Yes, but the focus is on the lower 16 bit's which are true bits, so NOS. That is to say that few chips manage to produce good oversampling compared to sticking with the 16 true bits. It also appear that 20-bit have a higher quality of over sampling compared to 24-bit... at least from my perspective going over the available material.
Doede explained why he choose the PCM1794 over other true 16-bit chips.
Attachments
I can’t think of any reason why such an experience would have anything to do with an increase in bit-depth from 20, to 24.
I'm just the messenger
... My question is if we are not just talking about subjective experience, but one were we have people that are comparing digital vs analog AND levels of digital quality .. the later ofc being comparing over sampling from one manufacturer to another. This makes the whole "which sounds better" a freaking nightmare since one is Apples vs Oranges and the other is Oranges vs different Oranges.Wonderful!
Is this a study? Where are the values of, for instance, phase noise? I have no problems to believe that using a good external clock leads to better results than the internal clock of some audio interface – it is sufficient that the latter is bad enough.
So please try again, this is not a study on the effect of phase noise on SQ. It is just marketing supported by an ad-hoc comparison.
Please produce a similar study where you (or they) measure the effects of phase noise and show at which levels the SQ becomes indistinguishable from further improvements. Then and only then we have something scientifically significant.
For a moment you tricked me (that's I wrote "Wonderful!") but now I am a bit disappointed...
If talking multibits when we already know which subjectivly sounds a little better than the others for yet reasons to find.
Then it comes for personal preference. For instance one will like better an ad1862 than a pcm 63, voth 20 bits.
For my likings the pcm I like the most are TDA1541As ; 1540, 1545 (still pcm?) , AD1862, PCM56.
One nice guy @miro1360 offered the possibility to benchmark some of them in a very close pcb...First time a member did that.
Of course it says not it all but maybe what sounds good with minimal efforts than to choose good suppies and parts around.
I did not see the experts helping to improve that non mercantil project btw; for the moment the member share more subjective likings around parts or i/v topologies. At least the gerbers are free to download.
Sorry as it becomes off topic....and without listening it becomes at the end just litterature to describe personal experiences.
Then it comes for personal preference. For instance one will like better an ad1862 than a pcm 63, voth 20 bits.
For my likings the pcm I like the most are TDA1541As ; 1540, 1545 (still pcm?) , AD1862, PCM56.
One nice guy @miro1360 offered the possibility to benchmark some of them in a very close pcb...First time a member did that.
Of course it says not it all but maybe what sounds good with minimal efforts than to choose good suppies and parts around.
I did not see the experts helping to improve that non mercantil project btw; for the moment the member share more subjective likings around parts or i/v topologies. At least the gerbers are free to download.
Sorry as it becomes off topic....and without listening it becomes at the end just litterature to describe personal experiences.
Do you happen to know if they gave a more detailed account of their test somewhere? I didn't find anything about it on the Grimm site, but maybe I didn't look at the right spot.
Ken, bit resolution does not give the same number as ENOB
https://www.analog.com/media/en/analog-dialogue/raqs/raq-issue-90.pdf
It’s fairly easy to check for bit resolution at home using single tone test frequency FFT and watching for the x dBFS test signal peak just emerging from the ground floor (practically, bit resolution= x/6).
Or better zoom-in at the lower range of a gain-linearity plot.
It is more involving for an amateur to specify ENOB
https://cdn.rohde-schwarz.com/pws/d...notes/1er03/ENOB_Technical_Paper_1ER03_1e.pdf
In any case, my ears fall short close to 12bits
George
I concur, George. This is why DAC chips spec’d to support a raw 24-bits compute to some different ENOB. The ENOB isn’t synonymous with bit resolutionKen, bit resolution does not give the same number as ENOB
https://www.analog.com/media/en/analog-dialogue/raqs/raq-issue-90.pdf
It’s fairly easy to check for bit resolution at home using single tone test frequency FFT and watching for the x dBFS test signal peak just emerging from the ground floor (practically, bit resolution= x/6).
Or better zoom-in at the lower range of a gain-linearity plot.
It is more involving for an amateur to specify ENOB
https://cdn.rohde-schwarz.com/pws/d...notes/1er03/ENOB_Technical_Paper_1ER03_1e.pdf
In any case, my ears fall short close to 12bits
George
No need to look for experts' marketing slides as the answer can be seen here. So Andrea's clocks offer only very slightly better SQ than Crystek 957. Andrea has told us that Crystek 957 is a poor oscillator. There are even <10$ oscillators that have lower close-in phase noise than Crystek 957 (e.g. NZ2520SD/SDA).
So a reasonable conclusion is that Andrea's claims regarding the importance of low close-in phase noise are exaggerations.
Reasonable? I think you know its not all that reasonable, and that's your point.
Furthermore, you make claims about close-in phase noise of NDK NZ2520SD/SDA which you might know overstated. The one phase noise plot published by NDK was so far as I can tell only ever included on a single Japanese language marketing document. If it was supposed to represent typical performance was not specified (again, so far as I can tell).
My only point is that if you want others to speak more carefully about they actually know verses what they believe, then you might find more success with that goal by setting what you consider to be a more exemplary example in the framing your own claims.
Last edited:
No, I really think it is reasonable conclusion. If a poor oscillator is only very slightly inferior to Andrea's clocks then it is a totally reasonable conclusion that low close-in phase noise is not that important.
Regarding ENOB: from an error budgetting point of view, you normally like the analogue noise to limit the signal to noise ratio of a sigma-delta DAC, because reducing the digital noise by adding an extra bit is much cheaper than reducing the analogue noise by 6 dB (for a given architecture, that typically means quadrupling the analogue area and current). Hence, you get an input word length well above the ENOB.
- Status
- Not open for further replies.