Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › Why is SPDIF better than USB?
New Posts  All Forums:Forum Nav:

Why is SPDIF better than USB? - Page 3

post #31 of 121

Imo from subjective listening experience, the way reclocking improves sound quality may not be easily audible, not as obvious as between dac's and amp's but not as subtle as vibration control differences. I believe it is more similar in degree of difference as recabling from a typical stock cable to a proper gauge and length cable, or adding a good power conditioner. I would not recommend a $100+ usb converter/reclocker unless the rest of your setup is ironed out, especially cables since you can DIY them for much cheaper than the cost of most reclockers, just buy a few dozen feet of 24 awg and 12 awg conductor size wires and hijack stock cable connectors. Imo a full recable from stock yields quite more benefit than jitter reduction.

post #32 of 121

 

Originally Posted by nick_charles View Post

I am still out over $100 on cables bought and resold at a loss or not resold (who wants to buy a cable when the seller tells you openly that they think it will make zero difference


so that's why all the audio gear reviews are always so raving? like the DacMagic that should sound horrid when you see what's in it, but only obscure blogs have the guts to call it no better than a uDAC.

 

Originally Posted by nick_charles View Post

 

You want to lend me a Hiface (as a matter of ethics I will not buy something knowing that I will return it) I'll happily run the tests you suggest and also measure the outputs which would be much more rigorous than my hearing as you could argue I did not want to hear the differences.

[..]

But let's make this more interesting. How about this. I will buy a Hiface with RCA Coax and test it against the raw USB using a 24/96 ADC on the analog outputs. I'll publish the graphs


I don't have an Hiface, and considering that I have a Firestone Bravo I would have no use for it.

 

And I'm not sure what you'd plan on measuring exactly? These transports are bit-perfect by essence, you need to measure their jitter to see a difference...we both know how expensive this kind of measuring equipment is. I can honestly say that the Bravo on its Supplier DPS w/ a short coax cable would sound far clearer and focused than your current toslink connection or the PCM2704 chip in your Zero. The tighter the incoming jitter, the more it'll chew the S/PDIF receiver job...CS8416 will do a far better job at recovering the S/PDIF clock off WM8804. I'm sure you could source those units from a shop w/ a return policy...and it's got nothing to do w/ ethics as far as I can tell, most shops clearly state that they have a return policy...it's not illegal or anything, it's the usual "try it before buy it" term.


Edited by leeperry - 9/16/10 at 6:49am
post #33 of 121
Quote:
Originally Posted by leeperry View Post

 

 

And I'm not sure what you'd plan on measuring exactly? These transports are bit-perfect by essence, you need to measure their jitter to see a difference...we both know how expensive this kind of measuring equipment is. I can honestly say that the Bravo on its Supplier DPS w/ a short coax cable would sound far clearer and focused than your current toslink connection or the PCM2704 chip in your Zero. The tighter the incoming jitter, the more it'll chew the S/PDIF receiver job...CS8416 will do a far better job at recovering the S/PDIF clock off WM8804. I'm sure you could source those units from a shop w/ a return policy...and it's got nothing to do w/ ethics as far as I can tell, most shops clearly state that they have a return policy...it's not illegal or anything, it's the usual "try it before buy it" term.


If the level of jitter makes a difference it will show up in the FR as differences in distortion in the decoded (analog) audio stream and it will be measurable as a deviation from the reference, it may be a small deviation i.e 10ths , 100ths or 1000ths or even 10,000ths of a db at any point but it will be there in the analog stream.

 

I know,  I have measured the FR on samples with different levels of jitter it does show up here and it is easy to say with confidence which sample has more jitter, assuming that the jitter level is the only thing altered and you have some reference point, I demonstrated this on a UK HiFi forum where I correctly assesed (without listening) which of 5 samples had which amount of jitter, between 0 and 100ns. It does not matter that you cannot easily measure the jitter itself as it appears downstream as it were. 

 

If you do not trust my word read Benjamin and Gannon, they did the same thing as one of their tests,  they added jitter and plotted the FR deviations for different levels of jitter. Again they did not find jitter audible until several 10s of ns were injected, but the differences were graphically apparent.

 

I know most stores have a return policy, most stores will also not sell as new a customer return and it is an inconvenience to reshelf and to lose money earned on a return. It is a matter of intent, I don't do "try and then return" when I am giving the seller the impression he has made a sale.

 

I did approach several cable companies with an open  try and then return proposition for my cables test , strangely I got not one reply, except for Warren Audio but he declined in the end once I explained my protocol.

 

 

 


 

post #34 of 121
Quote:
Originally Posted by nick_charles View Post


If the level of jitter makes a difference it will show up in the FR as differences in distortion in the decoded (analog) audio stream and it will be measurable as a deviation from the reference, it may be a small deviation i.e 10ths , 100ths or 1000ths or even 10,000ths of a db at any point but it will be there in the analog stream.

 

I know,  I have measured the FR on samples with different levels of jitter it does show up here and it is easy to say with confidence which sample has more jitter, assuming that the jitter level is the only thing altered and you have some reference point, I demonstrated this on a UK HiFi forum where I correctly assesed (without listening) which of 5 samples had which amount of jitter, between 0 and 100ns. It does not matter that you cannot easily measure the jitter itself as it appears downstream as it were. 

 

If you do not trust my word read Benjamin and Gannon, they did the same thing as one of their tests,  they added jitter and plotted the FR deviations for different levels of jitter. Again they did not find jitter audible until several 10s of ns were injected, but the differences were graphically apparent.

 

I know most stores have a return policy, most stores will also not sell as new a customer return and it is an inconvenience to reshelf and to lose money earned on a return. It is a matter of intent, I don't do "try and then return" when I am giving the seller the impression he has made a sale.

 

I did approach several cable companies with an open  try and then return proposition for my cables test , strangely I got not one reply, except for Warren Audio but he declined in the end once I explained my protocol.

 


 

 

You keep suggesting the same tests over and over ... do you understand the difference between frequency (what you are suggesting) and time domain? Jitter is essentially in the time domain, so it is pointless and stupid to suggest a test using only the frequency response.

 

I bought the Benjamin and Gannon paper (following your advice on another thread) and I find that paper pointless. It is an old research study that was conducted poorly. None of the equipment being used was of "audiophile" quality. How can you suggest that a paper that use unknow quality dac and headphone amps as being accurate?

The headphones being uses were cheap, entry level $50 sony headphones. I don't judge equipment by the price but there is a point where you can't make definitive and serious generalizations when the test equipment is so subpar.

 

Many audiphiles and (especially) spend a lot of time and money choosing the best dac chip, opamp, capacitor, piece of wire, power cord, headphone amp, headphone cable... all of which are considered as useless in such studies.

If you want to apply logic and science, you should, for the sake of generalization, assume that all the audiophool claims are true (even if you think they are not).

In order to prove that jitter do not exist, you cannot assume that all interconnects, opamps, capacitors, dac chips sound the same as long as they "measure" the same. You are just using a bunch of flawed studies to validate another flawed study.

When I will start seeing studies about jitter conducted in the same rigourous and open minded way as Kunchur conducted his studies (see here: http://www.physics.sc.edu/~kunchur/informat.htm#papers), then I will start taking those studies seriously.

It is not because you follow a "scientific" method that you always end up with the truth. When making the wrong assumptions (about what is audible and not) and using cheap and non representative equipment, it is easy to imagine how flawed such studies can be.
If you read Kunchur's paper, you will see that in order to test the temporal resolution (time domain) of humans, he couldn't find a CD player that was good enough. He had to use a high quality analog wave generator to test the human ears at the highest frequencies. What it means is pretty obvious if you can't (or don't want to) understand it: the human ear is better than most CD players out there. If CDs were good enough, he wouldn't have needed to look for something better.
Of course, he didn't just assume that a cheap $200 "pro" monitors were good enough because some cheap studios use it. He used a high quality ribbon tweeter and custom made electronics.

If he had used poor equipment (because they are good enough according to him) like Benjamin and Gannon, he would have reached a different conclusion.

Nick Charles, I really hope that one day you will try to do some serious research on the subject and not just throw at people outdated research about jitter. Recommending to people to buy an outdated research paper that used the wrong equipment and wrong assumptions doesn't reflect very well on you.

If you can find a paper that conducts a research using "serious" equipment (DCS, Esoteric ... dacs; Stax, HD800, LCD2, T1 ... headphones) then I will take those papers seriously. Untill then it is a bad advice to recommend people spending their money on a poor and outdated study, even if it one of the only ones on the subject.

post #35 of 121

For the case of Gamma2, I'm having difficulties discerning the sounds between S/PDIF and USB which is understandable considering the implementation in that particular DAC.

post #36 of 121

 

Originally Posted by nick_charles View Post

If you do not trust my word

 

I never said I didn't, and I'll be looking forward any measurements you would make...I'm not the one who needs to be convinced [:pieruslu]

post #37 of 121
Thread Starter 
Quote:
Originally Posted by K3cT View Post

For the case of Gamma2, I'm having difficulties discerning the sounds between S/PDIF and USB which is understandable considering the implementation in that particular DAC.


You are the first I've heard say so.  Since it would save me the additional cost of buying an adapter and cable, that's what I *want* to hear.  If you can discuss it in layman's terms (I see you are a DIY'er, and I'm just a listener), what do you mean by "considering the implementation in that particular DAC."  And is the Gamma 2 implementation likely to be similar to other USB implementations in the price category (in your opinion)?

post #38 of 121

I found the same thing. The USB and the SPDIF in the gamma 2 sound very similar. I have another 8471 the Ibasso D10 and on that the USB and the SPDIF sounds very different. The SPDIF is clean while the usb not so much without any extension in the top end.

post #39 of 121

What he means is the upsampler in the gamma 2.

post #40 of 121
Quote:
Originally Posted by baglunch View Post


You are the first I've heard say so.  Since it would save me the additional cost of buying an adapter and cable, that's what I *want* to hear.  If you can discuss it in layman's terms (I see you are a DIY'er, and I'm just a listener), what do you mean by "considering the implementation in that particular DAC."  And is the Gamma 2 implementation likely to be similar to other USB implementations in the price category (in your opinion)?


I'm still learning but I will try to explain it with the best of my abilities. In Gamma2, the PCM2707 (that's the USB receiver chip) is set to convert the digital data from USB to S/PDIF. This S/PDIF data is then sent to the CS8416 (that's the digital receiver chip) for clock recovery and conversion it to the raw I2S format. Finally, the I2S data is fed to SRC4192 (or AD1896 depending on the builder's choice) which serves as the DAC's Asynchronous Sample Rate Converter (ASRC) chip to re-clock and upsample the data to 24/96 before final conversion to analog data via the DAC chip, Wolfson WM8740/8741.

 

Basically you can see that all the data formats all end up as the raw I2S data before the final upsampling and D-to-A processes and from my understanding, this final asynchronous upsampling process is the reason why the discrepancies between the different inputs are minimized and how most of the jitters are eliminated. Even the designer himself is skeptical of the benefits of using popular asynchronous USB transports other than the ability to play 24/96 via USB. I suppose the use of ASRC chip also sets the Gamma2 apart from other DACs in its price range as that's the edge that truly makes it high-end.


Edited by K3cT - 9/16/10 at 12:33pm
post #41 of 121

 

Originally Posted by K3cT View Post

 

Finally, the I2S data is fed to SRC4192 (or AD1896 depending on the builder's choice) which serves as the DAC's Asynchronous Sample Rate Converter (ASRC) chip to re-clock and upsample the data to 24/96 before final conversion to analog data


Yes, many testimonials seem to confirm that ASRC kills the whole point of using a low jitter transport...but many ppl believe that a mandatory upsampling stage is uncalled for, only degrading and coloring the sound. Most recent DAC chips already include a very capable oversampling stage.


Edited by leeperry - 9/16/10 at 1:01pm
post #42 of 121

Good old lee eh? He might talk a load of bollocks but at least it's our bollocks.

 

Here is a good one.

 

 

Quote:
to answer your OP question...most USB inputs run off a 16/48 max PCM270x chip that sounds craptastic compared to any serious S/PDIF interface.

 

Here is a list of audio interfaces and the chips they contain. Not a lot of PCM270x.

 

http://carplay.it/VB/showthread.php?t=11668

 

The moral here is :-

 

Do not buy 'audiophile' grade gear any more. Like lee has noticed they are effectively a washed out industries pathetic attempt to sell you a cheap CD player split across three or 4  separate boxes (transport. DAC, digital in, power supply, headphone amp) all costing more than single a pro audio device which will do the same job, have better converters (usually AKM ) and include a whole load of useful extras.


Edited by RonaldDumsfeld - 9/16/10 at 1:10pm
post #43 of 121
Quote:
Originally Posted by RonaldDumsfeld View Post

Good old lee eh? He might talk a load of bollocks [..] 

Here is a list of audio interfaces and the chips they contain. Not a lot of PCM270x.

 

http://carplay.it/VB/showthread.php?t=11668


I wonder which one of us two talks the more bs tbh...most pro manufacturers will never disclose what they use in detail(your link only talks about the DAC chips, duh). And when a USB input is limited to 16/48, it's a good guess that it's using a PCM270x chip....these chips will always sound crummy in many ppl's experience.

 

Now, I'm talking from personal experience...are you? you feel like one of those so-called myth debunkers.

 

TC Electronic have their own jitter reduction technology, and indeed these might end up being excellent transports...at least they have a real R&D department, not like all the horror stories you can read about those Zero DAC's.


Edited by leeperry - 9/16/10 at 1:26pm
post #44 of 121

 

Quote:
when a USB input is limited to 16/48, it's a good guess that it's using a PCM270x chip....these chips will always sound crummy in many ppl's experience.

 

That's because the PCM270x generation of chips are cheap and don't require the manufacturer to do anything that might cost him money or require up to date skills and experience (like writing drivers).

 

So. Like I said. The answer is. Don't buy an obsolete 'audiophile' grade DAC. Buy one of the many on that list I posted that use a modern generation of USB.

 

Why spend $150 on an aftermarket USB > S/PDIF device to extend the life of a poor quality DAC. Spend the same money on a brand new DAC instead. Simples.

post #45 of 121
Quote:
Originally Posted by slim.a View Post



 

You keep suggesting the same tests over and over ... do you understand the difference between frequency (what you are suggesting) and time domain? Jitter is essentially in the time domain, so it is pointless and stupid to suggest a test using only the frequency response.

 

Yes I know the difference, thanks. Read Bob Adams (Analog devices) papers on jitter. He demonstrates how any jitter will be apparent in downstream distortions. B and G also very clearly show that jitter causes distortion in the frequency domain, you have seen the graphs yourself.,. See also Stereophile's measurements of jitter they always show jitter as deviations in the frequency domain as well, why because that is where the problem occurs from the listener point of view, it is the distortion sidebands and noise in that frequency domain that would be audible (except they are not in fact audible) not any ps or ns time shifting itself which is way beneath human discrimination.

 

 

I bought the Benjamin and Gannon paper (following your advice on another thread) and I find that paper pointless.

 

Okay I'll bite, why ?

 

 

It is an old research study that was conducted poorly. None of the equipment being used was of "audiophile" quality. How can you suggest that a paper that use unknow quality dac and headphone amps as being accurate?

 

Old does not mean redundant unless better studies have come along to challenge their findings, the only later real jitter research puts thresholds much higher (admittedly random , not signal correlated) , just as a thought there are research papers from the 1950s and 1960s that are still relevant today.

 

 

The headphones being uses were cheap, entry level $50 sony headphones. I don't judge equipment by the price but there is a point where you can't make definitive and serious generalizations when the test equipment is so subpar.

 

You obviously do use price as part of your judgment or you would not mention the price of the Sonys, which as I understand it have an admirably flat FR for headphones and low distortion, two characteristics I would think would be vital for serious tonal discrimination tasks, no ?

 

 

 

Many audiphiles and (especially) spend a lot of time and money choosing the best dac chip, opamp, capacitor, piece of wire, power cord, headphone amp, headphone cable... all of which are considered as useless in such studies.

If you want to apply logic and science, you should, for the sake of generalization, assume that all the audiophool claims are true (even if you think they are not).

In order to prove that jitter do not exist, you cannot assume that all interconnects, opamps, capacitors, dac chips sound the same as long as they "measure" the same. You are just using a bunch of flawed studies to validate another flawed study.

When I will start seeing studies about jitter conducted in the same rigourous and open minded way as Kunchur conducted his studies (see here: http://www.physics.sc.edu/~kunchur/informat.htm#papers), then I will start taking those studies seriously.

 

?????? - this is all about superconductors not audio ?

 

Kunchur's paper is interesting, he uses a source with 68ns jitter, so he obviously does not consider jitter a big problem or he would get better kit . I would be more interested though if he used music and not square waves. He does get some things a bit wrong though, for instance JND at 69 db is about 0.35db not 0.7 (Backus)  so is perilously close to the level differences caused by speaker displacement, but his lack of music based listening tests is the real flaw for us. I am also concerned about the effect of moving two powerful electromagnets (hum ?).  Even with all this he moves the threshold down 3 microseconds from 9 to 6 under extraordinary conditions that no-one outside of a lab will ever experience, is this a big deal ?

 

It is not because you follow a "scientific" method that you always end up with the truth. When making the wrong assumptions (about what is audible and not) and using cheap and non representative equipment, it is easy to imagine how flawed such studies can be.
If you read Kunchur's paper, you will see that in order to test the temporal resolution (time domain) of humans, he couldn't find a CD player that was good enough. He had to use a high quality analog wave generator to test the human ears at the highest frequencies. What it means is pretty obvious if you can't (or don't want to) understand it: the human ear is better than most CD players out there. If CDs were good enough, he wouldn't have needed to look for something better.
Of course, he didn't just assume that a cheap $200 "pro" monitors were good enough because some cheap studios use it. He used a high quality ribbon tweeter and custom made electronics.

If he had used poor equipment (because they are good enough according to him) like Benjamin and Gannon, he would have reached a different conclusion.

Nick Charles, I really hope that one day you will try to do some serious research on the subject and not just throw at people outdated research about jitter. Recommending to people to buy an outdated research paper that used the wrong equipment and wrong assumptions doesn't reflect very well on you.

 

Can you demonstrate to me empirically that jitter is audible at levels lower than that found by B and G ?.

 

Can you prove to me empirically that jitter detection is significantly improved by "serious" kit ?

 

If you can find a paper that conducts a research using "serious" equipment (DCS, Esoteric ... dacs; Stax, HD800, LCD2, T1 ... headphones) then I will take those papers seriously. Untill then it is a bad advice to recommend people spending their money on a poor and outdated study, even if it one of the only ones on the subject.


Edited by nick_charles - 9/16/10 at 6:56pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Dedicated Source Components
Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › Why is SPDIF better than USB?