Will a good external DAC make an Average Transport sound fantastic? (1 Viewer)

[Chu Gai]

once you get below 2 nanoseconds it's inaudible and companies are routinely achieving 200 picoseconds...the usual 10x less thing. ya got's to listen real real hard to pick up 2ns and that's with test tones.

 

[ChrisHeflen]

And just to quell any thought about my honesty...That is not in my living room unfortunately.

Sorry to interrupt the real topic. Please forgive.

 

[RichardHOS]

Brett, isn't Toslink just the optical form of S/PDIF?

 

[Michael R Price]

Chu,

I see what you mean about 2ns being a very low amount of jitter. Haven't tried listening for it myself but I can imagine that's inaudible. How's jitter measured, anyway?

Would this mean that all transports sound the same once they get below a certain jitter threshold?

 

[Craig_Kg]

once you get below 2 nanoseconds it's inaudible and companies are routinely achieving 200 picoseconds...the usual 10x less thing. ya got's to listen real real hard to pick up 2ns and that's with test tones.

You're looking at jitter in the wrong way. A 2ns jitter that is audible through a test tone is in the area of wow and flutter. Much lower levels of jitter will smear the sonic image that your mind tries to assemble, as the harmonics get out of alignment with the fundamentals of the signals. The biggest result is poorer imaging.

As for the topic, YES. I heard an excellent CD player (Arcam FMJ32 CDT) get blown away by a Chord DAC64 - up until then, I was a 'digital is digital' guy.

 

[Brett DiMichele]

Rich,

Yes essentialy TosLink and Coaxial both fall under the
Sony/Phillips Digital Interface but "high end" components
rarely ever use Optical. Optical has some advantages like
no worries about cable impedance issues or EMI interference
but there are drawbacks like needing to convert the Light
into digital pulses and then recieve those light pulses and
reconvert back into bit stream data. I guess people say
this is a good way to add more error to the mix (More complicated).

 

[Craig_Kg]

I can never understand how optical is considered to require extra conversion when the signal in a digital coax cable is modulated - either way, the digital signal is converted to analog for transmission and converted back when received.

 

[Brett DiMichele]

Craig,

And I am by no means against optical. It's just that certain
people are and most of the high end has followed suite.

Thier conjecture is that with coax you simply have an
electrical signal comming out of the transport and into the
DAC. And with Optical you have an electrical signal comming
out of the Transport being converted to pulses of light and
then sent down a fiber line where it is turned from light
back into an electrical signal. Is it more complicated? I'd
say yes.. But does that mean it is worse quality? I doubt
they could prove it to me.. Or anyone else...

It probably all boils down to Coaxial cables are thicker and
look better.. Ya know "Fat" cables = Quality..

 

[Craig_Kg]

PCM through SPDIF coax is definitely not just 0's and 1's sent as square edge transitions else you could not reliably extract the clock from the signal and the receiver would treat a silent portion (all 0's) the same as if the cable was removed - you can see from the display that it still recognises a PCM source signal. Although, then again, it could be Manchester encoded.. time for some more web surfing...

I think high end audio people are just used to copper(/silver/niobium/beryllium...) and distrust optical connections because they are new.


Whoops - I was wrong about the format: http://www.epanorama.net/documents/audio/spdif.html. It is square edged pulses - this actually gives the consideration of high-end digital cables more credibility.

 

[Chu Gai]

Well Craig, the research that I'm aware of doesn't support the audibility below 10 ns...oops I'd posted 2 ns above. Consider the following.

In AES Preprint Number 4826: "Theoretical and Audible Effects of Jitter on Digital Audio Quality" by Eric Benjamin and Benjamin Gannon, they found that so long as the DIR can lock on to the signal, that jitter was not a problem.
This would suggest things like jitter reduction boxes, while potentially effective, are worthless from the point of audibility. Plastic or glass toslink, for typical lengths used by an individual, is irrelevent.

Their findings were that the threshold for jitter is ~10 ns, peak to peak, for a sine wave. For musical sources, it's around 30-300 ns.
The reason why it's higher with musical sources has to do with masking in which case the jitter must exceed the masking threshold in order for it to be perceptible. Keep in mind here, that we're talking about 'just noticeable'.
One of the great things about toslink, is if you happened to have a very problematic EMI situation, toslink, and it's inexpensive, would be immune to it.
The following paper...
http://www.nanophon.com/audio/1394_sampling_jitter.pdf provides further information.

The 'j test' as initially developed by Julian Dunn, who passed away earlier this year, is currently one of the ways jitter's being looked at. Arcam, I believe uses it.

Now 'if' there are issues of audibility between two players, and they're not the result FR aberations or the result of people 'tweaking' and replacing components with completely unknown results, then perhaps we need to look elsewhere for the reasons. It was Dunn's thoughts, probably based upon measurements, that crosstalk from the servo into either the DAC reference or output posed a more significant problem.

Suffice it to say that issues that vinylphiles willingly accept and embrace would be torn into shreds by those using CD's. IMHO, CD players are quite remarkable for their general quality.

 

[Lewis Besze]

once you get below 10 nanoseconds it's inaudible

Pssssssttttt!
Don't tell that to Lee!

 

[Chu Gai]

Lee's a good guy and after all, what's audio to many people but sometimes picking a position and then agonizing over it? However the science would advance if the high-end mongers would apply science.
Time to put my CD player on chicken bones now and listen to some reggae

 

[Brett DiMichele]

Chu,

Don't forget to jiggle the cables.. Remember any electrons
left inside them will cause phase cancelation issues do to
the remainder of the free radicals contained within the
golden ratio quasi-cqonstant-q conductors.

 

[Lee Scoggins]

once you get below 2 nanoseconds it's inaudible and companies are routinely achieving 200 picoseconds

More misinformation from Chu. :rolleyes

Chu, we have been over this before-read my postings and respond in real debate. I have posted two key papers on jitter and how it has been PROVEN that it is audible to 2o picoseconds.

The Gannon paper is old research, Julian Dunn's paper is the relevent one.

I am tired of debating with people who don't want to learn or study the literature.

By the way, this is why transport quality matters even at the mid-priced level. Combined with DAC and Analog Output Stage quality, you have the three big sonic factors.

 

[Chu Gai]

Awww, now I've got to go and retract my he's a nice guy :frowning:

I mean, here I try to overlook that you constantly raise Georgie and AQ 'white papers' to the level of competent research. When little kids come to you and talk about their imaginary friends and discoveries, one smiles and says that's cute. When adults do it then it becomes anything but cute especially when the payoff is not your smile, but the handing over of your Visa card...don't leave home without it!

The Dunn paper to which I'm assuming you refer to (http://www.nanophon.com/audio/jitter92.pdf) was presented on Oct. 1992. Let's look at it a little more carefully. It speaks of 20 ps jitter at 20 kHz under a worst case sinusoidal situation. You do understand that these results as well as many other such audio thresholds were determined using headphones. I take it you further understand that these thresholds change for the worse once we deal with music and that it further changes for the worse once we move into listening with speakers with ambient noise floors that are significantly worse than headphones. Certainly you don't think either our hearing abilities are spectacular at 20 kHz or that there is significant high energy information at that particular frequency? Further that 20 ps # came from a playback level of 120 DB!!!! Want a perspective on 120 dB? A jet aircraft taking off with afterburners off an aircraft carrier at 50 ft was about 130 dB. My you must have an interesting CD collection.

Now let's fast forward to the Benjamin & Gannon paper which you say is old. After all, it can't be the Dunn paper, that's where you got your 20 ps from. That was done in 1998 and has NOT BEEN REFUTED.

Now that esteemed audio rag, StereoPhool, had this to say in their January 1999 issue.

For sine wave signals, the threshold of audibility of jitter in the experiment was 10 nanoseconds RMS at 17kHz and 100 nS RMS at 4kHz. (One nanosecond = 1000 picoseconds.)

The period of a 4 kHz tone is 250 microseconds, so, expressed as a percentage, we are talking 10 nanaoseconds divided by 250,000 nanoseconds, or 1/25,000.
This is 1/250th of a percent, or 0.004 percent. Expressed as a dB ratio, this is 88 dB.

Check out the jitter on this el-cheapo portable Panasonic CD Player. Damn, the jitter's 20 dB down from what Dolby found to be audible!!

Like I said, audiophiles love to agonize over everything.

Let us all drink a toast to the wondrous PT Barnum shall we?
After all it was he who said "No one ever went broke understimating the American public."

 

[Lee Scoggins]

Want a perspective on 120 dB?

Want a perspective of 120db. Visit my basement when I am playing Dark Side of The Moon.

Chu,

I like your comments and your quotes are almost always hilarious...but it is frustrating when you state matter of factly that 10 ns is the audible limit when Julian (rest his soul) has found different. Since we have discussed these papers numerous times and I have made copious postings on jitter as a subject, would it be unreasonable to at least throw in something like "some like Julian Dunn feel the limit may be 20 ps"? I spoke to him before his unfortunate death and he said that many researchers had been lowering the jitter limit of audibility every year to now around 20PS.

I am sorry if I sent a real angry reply above, but this is something where I have done real research and have been finding audibility between 200PS and 20PS using our Alessis deck. One cannot just look at Dolby's numbers on audibility either - many find fault with these. Remember these are the same people who said we could not hear above 20khz which was dead wrong.

As for Stereophile, read John Atkinson's lab reports in every issue where they review CD players. He feels different than that dated 1999 reference to the same paper you discussed before. He finds that differences in hundreds of picosecond can be a big deal. I am not the only one out there who hears this difference. I can point to Chesky engineers, mastering engineers, etc. It may seem unlikely at first glance that the ear can hear such admittedly split second difference in timing but that is why the neural network in our brain is so amazing.

I did notice that you did not refute anything in Dunn's paper...

 

[Chu Gai]

yeah yeah far as I know Floyd doesn't have any 120 dB licks @20K and besides I saw them several times, far too few I'm afraid, and saw them do what was to become Dark Side of the Moon about a year before the album came out. Good stuff

You're not understanding or misinterpreting the essence of what Dunn had said. Reread it carefully and understand the particularly special and exaggerated conditions he chose in order to ascertain the 20 ns level. Headphones, which lower the ambient noise floor, and test tones are invariably used when doing research.
Dunn's paper quite naturally led to the other which now sought to put a face on jitter from the practical aspect of music.
On Sept. 15, 1999 J. Dunn wrote in thread titled "Why aren't bits bits?" in Usenet the following.

It is likely that much of the misunderstanding about this comes from
the multitude of different ways that people have to measure jitter.

For example the cycle-to-cycle jitter measurement is very
insensitive to audio-band jitter. As a result audio frequency jitter at tens of ns amplitude (when measured with respect to ideal transition timings) could measure as cycle to cycle jitter at an amplitude of tens of pico-seconds on a 22MHz clock.

That level of jitter could be audible (as shown by Eric Benjamin and Ben Gannon's paper at last years San Francisco AES) but the conclusion that then jitter audibility in the ps region is possible is misleading by a factor of (possibly) a thousand.

Another misunderstanding comes from the assumption that any
imperfections in digital audio are a result of jitter. This can lead to an unhealthy interest in clocks at the expense of everything else.

Ahhhh...Mr. Atkinson has much vested in flawed methodology and hence has great appeal to people who subscribe to equally flawed methodology and that includes Chesky, Shmesky, and Pesky

There was nothing to refute in Dunn's paper Lee. You simply didn't understand it's relevance however it would seem that Dunn understood the relevance of Gannon's work

 

[Chu Gai]

Oh, and Dunn also wrote that cables have an imperceptible effect upon jitter.
Ever see Floyd do 'The Wall'?

 

[RobWil]

and that includes Chesky, Shmesky, and Pesky

The PEP boys??? I thought they sold auto arts!?!?

 

[Michael R Price]

Haha, that's wonderful.

But seriously though, I think these things are a bit more complicated than measuring a jitter number and saying it's below the threshold of audibility. I mean, come on, why does one opamp sound better than the other (and discrete components better than any opamp) when they all have multi-megahertz frequency response and distortion of -100db? What about the jitter and distortion caused by vibrations hitting the transport and slightly affecting the optical pickup and accompanying electronics? Or the noise placed on common power supply lines by the chips in there? Bet they don't measure those.