Why do some audiophiles spend thousands of dollars on a DAC? Are they searching for a "sound signature" they like, or just greater "accuracy"?

[mwhouston]

I’m sure with the amount of gear churning seen on SNA some just chases brand names. DACs, amps etc. I must admit my Audiophonics DAC hat on one of my Pi servers sounds different to my DIY DXD AK4497 DAC with my main Pi system. 
 

Both are very good but the AKM based DAC is superior. It should it has a substantial power supply and a superior process chip. Both under $600 new. 

[Anton Mullet]

Both and none of the above.

 

[muon*]

6 hours ago, aussievintage said:

 

Hmmm, and annoying.  Noone has actually put words to that "myriad" of other things though.  The problem is that vagueness suits those who don't like to commit  

Because it will go nowhere with you guys, you lot are cemented in what is and what is not and anything outside what you guys hold as being true is false, even if it may not be.

 

But you will work this to further cement the position you guys hold as true, and use it to attempt to discredit any other views.

 

Edit: It's like John says,  further discussion would be futile.

 

One main reason these threads never go well as it takes a strictly fixed objective line on a hobby that is largely subjective, for most.

I believe you guys when you might say "I heard no difference" but I also believe guys that say "I heard a difference" as different brains work differently in regards to perception, and no testing can be done that is not subject to confounding variables IMO.

 

Brings back memories of that WP guy who was of the belief that all dacs and cd players sound the same, Simon was his name I belive.

 

 

Edited November 6, 2020 by muon*

[Cafad]

I would say that "greater accuracy" is a sound signature.

[aussievintage]

1 hour ago, muon* said:

you lot are cemented in what is and what is not and anything outside what you guys hold as being true is false, even if it may not be.

Great, now I have been accused of being on both sides of this opinion -  and NEITHER is true.

Edited November 6, 2020 by aussievintage

[MLXXX]

19 hours ago, Cafad said:

I would say that "greater accuracy" is a sound signature.

I wonder where accuracy (rendering the digital signal as a faithful analogue signal) takes you  in terms of a "sound signature".  On audiophile forums I've often seen the view expressed that there is no "neutral" sounding DAC, that they all sound different.

 

However because even cheap modern DACs have such good linearity it doesn't seem a big stretch that "accurate" audiophile DACs would tend to converge towards a single neutral sound.

 

To my mind there would be the proviso that at a low sample rate, such as the rate used for standard CDs of 44.1kHz, there is some flexibility in how the high audible frequencies are reconstructed.

 

Thinking about CDs, they only have 16 bits of bit depth, and DACs can perform down to 20 bits. Theoretically you could get a DAC to play a 16-bit CD PCM stream, capture  and record that analogue output with an accurate ADC, and discover that the recording was bit for bit identical to the CD.  [To facilitate that happening, the ADC would probably need to be clocked in sync with the 44.1/16 PCM stream sent to the DAC.]

 

Edited November 7, 2020 by MLXXX

[aussievintage]

17 minutes ago, MLXXX said:

I wonder where accuracy (rendering the digital signal as a faithful analogue signal) takes you  in terms of a "sound signature".  On audiophile forums I've often seen the view expressed that there is no "neutral" sounding DAC, that they all sound different.

 

However because even cheap modern DACs have such good linearity it doesn't seem a big stretch that "accurate" audiophile DACs would tend to converge towards a single neutral sound.

 

To my mind there would be the proviso that at a low sample rate, such as rate used for standard CDs of 44.1kHz, there is some flexibility in how the high audible frequencies are reconstructed.

 

Thinking about CDs, they only have 16 bits of bit depth, and DACs can perform down to 20 bits. Theoretically you could get a DAC to play a 16-bit CD PCM stream, capture  and record that analogue output with an accurate ADC, and discover that the recording was bit for bit identical to the CD.  [To facilitate that happening, the ADC would probably need to be clocked in sync with the 44.1/16 PCM stream sent to the DAC.]

 

 

If you did record something bit for bit identical, then the immediate thought would be the DAC should have no sound signature.  However, if you take that analogue output from your recording ADC and connect it to a preamp or power amp, it is still possible the sound might change due to different loading and interaction with the new system.  

 

That does, though, mean that it is something of a coincidence that the loading and interaction with the recording ADC just happened to be perfect conditions for getting a bit for bit accurate recording.    Too much of a coincidence maybe?

 

Has anyone ever tried to do this re-recording of a DAC's analogue output to see what numbers they get, and how close they are?

[MLXXX]

28 minutes ago, aussievintage said:

Has anyone ever tried to do this re-recording of a DAC's analogue output to see what numbers they get, and how close they are?

That sounds like a question for a DAC expert. Presumably we might have some DAC experts on this forum [who actually get to read this thread even though it is in The Great Audio Debate part of the forum!] able to throw light on that question.

[mwhouston]

Ive been reading Stereophile mag for about 20 years and notice of late they are talking 0.0005% or better distortion from DACs. I know that doesnt always mean they sound good but it has to lend some weight. 

[akjono]

To contribute to the OP's question - I do know that I myself enjoy greater accuracy - details in a recording. Subtle nuances and spatial information of soundstage, not only more Bass and  scintillating vocals and higher highs etc. . I also assume that my system without a Dac can sound fine and enjoyable. (In the past I had a cassette player and turntable set up with the same components). I cannot answer why others like to spend $1000s on Dacs or etc... . I do know that many folks are bamboozled by stats and extremely low distortion as stated above. And of course the latest and greatest. But if one is not completely happy or satisfied with the status quo of one's music system, then the search continues. 

 

Everything we do/add/subtract to our system be it cables/dacs/sources will effect the final product = sound. But the most subjective factor of the equation is us. Human beings. As my friend Jim- who owned a String Shop down the hall from my office told me- (Guitars/violins/violas - sales and repairs) - once told me - we are the "most changing" instrument. Moods/attitudes/emotions affect us in greater ways than most imagine. Over the years I played $5000 Martins from his wall, and lesser known and priced guitars as well. $$ did not always dictate the better sound to my ears. (He knew I wasn't going to buy the Martin, though I eventually bought a Gibson Gospel arched-back). Suffice it to say our "internals" may change more rapidly than our externals (system). 

 

As @Cafad said above - maybe " greater accuracy" is a sound signature. I may be in that camp, but I do enjoy Timbre and Tone, snappy bass, and healthy snappy cymbals (and their decay) withing the soundscape. And I do know I want to enjoy what my ears hear, not particularly "measurements" of said equipment in a controlled environment that isn't in my house. Dac On! 

 

 

[aussievintage]

2 minutes ago, akjono said:

I do know that I myself enjoy greater accuracy - details in a recording.

 

2 minutes ago, akjono said:

 

As @Cafad said above - maybe " greater accuracy" is a sound signature. I may be in that camp, but

I do enjoy Timbre and Tone, snappy bass, and healthy snappy cymbals (and their decay) withing the soundscape.

 

Sounds like you and I both enjoy the same type of thing when it comes to the "sound".    I don't think what you are describing is necessarily  accuracy.  I call it just "detail"  ** , a word you even used in the first sentence above. Accurate, low distortion detail, always impresses me in a system though.

 

** I  only make the distinction because a little distortion can sometimes fool you into thinking there is more detail. otoh When the detail is accurate as well, then we are talking good sound.

 

[akjono]

@aussievintage Detail it is! As in the "Devil is in the details". I can live with that. Accuracy can be more subjective in regards to music, I agree for a multitude of reasons, distortion as you said one of them.

 

I will be seeing a mate in the next few days with a new addition, a dac, for his system, the Denafrips Ares 2.... his "old dac" is from 2013. Even the usb connection technology has made lightspeed evolution since then. It will be Interesting... . 

[davewantsmoore]

On 06/11/2020 at 3:25 PM, muon* said:

Because it will go nowhere with you guys, you lot are cemented in what is and what is not and anything outside what you guys hold as being true is false, even if it may not be.

 

But you will work this to further cement the position you guys hold as true, and use it to attempt to discredit any other views.

I'm not quite sure I underatand what you are trying to say here.

 

If you I do understand your words correctly.... then you couldn't be further from the truth

[kukynas]

9 hours ago, MLXXX said:

That sounds like a question for a DAC expert. Presumably we might have some DAC experts on this forum [who actually get to read this thread even though it is in The Great Audio Debate part of the forum!] able to throw light on that question.

I’m no expert other then knowing how to turn coffee machine on (with wife’s help of course) but if you tell me what to record I can do that

[MLXXX]

1 hour ago, kukynas said:

but if you tell me what to record I can do that

 

To ensure conditions are perfect for achieving bit to bit accuracy,  I think it will be necessary to be able to sync the recorder clocking with the PCM stream sent to the DAC. Can your recording device be synced in that way?

 

Nyquist tells us we can successfully sample a continuous wave provided the sample rate is no less than twice the frequency of the wave.  However the test I have in mind is not merely whether a waveform has been adequately sampled and later on successfully reconstructed.  It is a more specific test requiring that the ADC doing the sampling samples at exactly the same points in time relative to the analogue waveform cycles originally sampled.  That is only way the ADC will measure identical instantaneous waveform amplitudes.

 

Let me explain further. If you were to record a piano recital at a sample rate of 48kHz using ten ADCs of the same model running independently and with all connected to the analogue output of a microphone mixer, no two of the ten digital recordings would be the same at the bit level, even though they should sound the same as each other when played back.  Similarly if you play a CD transport ten times and record the DAC output ten times with an ADC that is not phase-locked to the CD transport stream, you will get 10 recordings that are each different at the bit-level, even though they should sound the same as each other when played back.

Edited November 7, 2020 by MLXXX

[kukynas]

Yes and no, my ADC RME ADI 2 PRO fs can control the DAC with spdif out unfortunately I don’t have 2 external DACs with spdif in at the mo, I could add RME DAC portion into the test but I don’t know how the FPGA’s clocking its own DAC and spdif output so can’t guarantee 100% consistency, I would assume that DAC output stage distortion would dominate but of course different clocking could cause unwanted variances and therefor 2 external DACs with spdif in would be prefered, now each and every DAC has differently implemented spdif input so you might be evaluating the DAC Input stage capabilities instead of it’s output quality and consequently judging its quality based on wrong measurement, I think the best comparison would be to control 2 USB in DACs by asio drivers from PC and record both again with asio drivers of the ADC, unfortunatelly my external DAC doesn’t have USB in so can’t do that either

[MLXXX]

An extract from the specs:

 

SPECS

ADI-2 Pro FS

 

ANALOG INPUTS

• XLR
• Input: XLR, servo-balanced Input sensitivity switchable +24 dBu, +19 dBu, +13 dBu, +4 dBu @ 0 dBFS
• Signal to Noise ratio (SNR) @ +13/19/24 dBu: 120.x dB RMS unweighted, 124.x dBA
• Signal to Noise ratio (SNR) @ +4 dBu: 119 dB RMS unweighted, 123 dBA
• Frequency response @ 44.1 kHz, -0.1 dB: 5 Hz – 20.5 kHz
• Frequency response @ 96 kHz, -0.5 dB: 3 Hz – 45.5 kHz
• Frequency response @ 192 kHz, -1 dB: 2 Hz – 92.7 kHz
• Frequency response @ 384 kHz, -1 dB: < 1 Hz – 124 kHz
• Frequency response @ 768 kHz, -3 dB: < 1 Hz – 180 kHz
• THD @ -1 dBFS: -116 dB, 0.00016 %
• THD+N @ -1 dBFS: -112 dB, 0.00025 %
• Channel separation: > 110 dB

TRS
As input XLR, but:

• Input: 6.3 mm TRS jack, servo-balanced, fully compatible to TS and RCA

___________

 

 

It's interesting to read the product description:

 

"RME’s anniversary reference AD/DA converter is a host of devices all put together into one unit, with a simple and mostly automated way of using it right out of the box. A unit with the industry’s biggest footprint per feature ratio, excellent tech specs, crystal clear and transparent sound signature, and two extremely powerful headphone outputs."  [my emphasis]

 

 

This does indeed seem to be a very high calibre device, suitable for recording the analogue outputs of DACs.

 

2 hours ago, kukynas said:

I could add RME DAC portion into the test but I don’t know how the FPGA’s clocking its own DAC and spdif output so can’t guarantee 100% consistency, I would assume that DAC output stage distortion would dominate but of course different clocking could cause unwanted variances

 

In a quick test with my own audio interface, a Behringer FCA1616, I found that 10 separate recordings of the playback of a source appeared to be in phase lock with each other as regards sample values, when using one of the audio interface's own DACs connected by analogue cable to one of the interface's analogue inputs.

 

 

_________

 

A the end of the day, some people might use an "any level of jitter may be audible" argument to dismiss a DAC that performs accurately down to the 16th bit and beyond. There is no answer to an "anything is possible" argument. 

 

However the bit test I have outlined could certainly enable DACs to be sorted into those that can at least deliver the bits with a flat frequency response and in a linear matter.  [I note that for testing linearity and frequency response using test waveforms you wouldn't actually need phase-locking between the DAC and the ADC.]

 

There are those who argue that the ability to reproduce test waveforms perfectly does not prove that music signals can be reproduced perfectly. So there would still be something persuasive about being able achieve a bit-perfect rendition of music from a CD track, compared with just a bit-perfect rendition of test waveforms.

 

 

_____________________

 

@kukynas, it occurs to me that the fact your ADC can operate at up to 768kHz might make it possible to make an asynchronous  recording at that very high sample rate and then "parse" the recording successfully so as to extract samples at say 44.1kHz or 48kHz with a very low phase discrepancy relative to the sampling of a reference waveform in the recording.

 

For example, insert 10,000 cycles of a sinewave at 11.025kHz or 12kHz at the start of the test file to be played back by a DAC, and manually examine a 768kHz sample rate recording of the analogue output of that DAC to find the timing of the recurring zero point crossing of the recorded test waveform.  Those crossing points would provide a reference for the extraction process needed for the remainder of the test file.  This technique would require that the absolute clock speed of the DAC did not drift appreciably (relative to the [different] absolute clock speed of the ADC).

 

Edited November 7, 2020 by MLXXX

[kukynas]

35 minutes ago, MLXXX said:

This does indeed seem to b a very high calibre device, suitable for recording the analogue outputs of DACs

 

yup, I'm using it for low voltage level audio device measurements, you won't find many ADCs on the market capable of such performance (cost no object) 

 

41 minutes ago, MLXXX said:

 

@kukynas, it occurs to me that the fact your ADC can operate at up to 768kHz might make it possible to make an asynchronous  recording at that very high sample rate and then "parse" the recording successfully so as to extract samples at say 44.1kHz or 48kHz with a very low phase discrepancy relative to the sampling of a reference waveform in the recording.

 

For example, insert 10,000 cycle of a sinewave at 11.05kHz or 12kHz at the start of the test file to be played back by a DAC, and manually examine a 768kHz sample rate recording of the analogue output of that DAC to find the timing of the recurring zero point crossing of the recorded test waveform.  Those crossing points would provide a reference for the extraction process needed for the remainder of the test file.  This technique would require that the absolute clock speed of the DAC did not drift appreciably (relative to the [different] absolute clock speed of the ADC).

 

if you can compile the test file, tell me which soft you'd like to use for playback/recording I can run it through and send you the output files

[kukynas]

51 minutes ago, MLXXX said:

A unit with the industry’s biggest footprint per feature ratio

 

that's because they using 10 layers PCBA, 10 freaking layers..., no issue with grounding and shielding, very short signal path, no parasitic noise etc.... you can see the outcome on measurements I did in another thread, how it handles incoming signal and PSRR  

[MLXXX]

1 hour ago, kukynas said:

 

yup, I'm using it for low voltage level audio device measurements, you won't find many ADCs on the market capable of such performance (cost no object) 

 

 

if you can compile the test file, tell me which soft you'd like to use for playback/recording I can run it through and send you the output files

 

Ok. Tonight I downloaded the 16 bit original CD version [in FLAC format] of the last title shown on the webpage HiRes Download - test bench.  That last title  is  Joseph Haydn: String Quartet In D, Op. 76, No. 5 - Finale - Presto, played by the Engegård Quartet.

 

Using Audacity, I added 1 second of 11,025Hz sinewave at maximum amplitude (0dB) to the beginning of the 44.1kHz 16 bit stereo file. I then saved the whole file, without dither, as a  44.1kHz 24 bit stereo file. The lowest 8 bits should be zeroes.

 

Here is a download link to the test file with introductory 11.025kHz tone:  Google Drive link to test file.

 

I will leave it to you to decide which software to use to send the 24 bit 44.1kHz PCM stream to a DAC.  Something that will send the digital stream intact without altering the bits is needed.  (I guess it would be possible to burn a CD just truncating the bottom 8 bits if no other method is available, and rely on a CD player transport.)

 

Similarly please use your own discretion as to which software to use for the recording.  If it is possible to sync your ADC to the stream sent to the DAC then a 44.1kHz sample rate would be a good sample rate to try.  If syncing is doubtful or not possible then a high sample rate could be chosen.  A multiple of 44.1kHz could make later extraction of bits easier, e.g. 441kHz would be convenient if that sample rate can be set.

 

The files created would best be in 24 bit integer format to facilitate comparison with the test file.

 

Obviously an initial level setting exercise would be useful so that 0dB output of the DAC corresponds to 0dB for the DAC (or slightly less).  If levels cannot be exactly set for 0dB then post-recording gain matching could be done in software at a 24 bit or better level of precision.

 

_____________

 

I did do some initial internet searches to see what this type of testing using a music file has revealed in the past, but couldn't find anything. Perhaps most people are content with test tone type verification that a DAC output is flat across the audible spectrum, and low in spurious products.  Or perhaps the syncing of the ADC to the DAC for bit-by-bit comparisons creates an impression of artificiality.

 

[kukynas]

I've just ordered Topping E30 from Amazon so hopefully could do the proper comparison next weekend, both have 0.5db volume increments so should be able to level match it except one is balanced and the other single ended, wanted to buy some PC DAC with volume control and usb powered for long time so it'll serve both purposes    

[MLXXX]

6 hours ago, kukynas said:

I've just ordered Topping E30

Extract of specs:

 

Output Voltage	2Vrms@0dBFS
THD+N	<0.0003% @1kHz A-weighting
Noise	<2uVrms @A-weighting
Crosstalk	-130dB@1kHz
SNR	>121dB@1kHz
Dynamic Range	119dB@1kHz
Channel Balance	<0.3dB

 

 

If the output of the E30 in combination with the gain of the ADC can be adjusted so that both channels are just under 0dB for the ADC prior to recording, it should be possible to make a fine adjustment to the levels in "post-production" to bring both channels to exactly 0dB.

 

Over the next few days I might try some recordings with my Behringer interface. It is not as highly specced as your ADI-2 Pro FS interface but despite that may be able to  give bit-for-bit results down to the 16th bit, over most of the audible range, when operating at only 44.1kHz. Ordinarily I record at 48kHz.

 

_____________

 

In passing I'd note that many integrated amplifiers these days routinely convert all analogue input into digital, process digitally, and then convert back to analogue.

[jt301]

On 06/11/2020 at 10:02 AM, MLXXX said:

One thing that prompted me to begin this thread was a test I'd had in mind of sending a PCM stream to a DAC and recording the output of an ADC connected to the output of that DAC by a short length of audio cable; but where the clocking of the ADC was synced to the clocking of the PCM stream sent to the DAC.

 

It seems very possible in that scenario to obtain 100% accuracy down to the 16th bit, over the whole audible range with the possible exception of frequencies near to the Nyquist limit, or at the bottom of the audible range.  However if the test did return a bit identical result all the way down to the 16th bit for almost the whole of the audible range, that would mean nothing to audiophiles chasing a distinctive sonic signature. To them, "mere accuracy" would not be good enough.

Exactly what are you trying to prove here?

[MLXXX]

4 hours ago, jt301 said:

Exactly what are you trying to prove here?

I had been  planning to test whether even a moderately priced DAC can produce an analogue waveform that is so faithful to the digital signal that when that waveform is captured with an accurate ADC all 16 bits for each sample for each channel of a CD can be recovered without loss.

 

That got me thinking as to whether that bit-perfect performance would satisfy high end DAC users in terms of a DAC for use in conjunction with a CD transport, or whether they were seeking more than "mere" bit-for-bit perfection.  Hence the title of this thread.

 

Edited November 8, 2020 by MLXXX

[eLicky]

Its a strange hobby, I spent $2k on RME-ADI2 to use with HD800 and let it go.

I now use an ODAC / O2 amp I picked up used for $100 and find it more dynamic and just as detailed with the same headphone.

Edited November 8, 2020 by Ewok209