Battle of S/PDIF vs USB: which is better?

[RayDunzl]

I've found that being a bit skeptical, even of my own superpowers, serves me well.

 

[amirm]

Sounds like an opportunity for Amir to get a new 'scope and compliance packages that do eye diagrams and jitter separation...

That remains my last Christmas wish but alas, as you know, they cost a fortune. To measure USB one needs a 4+ Ghz scope, differential probes, etc. I could buy multiple Audio Precision analyzers for the same price!

S/PDIF measurements though are much more forgiving and my current 500 Mhz scope should be up to task. So if there is interest, I may take a shot at such measurements.

 

[amirm]

Amir, make sure you have a Glassfibre optic cable when measuring and not a cheap plastic one.

The last one I bought was a few years ago and bragged about how it was low jitter, etc. Have to see if there is a brand on it or not to figure out what it is.

 

[svart-hvitt]

The last one I bought was a few years ago and bragged about how it was low jitter, etc. Have to see if there is a brand on it or not to figure out what it is.

I guess the glass vs. plastic toslink debate is audiophile BS. Jenving/Supra sell 20 meter long plastic toslink guaranteed for 32/384.

http://www.jenving.com/products/view/zac-toslink-optical-20m-1003100235

 

[Fitzcaraldo215]

I guess the glass vs. plastic toslink debate is audiophile BS. Jenving/Supra sell 20 meter long plastic toslink guaranteed for 32/384.

http://www.jenving.com/products/view/zac-toslink-optical-20m-1003100235

Yes, but does anybody's digital interface circuitry at each end actually handle that?

Even so, I do not think it is the road to the future.

I once used a Theta Digital player and DAC connected by their single mode glass fiber, which was allegedly superior to the more common AT&T multi mode glass fiber. Theta DACS then went up to 192k, and the single mode fiber supposedly had much higher bandwidth than that. How high, I do not know. Nice connector, by the way.

 

[Blumlein 88]

I guess the glass vs. plastic toslink debate is audiophile BS. Jenving/Supra sell 20 meter long plastic toslink guaranteed for 32/384.

http://www.jenving.com/products/view/zac-toslink-optical-20m-1003100235

I have a 30 footer from Monoprice. It shows the same results at the analog out as the 6 footer. Meaning whatever contribution is from the cable is below other contributors to noise and such. I don't have any gear that does over 96 khz with Toslink inputs, but the 30 footer plays 96 with no issues with that gear.

 

[RayDunzl]

Yes, but does anybody's digital interface circuitry at each end actually handle that?

Not mine!

Some of my interfaces do 192kHz, others limited to 96kHz, which is moot since it all gets reshuffled to 48kHz in the DSP then reshuffled again to 211kHz in the DAC.

 

[svart-hvitt]

Yes, but does anybody's digital interface circuitry at each end actually handle that?

I guess this is BACK TO THE FUTURE stuff.

 

[Blumlein 88]

Trying to assess the effectiveness of a digital link by measuring the output of a DAC is a bit of a flawed experiment. It may appear to be the 'all in one' measurement, but it is the result of multiple mechanisms, including (as mentioned before) shifting clock frequencies, random grounding schemes, stray RF, etc.

Studying the characteristics of the link itself and the mechanisms by which various DACs function would be far more useful. Then you could actually predict what was going to happen worst case rather than just sucking it and seeing.

Yes and no. Studying the contributions from the link would best be done looking at measurements of how that link compares. OTOH, if the contribution is low enough it may not impact the measured output of the DAC in the analog world which is all we get to hear. One big factor is PLLs or ASRCs used on SPDIF inputs filter out much of the jitter from the datastream. That likely puts the contribution from that one source (the datastream link) well below most other factors. Look at Benchmark for a heroic version of that filtering. So it is a near certainty that whatever level of jitter is due to the toslink cable itself is higher in the 30 footer I have vs a 3 footer. Is the contribution of the optical transmitters and receivers well above that in terms of where jitter comes in? I don't know, but would think it might well be. Is the filtering good enough that the jitter resulting from optical cable and circuitry good enough the former becomes relatively inconsequential? Results at the output make it seem very likely that is the case except for some of the very early gear using optical.

 

[Fitzcaraldo215]

Not mine!

Some of my interfaces do 192kHz, others limited to 96kHz, which is moot since it all gets reshuffled to 48kHz in the DSP then reshuffled again to 211kHz in the DAC.

Yup. My old prepro limited me to 48k with Audyssey on, although it and HDMI were 192k and also DSD64 capable with Audyssey off. Neither Onkyo/Integra or Audyssey or others bothered to tell the general public in any specs. It only came out years later in measured testing of a top-of-the-line Marantz prepro by David Rich in the Secrets of Home Theater forum.

Potentially minimal sonic benefits of ever increasing sampling rates aside, I did not like the idea or the sleazy omission by manufacturers.

I now use Dirac in a PC mainly at 176k PCM for converted DSD material or a max of 192k with other PCM. I can't swear it sounds better as a result of higher sampling rates, but I sleep better at those higher sampling rates. Dirac is also a better tool than Audyssey was, even with their Pro kit.

 

[Cosmik]

One big factor is PLLs or ASRCs used on SPDIF inputs filter out much of the jitter from the datastream.

I agree with all of your comment - it's precisely what I meant by "studying... the mechanisms by which various DACs function".

 

[Deleted member 65]

I guess the glass vs. plastic toslink debate is audiophile BS. Jenving/Supra sell 20 meter long plastic toslink guaranteed for 32/384.

http://www.jenving.com/products/view/zac-toslink-optical-20m-1003100235

It may very well be if you exclude cheap ...

 

[svart-hvitt]

It may very well be if you exclude cheap ...

What is the definition of cheap?

Maybe I’m a big spender...

 

[amirm]

What is the definition of cheap?

Whatever Sal buys!!!

 

[Thomas savage]

Whatever Sal buys!!!

What you pay the poor guy who digs your weeds.... well until he threw a wobbly and quit that is

Don't give theses DAC's away , you can get a scheme going...,, weeds for DAC's

 

[Blumlein 88]

What is the definition of cheap?

Maybe I’m a big spender...

https://www.monoprice.com/product?p_id=1419

6ft Toslink $1.99 probably that much more for shipping. I use one of these on the satellite TV box.

Now what I used in my posted test measurement was Premium Monoprice. $3.74 for 6 ft. And I suppose what I have been calling 30 ft must actually be 35ft. That will inflict a wallet wobbling $5.65 price. Quite possibly the above cable only with that outer protective layer and metal square ends.

https://www.monoprice.com/product?c_id=102&cp_id=10229&cs_id=1022904&p_id=2831&seq=1&format=2

 

[Ron Party]

What you pay the poor guy who digs your weeds.... well until he threw a wobbly and quit that is

Don't give theses DAC's away , you can get a scheme going...,, weeds for DAC's

Years ago I traded a Sony LCD projector for 2 bottles of anejo.

 

[3beezer]

I am coming late to this conversation, but I thought that I could add a few comments based on my experience that might be interesting to some readers. First, with respect to Amir's initial post, I agree with everything he said about the pros and cons of USB relative to S/PDIF. From an architectural standpoint, it is a better design to put the master clock close to the DAC chip, as is the case with asynchronous USB. With S/PDIF, the need to extract the clock from the audio stream makes the performance of the DAC susceptible to imperfections in the signal received and in the clock recovery circuitry. I believed in these principles fervently until I began to notice about two years ago a growing body of comments from customers of superior sound quality using S/PDIF. I had initially dismissed these comments as delusional, but eventually I was forced to take notice. I set up a critical listening test using an S/PDIF interface that was the best I could devise. To my amazement, the sound quality was noticeably better using the S/PDIF interface -- noticeable to me and several others even though none of us considered ourselves to have golden ears. However, further tests with other DACs sometimes revealed contradictory results. I am not in a position to perform such tests over a cross section of commercial DACs. Nor am I in a position to reach definite conclusions about what it is in the design of each DAC that accounts for the difference since manufacturers do not disclose technical details of their products. My theory relates to one of Amir's other points: USB requires circuitry that is much more complicated than the circuitry required to implement S/PDIF. As he said, that circuitry must be isolated from the sensitive DAC circuit. That isolation certainly requires separate grounds and power, and it might also require shielding against EMI. I concluded two years ago that some manufacturers were getting this isolation right and others were not. One would like to believe that today more manufacturers -- maybe all -- are getting it right. Certainly the ExaSound DAC that Amir measured gets it right, as does the Benchmark DAC that someone mentioned in this thread.

On the other side, more manufacturers seem to be getting the clock recovery circuitry required for S/PDIF interfaces right these days -- witness the aforementioned ExaSound and Benchmark DACs. My current thinking is that it is not possible to reach a general conclusion about which interface sounds better, so the best solution is to try both (assuming that your DAC and your sound source support both) and decide for yourself whether one sounds better. However, I think that the market is heading for a nirvana in which the technical challenges have been solved and both interfaces sound the same. Even then, I still like USB better for the other reasons that Amir provided: It is ubiquitous, it is enduring, it supports higher functionality, and it is philosophically superior (which matters to geeks like me). Oh, and one other thing: S/PDIF uses RCA connectors. Because RCA connectors are an abomination, S/PDIF should die. If you like the simplicity of S/PDIF, we can talk about AES-3id (it uses BNC connectors).

 

[svart-hvitt]

I am not sure that i understand your question, as i was referring to measurements done roughly 20 years ago on devices comparing the TOSLINK and SPDIF interfaces while archimage did a comparison of optical cables (though i´ve to admit to only gave it a short glance).

As Vincent Kars link shows a difference between the interfaces migh be measureable even on more modern players.

But in any case it is important to remember that there might exist differences in jitter performance between the various interfaces even on the same device and that these differences might be even more pronounced on older devices.
Equally important it is to consider that the performance overall depends on the jitter (level and spectra) delivered by the source, additional effects coming from the connection and on the jitter transfer function of the DAC at the end.

Jakob,

Archimago posts the result data and charts of his measurements of toslink, coax spdif and usb in three different articles. A comparison between them should be possible by jumping from one article to the next.

Toslink has obviously poorer channel separation!?

But jitter is quite low, and below hearing limit?

 

[svart-hvitt]

Trying to assess the effectiveness of a digital link by measuring the output of a DAC is a bit of a flawed experiment. It may appear to be the 'all in one' measurement, but it is the result of multiple mechanisms, including (as mentioned before) shifting clock frequencies, random grounding schemes, stray RF, etc.

Studying the characteristics of the link itself and the mechanisms by which various DACs function would be far more useful. Then you could actually predict what was going to happen worst case rather than just sucking it and seeing.

The question you raise is a potential blow to many tests. In quantitative science - and the softer as well - you try and control for other factors to distill an essence which consists of only one ingredient.

When a cable goes through a (black...) box, you lose control in the strict statistical sense. Alas, even the connectors - male and female - complicate the picture.

So what is a cable tester to do?

Do test equipment and procedures exist that test the cable only, with and without connectors?