[charlesE]

Maybe the experiment in the Banjamin and Gannon paper you mentioned had jitter injected into a DAC pre-PLL, i.e. on the line and not the sampling clock?

 

[nick_charles]

Quote:

Originally Posted by charlesE /img/forum/go_quote.gif Interesting stuff, one thing doesn't seem to add up though (or maybe I'm just missing something). Figure 9 of this AES article: http://www.google.com/url?sa=t&ct=re...A_DAC-swN1Id-g Seems to indicate that Deterministic Jitter (sinusoidal) above 3kHz is audible at amplitudes of less than 100ps. The article states in section 3.3 "This plot can be used as a specification for allowable sampling jitter in Nyquist sampled systems. At 20 kHz the peak to peak sampling jitter must be less than 20 ps, increasing at 6dB per octave for lower frequencies until approximately 500 Hz where the limit is 1 ns. Below 200 Hz the jitter may up to 500 ns in amplitude before the sidebands could become audible." It references the article "Considerations for Interfacing Digital Audio Equipment to the Standards AES-3, AES-5, AES-11" also by Dunn's as a source but those values are off by about 3 orders of magnitude from the other figures that you quoted.

Dunn's model does make jitter audible at the low ps level for 20K tones, this is true. The thing to remember is that this is 1) A theoretical model , 2) A theoretical model based on 120db listening levels assuming a zero noise floor and 3) Not borne out by any empirical listening tests anywhere to date.

Dunn's model just does not take sufficient account for masking.

Mathematically jitter will be audible at 10ps on a 20K tone for a 16 bit system, mathematically jitter at a few 100s of ps will degrade bit-depth from 16 bits down to 15 bits, higher levels will drop it to down to 14 bits, nobody will dispute this.

But, it has yet to be shown that anyone can actually detect such low levels of jitter. Join the AES, (I did) buy the Benjamin and Gannon paper, see what the results were for real world listening, student membership is 30 dollars and the B & G paper is $5.

 

[charlesE]

Ah, yes, from the Dunn article I got the impression that Figure 9 was based on testing, but it's essentially a plot of Equation 1. I should have read it more closely. Thanks!

 

[bigshot]

Nick Charles, it's good to do all the groundwork on this, but you might want to sum up the information more often. Most people don't cross reference separate posts well, so your context risks getting lost in the shuffle. The average person looks at a bunch of numbers and science and assumes it adds up to something very important. It would be good to sum up frequently in layman's terms.

See ya
Steve

 

[nick_charles]

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif Nick Charles, it's good to do all the groundwork on this, but you might want to sum up the information more often. Most people don't cross reference separate posts well, so your context risks getting lost in the shuffle. The average person looks at a bunch of numbers and science and assumes it adds up to something very important. It would be good to sum up frequently in layman's terms. See ya Steve

Okay, here we go...

-Models of jitter audibility do not seem to concord with empirical listening
tests.
-Masking renders even large amounts of jitter undetectable
-Deterministic sinusoidal Jitter is easier to detect than random jitter.
-Deterministic sinusoidal jitter creates speciific distortion sidebands
-Jitter is more detectable when applied to pure high frequency tones than
lower frequency pure tones
-Jitter is less detectable in music than against pure tones
-No controlled listening test has shown that humans can detect any
frequency of determinisic sinusoidal jitter at an magnitude of less than 20ns
peak-to-peak against a music signal
- The above threshold (20ns) is not typical, a more representative figure for
sinusoidal jitter detection against music is about 100ns
- Random jitter manifests itself as low level noise that degrades fidelity by
lowering the effective resolution (bit-depth)
-No controlled listening test has shown that humans can detect random jitter
at an magnitude of less than 250ns peak-to-peak against a music signal

 

[Solan]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Solen, Jitter is the least important feature of a DAC. The DAC chip itself and the analog section make the sound. To say that jitter isn't important does not in any way shape or form mean that the rest of the DAC adds no value. Your statement just personifies the ignorant path we have been led down by the experts in the last several years. The choice of DAC chip and analog circuitry used has a HUGE impact on the sound.

I know. Slightly tongue-in cheek, since I'm not exactly buying a $50 DAC myself. But what I've noticed is that the big selling points for the multiple kilodollar DACs feature "jitter" rather prominently. So the returns may be not just diminishing, but non-existant, if you pay $1000 to get rid of that last bit of jitter.

 

[Solan]

What about recordings that vary sound levels a lot, from mellow and almost quiet to full burst. Orff's Carmina Burana is an excellent example. At the high end of the dynamic range, no jitter may be heard since the dB difference between music and jitter is so large, but when the choir is barely mumbling, this dB difference is far smaller and jitter may be audible, again justifying the need for some care when evaluating jitter's impact.

 

[tfarney]

Quote:

Originally Posted by Solan /img/forum/go_quote.gif I know. Slightly tongue-in cheek, since I'm not exactly buying a $50 DAC myself. But what I've noticed is that the big selling points for the multiple kilodollar DACs feature "jitter" rather prominently. So the returns may be not just diminishing, but non-existant, if you pay $1000 to get rid of that last bit of jitter.

Jitter may be the reason discussed in reviews and product literature, but I think the big audible selling point for multiple kilodollar DACs is coloration in the ouput stage. "Warm," "smooth," and "euphoric" are presented as the elimination of distortion. They are actually the introduction of it, though sometimes for good reason. Digital media reveals flaws in recording and mastering that we want to smooth out and cover up. Hardware manufacturers are more than happy to do that for us, but when the color is built into the gear, it colors the good along with the bad. Better to turn down the treble. Well, if we hadn't been convinced that any tone circuits in our gear introduce unacceptable levels of distortion, which is another thread, another issue, another time.

Tim

 

[nick_charles]

Quote:

Originally Posted by Solan /img/forum/go_quote.gif What about recordings that vary sound levels a lot, from mellow and almost quiet to full burst. Orff's Carmina Burana is an excellent example. At the high end of the dynamic range, no jitter may be heard since the dB difference between music and jitter is so large, but when the choir is barely mumbling, this dB difference is far smaller and jitter may be audible, again justifying the need for some care when evaluating jitter's impact.

With choirs you are not dealing with the higher ranges i.e 11K plus where jitter is more audible , once you are in the range of human voices jitter audibility becomes much more difficult, also masking effectively hides jitter at these frequencies.

As another example , a picolo note at 2.5K played at -20db with a 1700hz 300ns jitter signal produces one sideband at -80db (60db down) , 100ns of the same frquency produces a sideband a further 10db down, at 30ns there is no observable sideband.

I will close with a quote from Benjamin and Gannon

Quote:

Jitter induced by the digital audio interface was not found to be an audible problem for any of the program material auditioned.

i.e they could not find any digital audio device that had audible jitter.

As another example Keith Howard showed in a UK AES lecture the effect of jitter on flute music he applied both sine and random jitter to a music segment that had a fundamental at 1500hz he used a 2K jitter signal , for the sine wave he created sidebands at -44db on signal which seems pretty bad , but to get this the jitter had to be at 1000ns.

 

[bigshot]

Actually, I meant translating all the numbers and science into something that might help a 21 year old kid decide whether to drop a paycheck or two on buying high end audio equipment to reduce jitter in his system... Most people reading here don't have any idea what a nanosecond is or what a distortion sideband sounds like. You've done a great job of chasing down all the particulars, but unless that is put into the context of a real world practical application, it isn't going to help anyone.

Say I'm living in a dorm, and I have a $300 Dennon CD player, a headphone amp and a nice pair of Grados or Senns... What impact would attempts to reduce jitter in that sort of system have on the sound? Would an external DAC improve the jitter levels to a more acceptable level? Would connecting it with USB sound better because it has less jitter than optical? Do I need to reclock to get better sound? These are the sorts of questions most people here have about jitter. A listing of the technical specifications of jitter, no matter how accurate or precise doesn't really begin to answer those sorts of questions. You need to give more context to the data.

Your quote about jitter's effects on various types of music from Benjamin and Gannon above is a great start. Now sum up the context of all that science when it comes to typical home audio equipment.

Not criticizing... just suggesting.

See ya
Steve

 

[tfarney]

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif Actually, I meant translating all the numbers and science into something that might help a 21 year old kid decide whether to drop a paycheck or two on buying high end audio equipment to reduce jitter in his system... Most people reading here don't have any idea what a nanosecond is or what a distortion sideband sounds like. You've done a great job of chasing down all the particulars, but unless that is put into the context of a real world practical application, it isn't going to help anyone. Say I'm living in a dorm, and I have a $300 Dennon CD player, a headphone amp and a nice pair of Grados or Senns... What impact would attempts to reduce jitter in that sort of system have on the sound? Would an external DAC improve the jitter levels to a more acceptable level? Would connecting it with USB sound better because it has less jitter than optical? Do I need to reclock to get better sound? These are the sorts of questions most people here have about jitter. A listing of the technical specifications of jitter, no matter how accurate or precise doesn't really begin to answer those sorts of questions. You need to give more context to the data. Your quote about jitter's effects on various types of music from Benjamin and Gannon above is a great start. Now sum up the context of all that science when it comes to typical home audio equipment. Not criticizing... just suggesting. See ya Steve

I'm not Nick, but I don't understand all his numbers, so I may be a good candidate for a plainspeak summary:

Jitter. You can't hear it, so fuggetaboutit.

Tim

 

[nick_charles]

Quote:

Originally Posted by bigshot /img/forum/go_quote.gif Actually, I meant translating all the numbers and science into something that might help a 21 year old kid decide whether to drop a paycheck or two on buying high end audio equipment to reduce jitter in his system... Most people reading here don't have any idea what a nanosecond is or what a distortion sideband sounds like. You've done a great job of chasing down all the particulars, but unless that is put into the context of a real world practical application, it isn't going to help anyone. Say I'm living in a dorm, and I have a $300 Dennon CD player, a headphone amp and a nice pair of Grados or Senns... What impact would attempts to reduce jitter in that sort of system have on the sound? Would an external DAC improve the jitter levels to a more acceptable level? Would connecting it with USB sound better because it has less jitter than optical? Do I need to reclock to get better sound? These are the sorts of questions most people here have about jitter. A listing of the technical specifications of jitter, no matter how accurate or precise doesn't really begin to answer those sorts of questions. You need to give more context to the data. Your quote about jitter's effects on various types of music from Benjamin and Gannon above is a great start. Now sum up the context of all that science when it comes to typical home audio equipment. Not criticizing... just suggesting. See ya Steve

In short

As far as I know there is no reliable i.e carefully proctored experimental evidence to suggest that any currently available commercial digital audio device has any form of jitter at a sufficient level likely to cause empirically audible degradation in music played at normal to loud listening volume.

Jitter added by coax or optical cables or by the addition of external DAC devices does not appear to be at a sufficient level to be problematic.

EDIT: I did not see Tim's answer when I posted this, but on a purely personal level I concur with his assessment.

 

[tfarney]

Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif In short As far as I know there is no reliable i.e carefully proctored experimental evidence to suggest that any currently available commercial digital audio device has any form of jitter at a sufficient level likely to cause empirically audible degradation in music played at normal to loud listening volume. Jitter added by coax or optical cables or by the addition of external DAC devices does not appear to be at a sufficient level to be problematic. EDIT: I did not see Tim's answer when I posted this, but on a purely personal level I concur with his assessment.

Nick, you are a silver-tongued devil.

Tim

 

[Pars]

When and if you guys get around to playing around with optical vs. coax digital cables, try different lengths of coax, particularly one at or above 1.5m vs. a short one. See if they sound the same...

Quote:

Originally Posted by nick_charles Jitter added by coax or optical cables or by the addition of external DAC devices does not appear to be at a sufficient level to be problematic.

Cables don't "add" jitter. Impedance mismatches do. This involves more than the cable and the connector type.

 

[nick_charles]

Quote:

Originally Posted by Pars /img/forum/go_quote.gif When and if you guys get around to playing around with optical vs. coax digital cables, try different lengths of coax, particularly one at or above 1.5m vs. a short one. See if they sound the same... Cables don't "add" jitter. Impedance mismatches do. This involves more than the cable and the connector type.

I have done some limited coax vs optical tests, I did not find any significant differences with my DAC. I do have a few different runs of digital coax I can test out indirectly none are terribly short though, but I have no way of verifying whether they really are 75ohm or not, nor though do I have any way of measuring the resultant jitter or if any differences in waveforms are connected with jitter.