CS43131 and CS43198 use a gimmick (Dynamic Range Management) to artificially increase DR spec. Also the noise shaping causes the noise floor to increase sharply over 40kHz so claiming support for up to 384kHz sample rates is misleading.
See e.g. https://www.diyaudio.com/community/threads/is-the-cs43198-a-same-fake.403352/post-7454355
See e.g. https://www.diyaudio.com/community/threads/is-the-cs43198-a-same-fake.403352/post-7454355
Nope, if it handles 384kSPS it handles 384kSPS and there's nothing misleading about that.
Noise-shaping is used by all sigma-delta ADCs and DACs to achieve low noise in the audio band, its a fundamental part of the technology, so if it didn't have noise shaping it would perform extremely poorly - in fact there would be no reason to use sigma-delta without it.
Noise shaping pushes quantization noise higher in frequency in order to increase the ENOB(*) in the audio band. So at some point in the frequency domain noise will start to rise rapidly, so long as that point is > 20kHz there's no issue. Noise is reduced in the audio band at the cost of noise increasing ultrasonically. This noise reduction equates to increased precision.
Audio ADCs and DACs are not designed for use as arbitrary signal generation at ultrasonic frequencies. Luckily some ADCs will work at low ultrasonic frequencies (and can be used in Bat detectors for instance). The noise increasing from 40kHz is entirely reasonable.
(*) ENOB = Effective Number Of Bits.
Depends on how you plan to use it. Most well performing DAC chips have flat noise floor up to Nyqvist regardless of sample rate. As I said in this case the good looking specs are misleading.
Missed off the PCM1794a/5/6?
Just bought a dual PCM1794 pcb from China without the opamps as I plan on using my Sowter transformer for I/V.
I;m swapping out the dual TDA1547 audio board in my VRDS10.
It wasa highly regarded dac chip and still in production.
true, but im not sure if its the noiseshaping or the overall implementation but there is something to CS chips audible-wise and its a fairly good performing modern chip
after all the noiseshaping peak still sits at -100db or so, imo from experiments its also better to have uniform noise than "random distortion", even if the noise is higher in level its far more pleasant to listen to (well if the 40khz noise would be audible to begin with...)
same goes for the subpar digital filters that dont attenuate right at 22,05khz but at 24khz (which can be "fixed" by using good resampling in software), AKM seems to have the same issue which is a strange coincident
i just tried out the CS43131 as a pure single chip implementation inside the "Moondrop dawn pro" , doesnt sound that bad but it tends to sound a bit fatiquing and bass could have more drive, will try next either "Onix Alpha Xl1" or "Fiio KA15" with CS43198 + SGM8626 output amp and see how it compares
purely from this test i wouldnt mess around with the CS43131 ... tho i have to see what the other implementation above sounds like, currently also looking for a good portable headphone amp...
purely from this test i wouldnt mess around with the CS43131 ... tho i have to see what the other implementation above sounds like, currently also looking for a good portable headphone amp...
In AKM dacs (including AK4191+AK4499) filters are only -6dB down at fs/2. Latest ESS dacs have brickwall and apodizing filters that have proper attenuation (i.e. stopband starting at fs/2). Despite this many consider AKM dacs better sounding than e.g. ESS dacs.
in portable usb powered world you will never hear what are they capable of
Not necessarily with an oversampling audio sigma-delta DAC, which is designed solely for its 0..20kHz performance.
CS/AKM chips have a noisefloor rise above 40khz, mainly because of the general design allowing a more direct DSD path as i understand it
