• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Question about ESS DAC "resample"...

OP
P

PenguinMusic

Senior Member
Joined
Dec 20, 2019
Messages
457
Likes
208
Does it?
You send a 48 kHz file and the display of the DAC says it receives a 48 kHz file, not to be mistaken for "output".
Of course the output will never be 48 kHz because the output of a DAC is always analog.

Just an example, my veteran Benchmark DAC 1 has a AKM DAC and internally resamples every thing to 126 kHz.
According to the developers this simply gives the best results.

Please observe that "bit-perfect" is in general not about DAC's but about avoiding the operating system to degrade the sound.

NOS are often based on a Philips TDAxxxx design from the 80's.
Indeed the first generation was NOS and imho sounded horrible.
https://www.bramjacobse.nl/wordpress/?p=2545


Hi,

Well, I am totally unaware of how things work.
The input shows indeed what I feed it.

But the DAC also has a digital output.
If I hook that to another DAC, the input of the second DAC shows the same as the input as the first DAC.
But maybe that digital ouput is just a "passthrough" ?

Anyways, as I said, I like what I hear coming out of my Amp fed by a OS DAC.
So I think I can live with it... even though I now have a little more knowledge and won't naively think the sound comes out "unaltered"...

Regards.
 

Mnyb

Addicted to Fun and Learning
Forum Donor
Joined
Aug 14, 2019
Messages
980
Likes
1,268
Location
Sweden, Västerås
They do not really manipulate sound they manipulate the data about the sound to an internal format more suitable for easy transformation to an analog waveform afaik and the sound is not “changed” I would not consider it “audio” until in the analog domain .
It’s really data that describes the audio that was captured not the audio itself , that’s why digital audio works so well
 

Jimbob54

Master Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
7,304
Likes
8,668
Hi,

Well, I am totally unaware of how things work.
The input shows indeed what I feed it.

But the DAC also has a digital output.
If I hook that to another DAC, the input of the second DAC shows the same as the input as the first DAC.
But maybe that digital ouput is just a "passthrough" ?

Anyways, as I said, I like what I hear coming out of my Amp fed by a OS DAC.
So I think I can live with it... even though I now have a little more knowledge and won't naively think the sound comes out "unaltered"...

Regards.

Thats the beauty of measurements for electronics- they are taken at the outputs- so it doesnt matter what happens inside the sausage machine from an end user perspective- and makes all the marketing blurb, lists of exotoc sounding features and components irrelevant. If you want the purest conversion possible, look at the measurements of the outputs.
 

Jimbob54

Master Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
7,304
Likes
8,668
Hi,

Thanks a lot for your answers.

So basically : all "common" DAC's do in a way manipulate the sound, but that doesn't really mean that the sound is degraded.

And to answer the question : I took the AUNE X8 because, after reading a lot on another thread, the common consensus was : it a certain threshold is reached, it doesn't really matter and advice was to choose according to aesthetics and features.
As I wrote a little earlier, the Aune x8 matches my aesthetical tastes AND it matches the other devices from same manufacturer.

I though that DAC's would not manipulate sound. I now know this is not the case. I've looked a little at NOS DAC's. Unaffordable and probably not as versatile as OS Dac's... And probably not really audible difference.

Regards.

There is no "sound" before the DAC to manipulate- there is digital data . In fact, there is no "sound" after the DAC either - but analog voltage impulses - but at least an amp can amplify that and transducers convert that to "sound"
 
OP
P

PenguinMusic

Senior Member
Joined
Dec 20, 2019
Messages
457
Likes
208
Hi,

Thanks a lot for your answers.

OK, to play the devil's advocate.
DAC are fed with DATA. And they convert that DATA into "sound".

So I can say that if DATA is manipulated, the outcoming sound will be also.

But indeed, if that manipulation does not alter the sound, I can live with that.
I was somewhat naïve.
I thought that a good DAC should leave the DATA it receives as untouched as possible.
I will go to sleep a little wiser tonight :)

And I can live with that knowledge.
As long as what I ear, I like... and I sure do.

Regards.
 

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
4,465
Likes
11,905
Location
Mid-Atlantic, USA. (Maryland)
OK, to play the devil's advocate.
DAC are fed with DATA. And they convert that DATA into "sound".

Playing along...

It isn't converted into sound until it becomes a pressure wave at your speaker. Before that, it is simply an easily measurable electrical signal. That is what Amir measures when he runs his tests, after all the upsampling, oversampling (whatever happens in the little box) and converting from digital to analog. Those measurements indicate that whatever is happening, it isn't having an audible impact on the output, when compared to the original input.

It's easy to get lost in the search for pure clean sound. There's a lot that just won't matter.

I would join the chorus here and suggest that you try to worry less about it.
 

Pluto

Addicted to Fun and Learning
Forum Donor
Joined
Sep 2, 2018
Messages
956
Likes
1,497
Location
Harrow, UK
if DATA is manipulated, the outcoming sound will be also
Untrue. Or at the very least, a broadly generalised statement that does not hold water.

If you create an artificial digital signal that you know to be a mathematically perfect whatever and you design a DAC that reproduces that whatever as close to perfection as you can measure, what more do you actually want your DAC to do?
 
OP
P

PenguinMusic

Senior Member
Joined
Dec 20, 2019
Messages
457
Likes
208
Hi guys,

Thanks to all of those that took some time to reply here.

As stated I was naïvely thinking that the DAC would just get an input signal and convert it with no other form of processing.
And that if it did, it would alter the sound in a really effective and audible way.
I was obviously plain wrong.

My concern was that in the first DAC's I tried, there was the Cambridge Audio Magic DAC (the really first release when the world of DACs was barely emerging to the mass market). ANd Cambridge Audio made such advertisement about the abitlity of the DZC to "upsample the sound to 192khz"... I heard it. Lasted 10 minutes in my system compared to the sound of my Arcam CD29 integrated CD Player.

So of course, I like what I hear, but I thought this was because the signal was not receiving lots of treatments...

I will follow advices : I'll stop worrying about how it goes. And will simply enjoy what comes out of my headphones.
After all, that is what I'm looking for : sound I like to listen to...
 

JustAnandaDourEyedDude

Senior Member
Joined
Apr 29, 2020
Messages
479
Likes
716
Location
USA
Just would like to have a simple explanation about the "resampling" thing in ESS DAC's and the absence of that in AKM DAC's.


Does that mean that ESS "manipulate" the sound that it is fed with to make it "sound better" ?

I thought if you "upsample" or "downsample" or "resample" stuff, then you necessarily add/substract information from original signal... Thus the sound cannot be as it is meant to be.


Or, as this refers to "synchronous digital inputs" does that mean that the input signal coming in 44.1 is resampled to 48khz because that is what the DAC accepts (or the other way round) ?

Just to reiterate the excellent info provided by the previous responders to your questions, and to elaborate a bit only in order to combat my reputation for concision and brevity in my posts. There are three distinct resampling processes that you may be lumping into one in your mind. Firstly, there is the sample rate conversion that the Operating System of the source may do, either to convert the sample rate of the source files/stream to match the sample rate capability of the DAC to which they are fed (either internal DAC or external), or to enable mixing of sounds with different sample rates (by conversion to nearest multiple of 48kHz), or else to limit the bitrate of the digital stream output to S/PDIF Coax or TosLink. Any such sample rate conversion is done by the OS audio layer when it receives data from the music player software and prior to sending the resampled audio data to the DAC device driver. I assume this sample rate conversion is done by the OS using its own routines and the CPU hardware, and when done for mixing sound channels or for output to S/PDIF it is independent of the particular DAC chip. This is where the whole "bitperfect" hoopla comes in, and people who are into HiRes audio fret about it more. Consumers have done a whole lot of digging into how to prevent the OS from resampling and instead send the bitperfect data directly to the DAC. It turns out this is difficult and frustrating, and involves the OS, the music player software and the DAC's device driver. Under Microsoft Windows, you have to get the music player to use exclusive mode of ASIO or WASAPI, and under Android it is even more difficult to bypass the resampling of the Audio layer. I have not bothered looking into it so far, except that the software on some of my DAPs is designed for bitperfect feeding of the DAC chip/s.


Secondly, there is the "resampling" (more resembling a reclocking) that Amir referred to in his Topping D70s review. This resampling happens, if it does, on the DAC chip side during the audio data transfer from the source to the DAC chip or a receiver chip like AKM4118. This is the ASRC done by the ESS DAC chips (which include a hardware ASRC unit) to the USB/Coax/TosLink stream received by them, but not by the AKM DAC chips. Post #37 in the D70s review thread explains that the resampling is more artful than just using the hardware ASRC for clock domain synchronization, and is more like an on-the-fly improvement of the syncing of source and internal DAC clocks. This is done to correct timing errors (jitter) in the audio bitstream, and gives the ESS DAC chips their superior jitter performance on Coax and TosLink inputs relative to the AKM DAC chips. In Post #4 Vincent Kars showed how use of ASRC improves jitter performance. While the above is true, in practice the jitter performance of AKM DAC chips is still excellent, and the jitter is inaudible, so nothing to worry about.


Thirdly, there is the upsampling/oversampling performed by all modern delta sigma DAC chips, including the ESS, AKM and Cirrus Logic families, an upsampling that happens within the DAC chip itself as it processes the audio data that it received from the source. As Vincent Kars mentioned, this upsampling is at the very basis of their superior performance compared with alternative DAC chip architectures such as R2R ladder resistors. The delta sigma modulation takes advantage of the engineering happenstance that the state-of-the-art analog relative time resolution of electrical circuit clocks is far more accurate than relative voltage level resolution of the analog output circuits. The delta sigma modulation upsamples along the time axis, and on the face of it downsamples the signal amplitude (though it does not actually lose amplitude resolution but instead maps it into the higher time resolution). This is followed by digital noise shaping and an analog low-pass filter that gets rid of the ultrasonic noise caused by very rapid oscillation of the DAC chip's unfiltered analog output among the coarsened amplitude levels. This is drastic manipulation of the data bits, very far from leaving the data untouched, yet maintaining audio signal fidelity quite well, but the upshot of it all is that it works very well, and recent generations of delta sigma DACs have proven quite superior in transparent performance and cheaper to make, as compared with the best examples of alternative DAC architectures.


I don't want any up/re/downsampling. Can I have that on a device that money can buy (decently) ?
With substantial research, you can avoid the software digital resampling performed by the OS (the first type of sample rate conversion). The second type, the ASRC performed for sync improvement, is actually beneficial in reducing jitter in Coax and TosLink connections, and there is no sense in going out of your way to avoid it even if you had the option to do so. I do not think it actually changes the sample rate; my understanding of it is that it just adjusts the sampling instants to more reliably read the digital audio data from the input stream of the DAC. The third type of digital resampling, the upsampling performed in modern delta sigma DAC chips, is also beneficial and is the enabler for the fidelity with which they convert digital to analog. The use of a NOS DAC or selecting a NOS filter on a DAC defeats upsampling capability, and with lower bit-rate audio data, it will audibly degrade the sound fidelity (but may be perceived by its afficionados as improving the musical quality). With very high bit-rate audio data (HiRes files or streaming), a NOS filter will still degrade the sound fidelity, but the degradation would be inaudible. If you are bent on avoiding this third type of up/re/downsampling, you can do so by choosing a R2R DAC instead of a ESS/AKM/Cirrus Logic one. Take a look at the performance and price of an excellent recent discrete R2R DAC reviewed by WolfX-700: Measurements of Musician Pegasus R2R DAC.

So, to summarize, as the others have pointed out, your Aune X8 is audibly transparent, and should cause no worry about its manipulation of the audio data. Enjoy your Aune X8.

But the DAC also has a digital output.

If I hook that to another DAC, the input of the second DAC shows the same as the input as the first DAC.

But maybe that digital ouput is just a "passthrough" ?

Yes it is a passthrough (except converted to S/PDIF format), and unrelated to the data manipulation that would be done internally by the DAC chip if you were to use its analog output instead. High sample-rate/bit-rate input data to any DAC unit would be downsampled to the maximum allowed S/PDIF "passthrough" (96kHz/24-bit or 192kHz/24-bit, I forget which; mfr may set lower limit), which content owners stipulated to prevent bitperfect copying of HiRes audio. HDMI output would not limit the bitrate, but mandates use of HDCP, in order to prevent bitperfect copying.
 
Last edited:

TheTalbotHound

Active Member
Forum Donor
Joined
Nov 23, 2020
Messages
154
Likes
92
The resampling refers to Asynchronous Sample Rate Conversion or ASRC in short. Yes, it does manipulate the data as in it can no longer be bit-perfect, but it's 2020, sample rate conversion is accurate enough to not have any audible impact whatsoever. So any concerns about the signal not being the 'original' are really not warranted.
Manufacturers like Schiit who claim their mega combo burrito filter "do not touch the original samples" are really just taking advantage of the (unwarranted) suspicion that input data being manipulated is inherently bad. It's not. As long as the algorithms used are of high enough fidelity there is no concern.

Do ESS chips, like say the 9038 pro, asynchronously convert to one specific sample rate? Like if i feed it 768khz, will it downsample it internally to a lower sample rate or not?
 

skyfly

Active Member
Joined
May 1, 2020
Messages
174
Likes
44
Do ESS chips, like say the 9038 pro, asynchronously convert to one specific sample rate? Like if i feed it 768khz, will it downsample it internally to a lower sample rate or not?

I heard ESS chips does the usual (synchronous, integer multiple) oversampling first, and then does an ASRC. Please correct me if I was misinformed.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
3,884
Likes
8,727
Location
Hampshire
I heard ESS chips does the usual (synchronous, integer multiple) oversampling first, and then does an ASRC. Please correct me if I was misinformed.
That is what the datasheets say. What they actually do is anyone's guess, but I've seen nothing to suggest that the docs are wrong in this regard.
 

JohnYang1997

Master Contributor
Technical Expert
Manufacturer
Joined
Dec 28, 2018
Messages
7,005
Likes
15,776
Location
China
How do you know?
Any resources?
OSF Bypass
The oversampling FIR filter can be bypassed, sourcing data directly into the IIR filter. ESS recommends using 8 x fs as the
input. For example, an external signal at 44.1kHz can be oversampled externally to 8 x 44.1kHz = 352.8kHz and then
applied to the serial decoder in either I2S or LJ format. The maximum sample rate that can be applied is 1.536MHz (8 x
192kHz).

This should be the frequency being fed into the modulator.
 

DeepSpace57

Active Member
Joined
Jun 6, 2019
Messages
242
Likes
106
@Veri hi Mate,

i know you ve got AF200 usb interface. Do you think such an well designed interface affective to reducing jitter for AKM Design-DACs through Coax, Aes, to slink input? My mind tells effective as reclocking does not take place on the board. I am curious about your thought.also @JohnYang1997 , pls
 

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
8,054
Likes
9,815
@Veri hi Mate,

i know you ve got AF200 usb interface. Do you think such an well designed interface affective to reducing jitter for AKM Design-DACs through Coax, Aes, to slink input? My mind tells effective as reclocking does not take place on the board. I am curious about your thought.also @JohnYang1997 , pls
Hey. AF200 is not too expensive and works well. To say it reduces jitter though, unlikely. A good device should have on-board reclocking. I use AF200 with older devices over BNC/coax input.
 
Top Bottom