESS chip sing given that SD chips are very algo dependent.

Wrt to recordings, we are screwed from start anyways if you really like music. Most of the good stuffs has been done and probably royally f***** up already so we work with what we have. Unless you have deep pockets, or like some members here who have access to unobtainium, and likes to listen to really esoteric stuff from the minor labels then maybe it'll work out. In the meantime, I'll buy all the cheap RBCD from Amazon UK and enjoy the music.

 

[opus111]

My guess is that it has less to do with the recordings and more to Lynn's choice of equipment. His Monachy oversamples and probably has some digital filter built in. This would, in comparison with a NOS design, already results in less dynamics and more noise modulation artifacts and phase issues.

Yes - this seems to be his personal taste - based on the review I just edited into my earlier post, he doesn't much like the sound of NOS. I am left curious to know which incarnations of TDA154x he listened to - perhaps they had the NOS droop which is what gave them the 'old fashioned' sound? Its this that makes me wonder if he's not partial to a tad of noise modulation. Also his technical analysis of what he considers to be flaws with the TDA DAC doesn't quite fit with my own understanding of them. From the studies I've done, the TDA1541A is the most technically perfect (traditional measurement-wise) of any 16bit capable DAC, including the offerings from TI/BB he seems to prefer. So all the stuff he's writing about being 'only 12 or 13 bits' does not ring true to me. He seems unaware that the TDA1541A does not rely on resistor matching, rather its using DEM which is a self-calibration scheme.

It took me sometime to get there and to me, this is the biggest difference between high end and the rest and no, you can't there just with a super DAC chip.

Yep, totally in agreement there - it takes time to appreciate the high end, certainly has taken me time to get there. And yes totally agree about the 'old world vs new world wine' thing - a very good analogy.

Btw, Resonessence is closely related to ESS as the founder is the brother of the CTO of ESS so maybe they have a better insight or secret sauce into how to make the ESS chip sing given that SD chips are very algo dependent.

Yep if these guys don't know how to implement the chip best, then no-one does. There's more to DAC design though than just the chip implementation and from the internal shots of the Invicta I've seen, its not done the way I'd do it, system-wise. But then I'd never start out with an ESS9018 so what I'm saying is kinda irrelevant...

In the meantime, I'll buy all the cheap RBCD from Amazon UK and enjoy the music.

+1 to that

<edit> A thought just occurred to me - perhaps Lynn listened to TDA1543 - this one most certainly is 12-13 bits and does indeed sound a bit old fashioned. But that's not the SOTA for NOS, its kindergarten level...

 

[BudP]

amazing stuff hiding in the dark

I can contact both Lynn and Rene Jaeger (Berkley designer) and see if they will post here. Lynn likely will, Rene', not so sure, but I will prod him tomorrow.

Bud

 

[NorthStar]

-- Hi Bud, and Welcome! :b

Bob

 

[opus111]

Hi Bud - are you the same Bud from over on DIYA, Mr EnABL ? If so, very warm welcome indeed

 

[BudP]

SHHHHH let's not start that fight over here too please. Three wars is enough! Thanks for the welcome guys!

 

[opus111]

Oh that's what makes WBF a little different, we're on the whole more into playing than fighting

 

[bbb]

There's more to DAC design though than just the chip implementation and from the internal shots of the Invicta I've seen, its not done the way I'd do it, system-wise.

I agree and that can be said for most of the modern dacs out there but then I'm old skool, I like irons, lots and lots of it.

 

[bbb]

Welcome Bud!

 

[LL21]

I'm not sure you've heard Lynn's amp designs - they're ones of his own design and construction, not conventional SETs (not that I know much at all about SETs). I haven't heard them so can't remark on whether they're euphonic or not, but I don't hear him dismissing all SS equipment. I've read his writings at length over on DIYA - this thread : http://www.diyaudio.com/forums/multi-way/100392-beyond-ariel.html (be warned its very long, took me a couple of weeks to get through it) and he does say that he's heard SS amps he likes the sound of, just not commercially available ones in the main.

What I hear him saying, putting his DAC article into the context of all his other writings I've digested is that its the low level linearity of his triode designs that he values, probably giving him the tonality and spaciousness he's looking for which helps him to overlook the other colourations that come alongside those qualities. I value those very same things that Lynn does but I'm seeking to realize them in SS designs which hopefully will free us from SET-type colourations.

Where I find myself differing from Lynn I found when I read this review he wrote of the Monarchy DAC: http://www.positive-feedback.com/Issue25/monarchy_m24.htm. Here's an excerpt :

On a subjective level, I've never liked the sound of the TDA 154X—I'm sorry, it's just too crude and old-fashioned for my tastes.

Always enjoy reading your posts Opus. I have read someone else (Mike Lavigne?) once post something to the effect that SET amps are surprisingly linear at low levels...its just that expecting them to produce linearity at higher levels of output is not what they were designed for...thus one must DEFINITELY look at system matching to get the best out of SETS for what they do.

I believe Lynn is suggesting Class A amplification (zero feedback) in general follows along similar basic principals which he champions regarding low levels of disortion, no feedback loops which have their challenges, no switching from Class A to B...but again, he points out that unless you want to drive a furnace...Class A DOES have to switch at some point unless we remain at lower levels of output.

I am NO techie...i have never heard SETs (i really must someday)...but i CAN say, that so far my favorite SS amps have been Class A (i did not know this when i heard them), and i do like the 1 NOS DAC i have heard and continue to prefer it to the other newer DACs i have heard (i think many of which are DSD)?

 

[LL21]

My guess is that it has less to do with the recordings and more to Lynn's choice of equipment. His Monachy oversamples and probably has some digital filter built in. This would, in comparison with a NOS design, already results in less dynamics and more noise modulation artifacts and phase issues. So, the difference with a SD dac would not be as apparent as a NOS one. Further, the Monarchy has a proper tube gain stage which will claw back some of the dynamics and contrasts lost with oversampling and filters which presumably led him to prefer multibits over the prevailing SD chips prior to the Invicta. Over our build of several dac platforms, we find that the digital section is not the ultimate arbiter of the final sound. If the implementation is so so, most times the ESS dac will win in a casual listening, it is simply more impressive upfront (like a good new world vs old world wine kinda way). Only when the implementation is good enough, then the difference is obvious. It also takes some training to discern the difference as most people would prefer the ESS sound as it ticks most audiophile checkboxes ie. more spacious, airy, smooth, detailed etc etc. But do they hear the phrasing and inflection from the throat box, the sticks on drum skins, the hammers and bow on strings and all the glorious resonance created that color each performances? Can they hear each individual notes, esp. piano starts and ebb without blurring into one? It took me sometime to get there and to me, this is the biggest difference between high end and the rest and no, you can't get there just with a super DAC chip.

Btw, Resonessence is closely related to ESS as the founder is the brother of the CTO of ESS so maybe they have a better insight or secret sauce into how to make the ESS chip sing given that SD chips are very algo dependent.

Wrt to recordings, we are screwed from start anyways if you really like music. Most of the good stuffs has been done and probably royally f***** up already so we work with what we have. Unless you have deep pockets, or like some members here who have access to unobtainium, and likes to listen to really esoteric stuff from the minor labels then maybe it'll work out. In the meantime, I'll buy all the cheap RBCD from Amazon UK and enjoy the music.

Another great read for people like me who are here to learn. Thanks. I'm all about your last sentence personally...i wish to enjoy my Redbook given that it represents 90% of my collection and 10% is hybrid SACD where i listen to the CD layer.

I have been surprised about how good the Oppos sound., particularly the modd'd ones...which i believe use the ESS DAC. I also have been incredibly impressed with the Stahl-Tek Vekian digital setup, and welcomed a few big-rig digital pieces to my home before. That said, over even 1 straight album, there is something in the back of my mind that hears something 'flatter' (i do not know how else to describe it)...that is very very difficult to hear in direct 5 second comparisons between 2 good digital setups. But its definitely there (for me).

I once was playing something thinking i was on the NOS DAC...walked away happy...but after 30 minutes i wonder why i was track flipping or stopping the album...stick in another one... suprised since i thought i enjoy that album...plus i rarely flip tracks and change albums. And then i realize after flipping 2 albums...ahhh...have the wrong input on the preamp. And flip back...and its back to 15 hours of listening to one album at a time while i work.

There IS something about bbb's description which i find as well in my own limited experience, but boy is it hard to 'spot it' in just a straight 'shoot out'.