NOS is almost synonymous with R2R/multibit architecture, which I think is the real preference we see in comparison.
4) DDDAC1794 - Doede Duma PCM1794A 24 bit sigma-delta dual mono sync NOS. Here the situation improves, nice musical to listen to but a too relaxed sound, lack of energy, the midrange remains a bit flattened, quite detailed but still narrow and shallow soundstage.
As the above description correctly states, the PCM1794A is a delta-sigma converter, operated in NOS mode (with it’s internal OS filter bypassed) in Doede’s DAC design. Yet, it was ranked above the #5 ranked AD5791 20-bit R2R based NOS DAC. The converter operating technology wasn’t the determining factor here.
Last edited:
I'm looking at the results broadly as I would never expect subjective preferences (or random DAC sound quality) to follow any sort of hard rules .
IIRC PCM1794 uses some type of hybrid R2R/DS architecture and happens to rank higher than Sabre.
IIRC PCM1794 uses some type of hybrid R2R/DS architecture and happens to rank higher than Sabre.
Last edited:
As you've shown, NOS can sound better for some compared to many OS filters, but high quality filters OS can sound even better sometimes.
@Markw4 If you are interested your listening demos could easily work out how the DAC-Lite performs in NOS vs OS?
From Ken's work the quality of the OS filter is paramount and PGGB plugin for Foobar is a good option if you have a Windoze machine
https://www.remastero.com/foo-pggb-rt-guide.html
If you want to be scientifically informed you could even do sinc 2M tap file processing
https://audiowise-canada.myshopify.com/products/pggb-it2?variant=39532161630268
and use Foobars DBT component https://www.foobar2000.org/components/view/foo_abx
How was the comparison done? I ask because it matters a lot. You can compare 2 of same components and they can sound different depending on how they are compared. So lets hear about this comparison details which brought out the audible difference.
Kazap is referring to a now inactive thread which had the objective of determining why OS and NOS seem to exhibit a characteristically different subjective sound, when, technically, it seems they should not. The ‘conclusions’ of the investigation are NOT properly scientific. This is due to a number of practical constraints, such as the lack of control over each participant‘s test set-up, and that the test group was comprised of voluntary ad hoc members of diyAudio. While the thread was rather lengthy, at over 1,900 posts, the following links are to the three primary summarizing documents resulting from the investigation.
Investigation Concluding Report (Ken Newton):
https://www.diyaudio.com/community/...s-sound-different.371931/page-90#post-6777645
FIR OS Filter Transparency Listening Tests (Hans Polak):
https://www.diyaudio.com/community/...s-sound-different.371931/page-89#post-6775491
FIR OS Filter Echo-Artifact Audibility Test (Marcel van de Gevel)
https://www.diyaudio.com/community/...s-sound-different.371931/page-76#post-6761338
I see, one of those, eh. OK.
So, what kazap wrote is just someone's personal preference. It could sound bad to someone with different taste or just was in a bad mood that day. It's too bad that such preference can't be directly shared on internet forums.
You may be retired, but that doesn’t seem to have improved your disposition.
I would like to note that Andrea had on his page a PCM1704 board as well. But it was not included in the test, while many other configurations yes. It makes me curious, given the inherent good sound of the chip..
This may indeed be a key issue. What I find somewhat baffling is that Benchmark DAC3 which is one of the very few dacs having custom filters (even fulfilling MarceldvG's intersample overshoot criteria) was found audible inferior by Markw4 compared to Topping D90 which has bog standard OS filters. The DAC chips are different but still it seems to confirm that these sighted listening tests are too subjective to draw any conclusions.
I have never used the D90 with it's built in filters.. never came into my mind. I suppose Markw4 did not use them as well.
Regarding DAC-3 verses D90, the external filter in DAC-3 was one of the best things about its sound. The reason D90 won was because of other factors. We could hear imperfections from both dacs, although the imperfections were different for each.
The other thing about DAC-3 is that my 2nd dac project, based on ES9038Q2M running synchronous DSD256 from AK4137, already sounded a little better than DAC-3 in some ways, but worse in other ways. Of course, development of the 2nd dac was based on having studied DAC-3 architecture and constantly comparing the sound of it to dac under development. When it came to playing DSD, DAC-3 could only manage DSD64 and the special filter could not be used. It sounded like a bog standard ESS dac in that mode.
Getting back to D90 verses DAC-3, we listened on the big ESL panels. D90 provided a wide and fairly deep soundstage at LF, and generally good, stable imaging at LF. As frequencies climbed up into the vocal range and beyond it increasingly sounded a bit off. Unnatural, a little distorted (not sure if it was actual distortion or something with a similar sound) maybe jittery and or noise modulated. By way of contrast, DAC-3 soundstage was narrow and forward (closer to the listener, more to the front between the speakers rather than extending behind them and out to the sides). It also sounded more distorted or distortion-like as compared to D90. Right now I forget exactly how many clocks there were in DAC-3, but there quite a few for dac. So far as I could tell they were not in the same league as Accusilicon used in D90 (although D90 clocks do pass through a CPLD, so there's that).
The other thing about DAC-3 is that my 2nd dac project, based on ES9038Q2M running synchronous DSD256 from AK4137, already sounded a little better than DAC-3 in some ways, but worse in other ways. Of course, development of the 2nd dac was based on having studied DAC-3 architecture and constantly comparing the sound of it to dac under development. When it came to playing DSD, DAC-3 could only manage DSD64 and the special filter could not be used. It sounded like a bog standard ESS dac in that mode.
Getting back to D90 verses DAC-3, we listened on the big ESL panels. D90 provided a wide and fairly deep soundstage at LF, and generally good, stable imaging at LF. As frequencies climbed up into the vocal range and beyond it increasingly sounded a bit off. Unnatural, a little distorted (not sure if it was actual distortion or something with a similar sound) maybe jittery and or noise modulated. By way of contrast, DAC-3 soundstage was narrow and forward (closer to the listener, more to the front between the speakers rather than extending behind them and out to the sides). It also sounded more distorted or distortion-like as compared to D90. Right now I forget exactly how many clocks there were in DAC-3, but there quite a few for dac. So far as I could tell they were not in the same league as Accusilicon used in D90 (although D90 clocks do pass through a CPLD, so there's that).
Last edited:
You are talking about the chip differences while twiddling around with the rest of the implementation. Stop pretending that you have a grip of "chip characteristics" or present scientific evidence that you have done a "fair" comparison by posting schemas and pictures for implementation along with the an architecture description on how and on what level the chips are compared. You speak serious, now it's time to act at the same level as you speak - or hold your piece...
//
//
As I have explained to beginners, dac performance is mostly determined by the circuitry around the dac chip. In that sense what the dac chip does is set an upper limit on the best performance that can be obtained. There are many variables involved. Its not just the chip, not just the filter, not just the clocks.
Regarding demands for proprietary information obtained by a lot of hard work, I will choose if and when to disclose details.
Regarding demands for proprietary information obtained by a lot of hard work, I will choose if and when to disclose details.
Last edited:
I agree but somehow all these comparisons end up making generalizations: e.g. DIYINHK (or TP Buffalo) ES9038 implementation no matter how poor in this test represented all DS dacs so the results are used as proof that R2R sounds better than DS. Or that NOS sounds better than OS.
Regarding NOS vs. OS the super slow filter in AK449x DACs should be same as NOS so using that when comparing to other NOS dacs could at least theoretically eliminate one variable.
IMHO the reason people tend to make generalizations from sparse data is well explained at: https://www.apa.org/monitor/2012/02/conclusions
IOW its just how the human brain works.
EDIT: If its more convenient to read it here, a quote:
"System 1 is a storyteller. It tells the best stories that it can from the information available, even when the information is sparse or unreliable. And that makes stories that are based on very different qualities of evidence equally compelling. Our measure of how "good" a story is—how confident we are in its accuracy—is not an evaluation of the reliability of the evidence and its quality, it's a measure of the coherence of the story."
IOW its just how the human brain works.
EDIT: If its more convenient to read it here, a quote:
"System 1 is a storyteller. It tells the best stories that it can from the information available, even when the information is sparse or unreliable. And that makes stories that are based on very different qualities of evidence equally compelling. Our measure of how "good" a story is—how confident we are in its accuracy—is not an evaluation of the reliability of the evidence and its quality, it's a measure of the coherence of the story."
Last edited:
I friendly hint to you is then to hold on to your "teasers" a bit or you will find yourself being added to the "max headroom" group by readers here 😉
//
I stopped listening to my Tannoy K3838's well over a decade ago, studio monitors or not. They are good, but not great, and far from neutral imo.
Mark, I think everyone reading this thread knows that everything matters. The chip, the pcb, power supplies, the IV, the entire chain, the room and the ears and brain of the listeners.
At the end of the day the original post is nothing more or less than a few guys listening to a few specific dacs and commenting on their personal subjective thoughts about what they heard. At least they posted some pictures so anyone curious can more or less piece together the dac implementation they used. Not sure it needed 24 pages of chest pounding or proprietary implementation knowledge but each to their own.
It is what it is. I ate at this restaurant. Here's a pic of what I ate. I liked it. That other dish, not so much.
Ken, the reference to your tests was an interesting read. Appreciate the link to www.remastero.com. 🙂
- Status
- Not open for further replies.