Yeah... 4 x oversampling fixes all the issues.
A nicely executed NOS DAC can sound really great with high-resolution source files, in which case there's no need for oversampling. I find this option (with high-res source files) to be better sounding than any oversampling.
True. These days, I find the NOS sound of 44.1 (48) kHz material unlistenable with good speakers.
@elviukai I see, thank you for proper read. Now, what i can conclude from this is that it would seem that middle is pushed more toward your perception due to "rolloff" of the highs that you can hear. Add to that NOS dac droop, which further have high frequency rolloff, coming all the way down to 5khz. It happens due to sinc masking caused by sample and hold, which visually you can see as steps on unfiltered dac, and in turn it is supressing high frequencies a bit. And higher the sample rate, the less of it is pronounced (hence why OS sounds much better to you). There is a simple solution to it, sinc correction in output stage.
Now, there a common missconception regarding jitter and multibit dacs, saying they are resilient to it. Which simply is not true and a loooong game of chinese whispers (telephone game). The less jittery digital front, the better constructed output and you won't have the issues of those collapses of classical music. Especially with sinc boost, and increased dynamic range it brings.
On top of making sota layout for each individual dac chip, and reduce issues that arise from that.
Now, there a common missconception regarding jitter and multibit dacs, saying they are resilient to it. Which simply is not true and a loooong game of chinese whispers (telephone game). The less jittery digital front, the better constructed output and you won't have the issues of those collapses of classical music. Especially with sinc boost, and increased dynamic range it brings.
On top of making sota layout for each individual dac chip, and reduce issues that arise from that.
I dont think so. oversampling is bad, non oversampling is bad as well, upsampling (in PC) is also bad. thats life. all solutions is bad. for me oversampling (low tap or higher) with very simple analog filter is best compromise. Upsampling probably is better but it need to be executed NOT in PC.
I agree. and F1 car is perfect in track but horrible in streets. I have tried that path- 1) tried use PC upsampling via foobar- vrey bad. 2) tried upsample records (44 to 88 or 176 and 48 to 96 or 192)with variuos sofware- worse tan original song. i dont have golden ears but i can hear Kernel vs Wasapi vs DS . and i also hear those manipulations.
that was my test- 44khz records on NOS /OS configuarion in same dac. I dont claim anything less o anything more. just what I heard in actual DAC.
maybe problem is my job. I work with speakers for two decades. have heard white noise at leat 70 000times. usualy i can draw speakers frequency response on paper within +-3db form just one white noise signal. when i hear -5db drop in 14khz its hughe loss in airiness. when i was 20 years younger --5db drop @14khz was totaly aceptable.
time to time i hear audo stuf which have technical flaws- for example classe cdp0.3 or m2tech evo usb converter(yes a usb spdif conveter had drop in 12-14khz) . friends of mine enjoy the sound. i came listen and said something is broken i took CD player look inside. look in schematics- and compare with real parts fitted- a wrong capacitor in salen key is inserted by asembly line. probably all classe cdp0.3 have such failure. people listen, enjoy, review it. even audio magasines.
when your job is to hear subtile nuanses you simply are very sensitive, because ear are trained. and not nessesary ears should be perfect. my ear is trained for hearing speakers flaws but probably as a side efect of that i hear dac flaws as well and probably my brain amplifies those flaws much more than just simple lsitener.
I do not know other answer or theory why for me NOS sounded so wrong.
time to time i hear audo stuf which have technical flaws- for example classe cdp0.3 or m2tech evo usb converter(yes a usb spdif conveter had drop in 12-14khz) . friends of mine enjoy the sound. i came listen and said something is broken i took CD player look inside. look in schematics- and compare with real parts fitted- a wrong capacitor in salen key is inserted by asembly line. probably all classe cdp0.3 have such failure. people listen, enjoy, review it. even audio magasines.
when your job is to hear subtile nuanses you simply are very sensitive, because ear are trained. and not nessesary ears should be perfect. my ear is trained for hearing speakers flaws but probably as a side efect of that i hear dac flaws as well and probably my brain amplifies those flaws much more than just simple lsitener.
I do not know other answer or theory why for me NOS sounded so wrong.
It's subjective and expected, we all hear a little differently. Plus limiting the upper range with age.
Can we just not accept that each of us hears differently, has different speakers, rooms, etc and enjoys different music?
I find it interesting to read about others' experiences, but i know there is no way they can relate fully to my own. And that's fine.
Elviukai's contributions to this thread have been nothing but interesting and informative, for which i am grateful.
As for my own experience with NOS, historically it's been a mixed bag. Poor memories of late 90ies experiments with 1543 dacs and 1541 cdps. Stuck to delta sigma ever since, until building a version of Miro's 1862. Which wasn't a blazing success overnight. Initially it sounded rolled off, with a poor soundstage, not very good at large scale classical and quite poor at well recorded early music with a lot of natural ambience. Relatively small tweaks and re-voicing turned it into my favourite dac. It's certainly not perfect, but has cooled down much of my enthusiasm to try out the latest AKM chips. And ESS is not even on the horizon.
As for over/up-sampling and analogue filtering... none of what i've tried sounds better to me than nothing at all. Native hires? For some reason, the improvement, if any, is very subtle. Fwiw i hear a much bigger difference between redbook and higher res using sigma delta dacs.
Also spent some time with the offline HQP upsampler, exploring the multiple options, as there is always a suspicion when upsampling is done in real time.
Of course, it was just my personal conclusion to the sound he is hearing with his hearing measurements. Nothing of it was malicious, and i thanked him for his detailed write up on the matter/question, and my question was genuine. Unlike many, i try to mix both something tangible (measured, tested) and untangible (percepction), in hopes of middle ground and to improve upon designs. I'm not extremist in neither of the camps 🙂
Makes me want to have the same test to my hearing, and compare, try to figure out the differences that do not derive from personal bias (elvi analogy on dsp and brain is on point).
Upsampling is never going to be a great thing technicaly, having something imaginary put inbetween samples, derived from an algorithm, is not what the recorded music sounded like. Hence the NOS for me is a true way to listen. Be that as it may, NOS can be improved, a lot. Digital front must be impecable, reclocking is a must (with very good oscillators and well designed surrounding circuitry), it doesn't make it perfect but reduces problems that come out of the dac in layman terms. A picture, bit of an exaggeration, but it's easier to figure it in ones head.
As for analog side and filtering, i opt for sinc correction for 44.1k/48k recordings (very easy to switch on and off for preferences), it very much so brings dynamics up, and that percieved airness which some may find lacking on nos dacs (described above). It doesn't take much (if any, at least for me) from the signal, but contributes a lot to the optimization of nos dac. Anything else, i feel the energy of the signal gets damped quite a bit. Sota layout for each dac is a must, they simply cannot be replicated in the same circuit as it is not optimized for it. And these chips have been made with different tech over the years, and with quick course from reputable sources you can see why you cannot make the same layout and circuitry for cmos vs bipolar.
What about this one?
https://www.ebay.com/itm/2250579208...y2w8jUGrRFP9OOBsfqWkNl+Q==|tkp:Bk9SR6jHx9DoZQ
https://www.ebay.com/itm/2250579208...y2w8jUGrRFP9OOBsfqWkNl+Q==|tkp:Bk9SR6jHx9DoZQ
First, the cheap OP IC is used as LPF (low-pass filter) and the vacuum tube is used as the output component to completely avoid the influence of IC and the loss of detail caused by a large number of negative feedback. The circuit architecture uses two 6dj8 as SRPP continuous output, with large output amplitude, high S / N ratio and low output impedance. More importantly, the whole analog signal output is completely processed by a simple vacuum tube without any trace of crystal or IC. 🙄
Overall it doesn't seem like something I'd waste my precious PCM63 on.
Overall it doesn't seem like something I'd waste my precious PCM63 on.
I told you there are 2 DACs in the chip and you said I have a different maker of the chip.
This colinear, is my hypothesis why the dac is not neutral and elviukai describes the sound as such.
The datasheet doesn't explain the choices and reasons behind the choices of the design and the pro/con of this.
With a trained ear you will hear that under DSD white noise follows the signal in the background, it is awful.
You don't need to hear any high ultrasonic frequencies to appreciate audio, a violin or soprano won't go past 4k, and the important harmonics are very low level at 8k. I rarely see anything above 6khz. Above 12khz it is mostly room boundaries.
Most audio engineers filter high frequencies out of the recordings because it is annoying information which is not musical,
such as something dropped, thermal expansion of objects, sheets of paper turning of falling or electric motors noises.
People sneezing and coughing, that creates loud sounds into the 18k +, so it is obviously filtered.
The very highest note is C7, 4180hz, as long as you can hear to 13khz you don't miss anything important.
( people will get accustom to hear a deficient sound system, n.o.s., passive i/v etc.)
Most audio engineers filter high frequencies out of the recordings because it is annoying information which is not musical,
such as something dropped, thermal expansion of objects, sheets of paper turning of falling or electric motors noises.
People sneezing and coughing, that creates loud sounds into the 18k +, so it is obviously filtered.
The very highest note is C7, 4180hz, as long as you can hear to 13khz you don't miss anything important.
( people will get accustom to hear a deficient sound system, n.o.s., passive i/v etc.)
Hi,
Not sure about what we are talking. Is it the filter from the I/V in the feedback loop or passive or active filtering after the - 6 dB of the Cfb in the "virtual" ground of op amp transimpedance stages ?
How about - 6 dB more with simple passive RC between the I/V output and a buffer or at the output after a buffer or directly after a standalone op amp I/V ?
Can you hear -12 dB -3dB low pass at 22 K Hz in a NOS conf VS only - 6dB as far no oscillation is involved?
I wonder what about the nasty RF spreads from the current output dac chips VS the alliasing in what we hear to be nasty.
I have no harshness with my NOS experiments but I spend a lot of time to voice my DACS benchmarking the op amps, passive parts and layout in my DACs I like to sound natural but still experience the particular sound of a trumpet or a bell as I do like jazz music a lot. I am not experiencing fatigue with NOS. Soundstage is fixable in the voicing work but ask a lot of testing work which only DIY people can affoard to spend. And power supply is indeed not a poor part as already noticed in this thread.
IMHO and YMMV for sure.
Edit : about car and bodywork, I do find italians are still making great fresh things !
Last edited:
Maybe with a loudspeaker with a straigth spl curve à la monitors. But not with a high end good sounding loudspeaker in my experience at least to my ears in a damped room. Anyway those straigth spl curve monitors sounding like are only bearabler in special treated rooms. Or it is fatiguing as well as with x16 oversampling.
I only like the loudspeaker in civil environments having a progressive drop after 1 to 2 K hz to target something to -5DB to -10 dB at listening position for 20 K hz. I always wondered if the culpritt of fatiguing sound didn't come from not enough atenuated harmonics of the main area frequency where most of the music is whatever it is already took naturally by the mic (but with electronic music) and should theorically work with a straigth wire sounding monitors? But that's off topic here from me, just an opinion.
Last edited: