R2R vs Delta Sigma DACs - same results?

[Vintageadict]

@Roland68

Thanks for the reply. That is what I have been thinking. I am shopping DACs now and may just use the Yamaha as the CD transport. I have a Cambridge CXN-V2 for streaming and will be able to try the new DAC with the CD player and the Streamer. I have no issues with the Yamaha as a transport and a top loading cd player doesn't work well in my component rack. I did try using the Cambridge DAC for CDs by running a digital out from the Yamaha into the Cambridge while having the rca out from the Yamaha still hooked up to my pre. Toggling back and forth between the two inputs produced no discernable results. Seems like technology may have passed some of my digital equipment.

 

[NTTY]

Interesting thread.

I just wanted to come back to the claim I've seen repeated so often that with 16bits and shape dither, we can reach 20bits resolution. Well, actually no, unless we go for a very narrow passband, in which case 16bits undithered signal reaches 18bits.

Let me illustrate with digitally analyzed test tones.

This is 16bit undithered, 999.91Hz @0dBFS:

The software is a little optimistic, but close enough. SNR should be closer to 98dB (6.02*N + 1.76 = 98.08dB), but ok.

Now let's use triangular dither from Audacity that uses only half a bit:

We indeed lost half bit of resolution (and 3dB in SNR).

Now again the same sine but with shape dither:

If the noise is accounted for up to 20kHz, then SNR is reduced to 83.8dB and ENOB is down to 13.6bits.

The only way to make the above 20bits resolution is to high pass at 200Hz and low pass at 2kHz... In which case we indeed get 20bits, because not much is taken into account, as far as noise is concerned:

And if I do the same with the initial undithered test file, then I get 18bits:

All that said, in my CD players testing, I use shape dither test tones and a restricted span (20Hz to 6kHz), to see what the internal DAC is capable of, which is a max 18.7bits as illustrated below:

Actually, I don't run this test at full scale, but at -12dBFS, because that level is discriminatory between R2R and Delta Sigma DACs, per my experience. And so:

ENOB is calculated at 16.7bit + 12dB (2bits) = 18.7bits too.

Now, real life examples.

The Denon DCD-3560 with one of the best ancient R2R conversion I've encountered:

That is 15.2bits + 2bits = 17.2bits, 1.5bit short of the possible maximum.

And the excellent SMSL PL200:

16.7bits + 2bits = max 18.7bits with that test.

Cheers

 

[Vintageadict]

Excuse this question but trying to understand the value of what you are showing. So is the point in these measurements to show that the DS configured DAC in the SMSL player can achieve a higher bit resolution than the best CD player with an R2R DAC that you have access too? I'm not a measurement guy so trying to keep up here.

 

[voodooless]

I just wanted to come back to the claim I've seen repeated so often that with 16bits and shape dither, we can reach 20bits resolution. Well, actually no, unless we go for a very narrow passband, in which case 16bits undithered signal reaches 18bits.

Nobody claimed if was for the full bandwidth Obviously shaped dither is a trade off.

For fun, do A-weighting on all of these. You’ll see how much of a trade-off it is to our hearing.

In any case, you won’t have that much DR at high frequencies anyway, because music doesn’t contain that much energy there.

 

[danadam]

Nobody claimed if was for the full bandwidth Obviously shaped dither is a trade off.

For fun, do A-weighting on all of these. You’ll see how much of a trade-off it is to our hearing.

Exactly. The claim is, I presume, more in terms of how much difference it makes perceptually. It means, when you set volume such that you barely hear the flat dither, how much can you increase the volume until you start hearing the shaped dither. In the attachment there is an example of high-shibata from SoX:

This is flat dither with 20 dB gain and high-shibata dither with 40 dB gain. The first 4 seconds are to calibrate the volume, so that you barely hear it. Then you should also only barely hear the high-shibata, at least that's how it is for me.

20 dB difference is more than 3 bits but less than 4 bits, so that's not full 20 bits with shaped dither, but maybe there are better (more aggressive ) shaping filters out there.

 

[Strumbringer]

I'm intrigued, curious, and perplexed by all the R2R interest. Elsewhere, these DACs are all the rage and apparently are being manufactured and purchased like there's no tomorrow.

I'm no audio engineer, but what I've learned and read is that the R2R/NOS category provides more distortion, introduction of more noise/aliasing, significant high-frequency roll-off, and worse SINAD. Do I have that all right? I've never heard an R2R DAC, but if I wanted these features in my RME ADI-2, I suppose NOS mode and playing it through a tube amp is the rough idea?

The analogy that instantly comes to my mind is if I wanted to go back to a 4:3 television and watch reruns of Gilligan's Island. I suppose this is nostalgic and desirable to some.

That being said, I need to hear one of these R2R "warm" or tube-buffered DACs.

 

[Mart68]

've never heard an R2R DAC,

You probably have, they were commonplace in CD players back in the 1990s.

 

[solderdude]

Yes, but these all were oversampling (4 to 8x) and had proper reconstruction filters.

The so called 'NOS' DACs we see today omit that filter and do not oversample.
They basically (mis)use the chip (or a poor performing discrete version of it) and claim 'superior sound', more 'analog' (whatever that may be).

Of course there are also proper (oversampling with good digital filter) R2R DACs (and derivatives) that, in a blind test, are indistinguishable from any DS DAC.
There even are some out there that rival DS DACs in harmonic (and IM) distortion.

 

[Roland68]

Yes, but these all were oversampling (4 to 8x) and had proper reconstruction filters.

The so called 'NOS' DACs we see today omit that filter and do not oversample.
They basically (mis)use the chip (or a poor performing discrete version of it) and claim 'superior sound', more 'analog' (whatever that may be).

Of course there are also proper (oversampling with good digital filter) R2R DACs (and derivatives) that, in a blind test, are indistinguishable from any DS DAC.
There even are some out there that rival DS DACs in harmonic (and IM) distortion.

This reminds me of how many blind tests we've had with (audiophile) guests, all of whom claimed they could easily distinguish analog from digital sources. They were mostly wrong, at best 50/50.

I own a CD recording with superbly digitized tracks from vinyl that don't exist in digital form. Everyone who heard it in a blind test swears it's analog and not played from a CD.

 

[oleg87]

The analogy that instantly comes to my mind is if I wanted to go back to a 4:3 television and watch reruns of Gilligan's Island. I suppose this is nostalgic and desirable to some.

Not even, aside from the power of suggestibility. I used some 90s r2r Burr Brown chip in my system for years and years, it wasn’t anything other than sonically transparent. Now it’s the latest fashion in an industry driven by fads.

 

[cay-uwe]

Can I assume that exceptional execution of the design while incorporating quality parts can make a difference in a DACs presentation?

My experience is that a technically "perfect" component is not necessarly the one one likes better.

Reacently I started measuring different kind of CD-Players / DAC, ranging from about EURO 150,00 - 4.000,00. In a listening session, in which I did not told the listener which DAC / CD-Player was used, a friend of mine liked the DAC for EURO 150,00 best. An over 35 years old PHILIPS CD-Player he ranked as #2. Both did not measure very good, altough from audible point of view they were very good, meaning very low THD, THD+N, etc.

 

[Joe Smith]

Just recently swapped in a 16khz only handmade dac from about 15 years ago ("Hot Audio," made in Canada) that only does 16/48 max, still sounded quite good. Uses the TI PCM2705 chip, I believe. No sonic difference for my old ears from my current Schiit and JDS Labs dacs. It had the salutary effect of making me NOT order a new "mesh" dac from Schiit, I'd be chasing pixie dust.

 

[AndyBell]

A while ago, I did a semi-blind, volume matched, test between the FiiO K11 R2R in NOS mode and the Topping dx2 pro plus using its filter #2. If I understand things aright, that’s comparing the K11 at its worst to the Dx2 at its best.

I played a range songs I knew well and had my set up such that I could switch between the DACs whilst listening to them. I knew when the change was made but didn’t know which DAC was playing.

It may be that my hearing at my age isn’t that sensitive but I could not hear any difference on any song between the two DACs. Based on what I had read (elsewhere) I was expecting significant differences between the two. Now I’m on ASR I understand my test results better…

 

[Svend P]

It may be that my hearing at my age isn’t that sensitive but I could not hear any difference on any song between the two DACs. Based on what I had read (elsewhere) I was expecting significant differences between the two. Now I’m on ASR I understand my test results better…

That makes sense. You don't have to read all 547 pages to get an idea of why your result is no surprise, but this could give you a better understanding:

How can DAC's have a SOUND SIGNATURE if they measure as transparent?

I know there are people out there that think cables affect sound, which is much worse, but there really is no response to something like that, but just to smile and nod. But what about people who talk about DACS as if they were headphone drivers or speakers, and talk about the SOUNDSTAGE...

www.audiosciencereview.com

 

[Roland68]

A while ago, I did a semi-blind, volume matched, test between the FiiO K11 R2R in NOS mode and the Topping dx2 pro plus using its filter #2. If I understand things aright, that’s comparing the K11 at its worst to the Dx2 at its best.

I played a range songs I knew well and had my set up such that I could switch between the DACs whilst listening to them. I knew when the change was made but didn’t know which DAC was playing.

It may be that my hearing at my age isn’t that sensitive but I could not hear any difference on any song between the two DACs. Based on what I had read (elsewhere) I was expecting significant differences between the two. Now I’m on ASR I understand my test results better…

The people who claim to hear such large, audible differences between DACs wouldn't pass a blind test. But their excuses for suddenly not hearing the differences would be spectacular.
I've even been accused of swapping CDs or files during such tests.

And in reality, it has little to do with their hearing. My father, at 65 years old and with a significant hearing loss between 3000 and 6500 Hz, could reliably hear subtle differences more reliably than trained listeners.

 

[AndyBell]

That makes sense. You don't have to read all 547 pages to get an idea of why your result is no surprise, but this could give you a better understanding:

How can DAC's have a SOUND SIGNATURE if they measure as transparent?

I know there are people out there that think cables affect sound, which is much worse, but there really is no response to something like that, but just to smile and nod. But what about people who talk about DACS as if they were headphone drivers or speakers, and talk about the SOUNDSTAGE...

www.audiosciencereview.com

Yes, I’ve read some of that.

I repeated the same test with a variety of DACs including some USB dongle DACs. Even when the tests were sighted I couldn’t hear any differences provided the volumes were matched.

My background is IT - I’m a c++ programmer currently in the semiconductor industry where measurements and data trump opinion and feelings.

I did go into my tests expecting to hear differences and was surprised to find none. My first tests were sighted - but I’m a very analytical sort of person, so maybe that helped to combat expectation bias. Then again, these devices are not that expensive, so maybe there was less pressure - had I spent £1000s I surely would have expected more…

I did the same tests with upsampling and had the same results.

It’s a nice place to be - I’m no longer searching for ‘better’ gear in this area. Based on feedback from this site, I added a small subwoofer to my small office - I could certainly hear that! And I’ve invested in a measurement mic and have done some room correction. I can hear the difference there too, but I think I’ve a way to go in that area.

 

[cay-uwe]

The people who claim to hear such large, audible differences between DACs wouldn't pass a blind test.

Some years ago I had a SOEKRIS R2R in house for testing. In those days a friend of mine visited me and he is blind. So there was no need for me to hide something while switching between the DACs I had in those days. He was surprised when I switched to the SOEKRIS R2R and found it to sound "cleaner", less stressy and with a fuller bass when compared to the build in ESS Sabre DAC in my cocktail Audio X40 has. Actually, I had the same impressions...

For me it gets very obvious when listening to actual Pop & Rock recordings, which actually are recorded at a very loud level. Practically with any DAC those recordings sound horrible, but a bit "cleaner" across a R2R DAC.

 

[peng]

I appreciate your explanation. So the blanket statement, "all DACs sound the same", which is often encountered is not necessarily true, correct?

Often encountered, is that even true? You may be right, but I don't recall seeing such blanket statement ever made by serious ASR members though. Some DACs in the below $35 may not sound the same as DACs in the $100-$200 range. Such "blanket" statements, if ever made about DACs, should not be taken seriously.

 

[voodooless]

In those days a friend of mine visited me and he is blind. So there was no need for me to hide something while switching between the DACs I had in those days.

Of course you have to. Being blind does not absolve you from bias.

 

[Svend P]

He was surprised when I switched to the SOEKRIS R2R and found it to sound "cleaner", less stressy and with a fuller bass when compared to the build in ESS Sabre DAC in my cocktail Audio X40 has. Actually, I had the same impressions...

Level matched?