I'm not here to debate anybody...just to ask questions and learn a bit. Generally, i focus on PCM because 90% of what i own is CD, and 10% hybrid SACD, so i choose to optimize Redbook playback. I am curious thought about Ref Recordings, who is generally well regarded, using a NOS ADC...anyone know why he has made this choice?
DSD comparison to PCM.
- Thread starter U47
- Start date
I'm not here to debate anybody...just to ask questions and learn a bit. Generally, i focus on PCM because 90% of what i own is CD, and 10% hybrid SACD, so i choose to optimize Redbook playback. I am curious thought about Ref Recordings, who is generally well regarded, using a NOS ADC...anyone know why he has made this choice?
Just a very quick point of clarification. KOJ uses the same ADC as me. This particular ADC is only non-oversampling at 4fs rates, i.e. 176.4 and 192 KHz (see my earlier quotes form MR and KOJ). At these rates KOJ believes (and I agree) that it's not necessary to use over-sampling - the SR is high enough to get away without it.
Mani.
FYI:
It may not be the most obvious to everybody that at least as how I see "NOS/Filterless" DACs, it is not allowed to feed them with Redbook unfiltered. IOW, when we compare DSD to PCM and for PCM try to follow a chain which does not convert to "DSD" anywhere and it is about RBCD, this still *needs* upsampling/filtering.
If we like to compare apples to apples, the filtering is not allowed to be of any of the common types which exhibit at least post-ringing (and most often also pre-ringing) because it would fairly much come down to what an SDM does (for its necessary filtering).
While it is a (my) design choice to use genuine interpolation filtering which is 100% OK to the time domain and which is not 100% OK at all to the frequency domain, it could be important to see that now at least the time domain resembles that of native (!) DSD (but DSD is not 100%).
When Redbook is not filtered the time domain is still OK, but the frequency domain is a total mess.
... which is of course what NOS/Filterless lovers listen to and still like ...
When (PCM) Hires is played through a NOS DAC you get the real merits of it, unlike SDM based DACs (which always are to be considered OS) which re-apply filtering (not necessarily or necessary, but they do). It is there where 99% of PCM Hires flaws because you hear it as it was recorded (or engineered if you like); we'd be dependent on the ADC and further means used. It will, therefore, be no coincidence that I myself could not find any "HDTracks Hires" which sounded better than their Redbook counterparts which is just because I hear the real merits through a NOS DAC. Notice that this was two years before the Audacity Cowboys came about and let's say I'm well known for my views in this realm.
Disclaimer : if anyone tells me that "of course" Hires needs the filtering at playback just the same, then I'm out of this game. But I don't think this is so.
(and before you don't understand : what's coming out from HDTracks has been available years before they "obtained" it and I have it by the hundreds)
The above contains many many subjects assumed to be clear and understood by everybody, while I know this is not so at all and I too am seeking for real merits in some subjects. What I do like to express with it though, is that it is the most tough to compare apples to apples when it is about comparing PCM to DSD;
Mani showed some chains, some avoiding all conversions. But it doesn't tell all because there is too few means to compare; Therefore, consider this :
- What D/A chips are available which allow true NOS hence how many DACs can exist that allow comparing "glitching" assumed that were key to better sound (Opus' subject);
- When we have found a few DACs, how many are setup such that we can apply our own filtering means (and take it from me please that *that* makes the sound).
- When we take the DSD chain without conversions, how many DFFs etc. would we have of which we know that they are still native DSD by guarantee ? (suggestion : count on one hand).
And this is all outside of the debate which of the formats is technically better, with the knowledge that both are so different that it really needs 100s of posts to sort out that alone, and then still there's no real conclusion. This by itself makes me think that there is one solution only : listen. Yeah, well, that is what we all try to do. But now the remainder of my post above counts and I say that nobody is able to compare the apples to the apples because *if* we were able to find two setups which compare nicely, then it's still about the selection of DACs which hardly is there. So, Mani uses a PMII for PCM recording; I don't own one. He uses a NOS1 for PCM playback. Good, I have that, but there is nothing else that I know of might you not like that. He uses a Mytek for DSD playback; maybe not the best. And native DSD in there ? personally I even doubt that but alas. A Korg for recording ? Mwah. Tascam ? looks better but who tells it is the best. Using recording from vinyl to judge ? sure, I don't hear the difference when played through the NOS1 (and I tend to hear differences everywhere). But LP ? Not the best base in my view.
Moral : Someone like Mani really tries everything and all to find the real merits. But it hardly can be done by a human being;
When we have conclusions, it will be based upon that coincidental chain, the devices used (one is better than the other) and even the material we like to play. And to take this latter to the extreme : I can't find any PCM Hires that sounds better than Redbook. Think about how controversial this is and what it all can or has to take to be correct on it. Yes, include my ears please.
Peter
It may not be the most obvious to everybody that at least as how I see "NOS/Filterless" DACs, it is not allowed to feed them with Redbook unfiltered. IOW, when we compare DSD to PCM and for PCM try to follow a chain which does not convert to "DSD" anywhere and it is about RBCD, this still *needs* upsampling/filtering.
If we like to compare apples to apples, the filtering is not allowed to be of any of the common types which exhibit at least post-ringing (and most often also pre-ringing) because it would fairly much come down to what an SDM does (for its necessary filtering).
While it is a (my) design choice to use genuine interpolation filtering which is 100% OK to the time domain and which is not 100% OK at all to the frequency domain, it could be important to see that now at least the time domain resembles that of native (!) DSD (but DSD is not 100%).
When Redbook is not filtered the time domain is still OK, but the frequency domain is a total mess.
... which is of course what NOS/Filterless lovers listen to and still like ...
When (PCM) Hires is played through a NOS DAC you get the real merits of it, unlike SDM based DACs (which always are to be considered OS) which re-apply filtering (not necessarily or necessary, but they do). It is there where 99% of PCM Hires flaws because you hear it as it was recorded (or engineered if you like); we'd be dependent on the ADC and further means used. It will, therefore, be no coincidence that I myself could not find any "HDTracks Hires" which sounded better than their Redbook counterparts which is just because I hear the real merits through a NOS DAC. Notice that this was two years before the Audacity Cowboys came about and let's say I'm well known for my views in this realm.
Disclaimer : if anyone tells me that "of course" Hires needs the filtering at playback just the same, then I'm out of this game. But I don't think this is so.
(and before you don't understand : what's coming out from HDTracks has been available years before they "obtained" it and I have it by the hundreds)
The above contains many many subjects assumed to be clear and understood by everybody, while I know this is not so at all and I too am seeking for real merits in some subjects. What I do like to express with it though, is that it is the most tough to compare apples to apples when it is about comparing PCM to DSD;
Mani showed some chains, some avoiding all conversions. But it doesn't tell all because there is too few means to compare; Therefore, consider this :
- What D/A chips are available which allow true NOS hence how many DACs can exist that allow comparing "glitching" assumed that were key to better sound (Opus' subject);
- When we have found a few DACs, how many are setup such that we can apply our own filtering means (and take it from me please that *that* makes the sound).
- When we take the DSD chain without conversions, how many DFFs etc. would we have of which we know that they are still native DSD by guarantee ? (suggestion : count on one hand).
And this is all outside of the debate which of the formats is technically better, with the knowledge that both are so different that it really needs 100s of posts to sort out that alone, and then still there's no real conclusion. This by itself makes me think that there is one solution only : listen. Yeah, well, that is what we all try to do. But now the remainder of my post above counts and I say that nobody is able to compare the apples to the apples because *if* we were able to find two setups which compare nicely, then it's still about the selection of DACs which hardly is there. So, Mani uses a PMII for PCM recording; I don't own one. He uses a NOS1 for PCM playback. Good, I have that, but there is nothing else that I know of might you not like that. He uses a Mytek for DSD playback; maybe not the best. And native DSD in there ? personally I even doubt that but alas. A Korg for recording ? Mwah. Tascam ? looks better but who tells it is the best. Using recording from vinyl to judge ? sure, I don't hear the difference when played through the NOS1 (and I tend to hear differences everywhere). But LP ? Not the best base in my view.
Moral : Someone like Mani really tries everything and all to find the real merits. But it hardly can be done by a human being;
When we have conclusions, it will be based upon that coincidental chain, the devices used (one is better than the other) and even the material we like to play. And to take this latter to the extreme : I can't find any PCM Hires that sounds better than Redbook. Think about how controversial this is and what it all can or has to take to be correct on it. Yes, include my ears please.
Peter
Hi
Guess I don't hear much of PCM artifacts. For one I don't find the better PCM less dynamic than LP. TO me it is the contrary. I am not sure I follow but it seems that NOS is the Hoy Grail. My questions:
What do you think of these DACs:
Berkeley Alpha
Weiss DAC 202 (Maybe not the exact number)
I would be interested in both the technical merits and the listening experience if any. I am a little more than an uninterested party. I like very, very much what I am hearing from the Berkeley. Recollection from the Weiss is also extremely positive. There will be more questions for sure this subject is of interest to me.
Guess I don't hear much of PCM artifacts. For one I don't find the better PCM less dynamic than LP. TO me it is the contrary. I am not sure I follow but it seems that NOS is the Hoy Grail. My questions:
What do you think of these DACs:
Berkeley Alpha
Weiss DAC 202 (Maybe not the exact number)
I would be interested in both the technical merits and the listening experience if any. I am a little more than an uninterested party. I like very, very much what I am hearing from the Berkeley. Recollection from the Weiss is also extremely positive. There will be more questions for sure this subject is of interest to me.
I believe the Berkeley is using the AD1955, a current-out S-D chip which I have some experience of developing with. I've also studied the PCB of the Berkeley and its exemplary in noise control. From a technical point of view, its weaknesses seem to me to be two-fold: noise modulation from the S-D modulators in the DAC chip and the use of opamp I/V. Opamps simply aren't linear enough when exposed to the high level RF emissions from small geometry CMOS ICs.
Opus111
I don't find the Berlkeley lacking in dynamics by any stretch... Not do I find it noisy. I could be missing something and hope to be educated.
Would sensitivity to noise modulation in DACs be a trait comparable to the rainbow in single-chip DLP PJ, some people are not sensitive to this , I, OTOH, can get an headache from many single DLP PJs, thus my preference for 3-chip DLPs or DILA and its variants.
I don't find the Berlkeley lacking in dynamics by any stretch... Not do I find it noisy. I could be missing something and hope to be educated.
Would sensitivity to noise modulation in DACs be a trait comparable to the rainbow in single-chip DLP PJ, some people are not sensitive to this , I, OTOH, can get an headache from many single DLP PJs, thus my preference for 3-chip DLPs or DILA and its variants.
Frantz,
I don't regard myself to be allowed to have contentual expressions about both DACs you mention, but the both are the most extreme apart from eachother. To me it seems unlikely that you can like both.
Less than 1c
Peter
I don't regard myself to be allowed to have contentual expressions about both DACs you mention, but the both are the most extreme apart from eachother. To me it seems unlikely that you can like both.
Less than 1c
Peter
I lived with various incarnations of AD1955 DACs for a year or more and never noticed any lack of dynamics nor any noise. That is, until I had another point of reference - a multi-bit DAC. These aspects are only things I've picked up on relative to what sound to me to be more transparent performers.
Yes perhaps, though others on here are still hearing the effects of the noise modulation - just preferring it to its absence. Kind of like when I first tasted Chinese food in China, I found the lack of salt as compared to my western diet rendered the food rather bland. But over several months, I got used to it and appreciate now how having less salt allows the real flavours of the foods to shine though better. I can't go back to western levels of salt these days, its too 'in-yer-face' for me now.
Hi
I remember good things about the Weiss, experience with the Berkeley is current. haven't compared them side by side.
I am entirely vested in digital .. I don't see DSD as a technically correct way of doing things. On top of that 99.9984% of music is recorded in PCM, I will thus not invest in anything but PCM. I am interested in knowing what to listen to, IOW for the best digital available.
I understand Peter makes the Phasure where can one audition it in the US? Also does it require the XXHighENd software of are same results achievable with other players?
I remember good things about the Weiss, experience with the Berkeley is current. haven't compared them side by side.
I am entirely vested in digital .. I don't see DSD as a technically correct way of doing things. On top of that 99.9984%
of music is recorded in PCM, I will thus not invest in anything but PCM. I am interested in knowing what to listen to, IOW for the best digital available.I understand Peter makes the Phasure where can one audition it in the US? Also does it require the XXHighENd software of are same results achievable with other players?
Hi
This whole talk about NOS tends to bring back to a design based on an inexpensive DAC modification from one person here .... Hi Opus ... Must say that I heard a NOS DAC back 6 or more years, I can't even remember. To say that I wasn't impressed would be an understatement
This whole talk about NOS tends to bring back to a design based on an inexpensive DAC modification from one person here .... Hi Opus
... Must say that I heard a NOS DAC back 6 or more years, I can't even remember. To say that I wasn't impressed would be an understatement 
What is an NOS DAC?
1.A DAC with no reconstruction filter at all? eek!
2.A DAC with a non-oversampling reconstruction filter.
3.A DAC, like Phasure, that doesn't do any oversampling/upsampling inside the DAC, but does apply some kind of oversampling or other filter server side?
The problem is that the label NOS is used with any number of these designs and more. There are many flavors. Just generalizing about NOS DACs is pointless.
1.A DAC with no reconstruction filter at all? eek!
2.A DAC with a non-oversampling reconstruction filter.
3.A DAC, like Phasure, that doesn't do any oversampling/upsampling inside the DAC, but does apply some kind of oversampling or other filter server side?
The problem is that the label NOS is used with any number of these designs and more. There are many flavors. Just generalizing about NOS DACs is pointless.
I go for the most basic meaning - a 'No OverSampling' DAC which means a DAC where the output symbol rate is the same as the input sample rate and there's no digital filter inside. This definition necessitates it being a multi-bit one and makes no stipulation about anti-imaging (reconstruction) filters or flatness of FR.
As regards the Berkeley Alpha, the PMD200 is a digital filter chip (based I think on the DSP56000 family), not a DAC chip.
Sorry about that. That's why I deleted my comment. However, I didn't think the Berkeley used a SDM chip. Isn't the filter the same as the PMD 200 though?
Your definition would include the Phasure in the NOS camp then?
When I examined the PCB of the Berkeley (a hires pic Google found for me) the pinout of the chip did match the AD1955. That's not a 100% secure conclusion, but close enough for me. As far as I'm aware the filter's done in DSP firmware so could indeed be the same as that in the PMD200.
I'm not totally familiar with the Phasure - perhaps PeterSt would confirm that its NOS under this definition, it would be best to have it 'from the horse's mouth'. From my understanding it does have a digital filter but that's defeatable under user control - if so then I'd say it has the capability of acting as a NOS DAC in one configuration.
I'm not totally familiar with the Phasure - perhaps PeterSt would confirm that its NOS under this definition, it would be best to have it 'from the horse's mouth'. From my understanding it does have a digital filter but that's defeatable under user control - if so then I'd say it has the capability of acting as a NOS DAC in one configuration.
It is different;
There is no digitial filter nor an analog filter (the latter is easy to apply by anyone but is moot anyway). With this comes the *rule* that filtering is to be applied in-(PC-)software be it in real time or offline. In the mean time there's about nothing in the signal path and let's say that the sheer goal was to have it completely neutral. But why actually that ?
So any Hires could be compared to Redbook while the DAC behaves 100% the same electrically in all cases. This by itself is to be controlled by the PC software again which decides to upsample/filter to what rate. This is really how it started out for myself as a DIY project and only to get the real merits of Hires. That in the mean time I was intrigued by how horrible measuring could lead to listening pleasure was another part of the project and from that came the non-ringing filter.
Anyway, when Redbook is upsampled to e.g. 88.2 and compared with a Hires of 88.2 (which would be DSD based to stay on topic somewhat) the NOS1 by guarantee is electrically 100% the same in both cases. Also notice the importance of buffer sizes which in case of the predecessor of the NOS1-USB from today (the "old one" which Bruce obtained) could be as small as 1 2ch sample of 32 bits @ 384KHz. And since it won't go lower than one sample, that too could not be influencing (try a Weiss and see the difference).
For you Opus, the perceived glitching at the higher rates (PCM1704) is diminished by the higher upsample rate (thus better filter) which is the most easy to see by the THD figures already. So, most probably the higher distortion from R2R glitching will be audible (at least that is what I think) but with better THD figures to begin with, what to say. Next though, because NOS = NOS and filterless, here too it can the most easily audibly be compared because when we'd be afraid of higher distortion at the higher rate, we just dial in the lower rate. It is all too simple actually, but, part of a strategy of course. And in this case this includes the in-PC software which is 100% part of the game. Without that it's "stupid" NOS and I learned not to like poor figures.
Lastly, since I think you are interested in it, it is not much known (because I don't tell about it) that a 24 bit 192KHz (not upsampled) 1KHz test signal (@ -3dBFS) measures 0.00034% THD+N. This is is indeed way better than the TI spec, but a. I use 4 per channel with some tricks, b. the couple of 1000 PCM1704U-K I obtained are *better* than the U-K ratings (had them tested by TI themselves) and c. there is more than D/A chips alone.
One thing : The figures we look at through NOS are genuine and include the impulse resonse (which is "infinite" disregarding any sample rate but regarding the used slew rate and all), while the 0.00000less% figures of something like a Sabre ring the room apart.
Peter (sorry to be a bit enthusiastic
)
...
One thing : The figures we look at through NOS are genuine and include the impulse resonse (which is "infinite" disregarding any sample rate but regarding the used slew rate and all), while the 0.00000less% figures of something like a Sabre ring the room apart.
Peter (sorry to be a bit enthusiastic
Hi Peter,
First, your knowledgeable participation is greatly appreciated.
Second, enthusiasm is a very good living quality.
Last, "response"; you forgot the "p".
Highest regards, and I Welcome you,
Bob
Actually, I can name maybe 500, or more. There's the entire Telarc catalog, which were all recorded, mixed, and mastered on Sonoma workstations with mixer, several Pentatone and Harmonia Mundi, which were also mastered on Sonoma and Genex DSD mixers, the Blue Coast catalog which are recorded to tape, and mixed in analog, then archived to a Sonoma in DSD, and the entire Channel Classics Stereo catalog, which were balanced, and mixed in analog at the session, and butt edited with Pyramix in DSD.
There's also a number of DSD recordings that were mixed on SADiE workstations that also contain the Sony E-Chip, allowing editing and mixing in DSD-Wide (same as the Sonoma) without decimation.
Worth a quick mention here that 'DSD-Wide' is a marketing name for noise-shaped PCM. I'm presuming by 'entire Telarc catalog' you're only referring to their DSD mastered catalog - Telarc has been in business much longer than DSD has existed.
Bruce B
WBF Founding Member, Pro Audio Production Member
I can live with a few miliseconds of signal convertered to DSD-wide at the edit points... "IF" the DSD recording needed editing in the first place!
Similar threads
 | Steve Williams Site Founder | Site Owner | Administrator |  | Ron Resnick Site Owner | Administrator |  | Julian (The Fixer) Website Build | Marketing Managersing |