Is there any other point to this how many bits discussion except for wishful thinking?
It's easy to agree more bits would be fine, nut as Frank pointed out quite rightly, more bits does NOT automatically mean better sound. As we all know from our 16 current crop, most still haven't mastered even that technology. Even 16 bits can sound impressive, but doesn't far too often.
It's easy to agree more bits would be fine, nut as Frank pointed out quite rightly, more bits does NOT automatically mean better sound. As we all know from our 16 current crop, most still haven't mastered even that technology. Even 16 bits can sound impressive, but doesn't far too often.
Dither, as Wrinkle said.
Now run both through the output filter - how does it look then?
Jan
I don't see four Sabre D/A ICs there, just one. Maybe the other three are on the other side of PCB?
What I do see is SRC4392 sample rate converter on the input (that makes two SRC processes now), switcher PSU and a lot of lovely ceramic caps. All purist audiophile stuff..
Benchmark DAC2 HGC impressions (and Linux setup notes) - 4season - Digital Drive
No blood and offal. You need a pretty large Thai expat community here making food essentially for each other to get it.
Yup, right from SRC4392 DS data sheet, "accepts up to 24 bit audio data". Specs are right on 24 bit numerical noise.
OK, thanks. What this basically mean is that enyone else who's going to use this chip for 2 tracks will also use remaining six channels in one way or another. No need to toot marketing horn about multiple D/As per channel as if it is something unique.
Wrinkle
Can you please elaborate on the calculations over bit depth, bandwidth and power dissipation on a DAC?
I have looked at many DAC data sheets and I don’t see high W numbers.
In the first attachment, there are some relevant detailed numbers from a 14bit multibit Sigma Delta DAC (It is not for audio applications, working intentionally with fold-back alliasing components, but the issue you are raising should hold along any Nyquist zone operation).
In the second attachment there are some Power/Dynamic Range/Bandwidth comparisons from (mostly) Audio DACs
Distortion in ADCs or DACs in practice, does not have to do only (or not at all) with number of bits. (e.g. Integral and Differential Nonlinearity Distortion Effects)
(Chapter 3, pages 159-162)
ADI - Analog Dialogue | Data Conversion Handbook
There is no restriction to DC applications only.
The technique is used to build logarithmic DACs too.
George
Attachments
Which datasheets did you find for DACs with 28bit noise floors (-170dB) ? I rather suspect wrinkle's making the point that the thermal noise floor is going to demand high currents into low impedances to get the SNR up to 170dB.
None.
I have found ADCs from 8 up to 24 bits and DACs from 8 to 20bits
From these, I don’t see a steep power consumption escalation vs bit depth .
I can see though important changes due to DACs architecture.
Maybe.
I miss the theory, that’s why I asked Wrinkle if he could provide some info
George
Last edited:
Probably using electronic parts made in mainland China ..
Yeah, blame Canada!
http://www.youtube.com/watch?v=5CTvjwt1bvY
Yup, quantization error is an additive processes not transfer function linearity related. In practice the INL and DNL limit the performance and more bits do nothing.
The quantization error is decorrelated by dither, first principles stuff no need to go there again.
Last edited:
Judging from myself, the “first principles stuff” is always the easiest to overlook, misunderstand or both.
As to ADCs/DACs fundamentals and details, there is a lot of things I had (have) to read again and again for to understand them properly.
Questions and explanations from members here help.
George
Finally, some good info! A lot of assumptions about what I know or dont know or want to know or what I have measued. What I do not want is the same old stuff.
Of course there are many, many details.... none I want to rehash, though. The main difference that I hear between excellently executed designs of 16 and 24b though is the Bits.
Others will have to dispute that for and to themselves.
Maybe it would be possible to do better with another topology, such as those you indicated/show? Can that be done on the DAC side for audio?
THx-RNMarsh
Last edited:
I haven't seen you mention an apples to apples comparison yet. just comparing different masterings. Not sure there is proof that it is the bits? But guess you are convinced it must be the bits?
http://www.ti.com/lit/an/slaa013/slaa013.pdf
An excellent treatise on data converters for anyone wondering what some folks are hemming and hawing about.
I guess the better question is: are newer, high-bit DAC/ADCs simply more linear than their predecessors? Those nonlinearities can very easily swamp the entire difference between 16 and (ostensible) 24 bit depth.
An excellent treatise on data converters for anyone wondering what some folks are hemming and hawing about.
I guess the better question is: are newer, high-bit DAC/ADCs simply more linear than their predecessors? Those nonlinearities can very easily swamp the entire difference between 16 and (ostensible) 24 bit depth.
- Status
- Not open for further replies.