|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
70.168.149.115
In Reply to: RE: Humility and Prayer posted by John Marks on April 20, 2016 at 20:57:26
nt
Follow Ups:
Seriously.
I was speaking with a major big-shot household name digital guy, and I--honestly, not being provocative or setting a trap--I asked, does anyone know offhand, converting time to distance, how LONG a SINGLE positive PCM pulse would be, if it were frozen in time?
In other words, given the interchangeability of frequency and time, and time and distance, if one froze a single positive PCM pulse in time, how long would a S/PDIF cable have to be, in order to hold a single frozen positive PCM pulse?
He laughed genially and said that he had never ever heard anyone even think about that, it could not possibly matter.
I replied, I am hearing differences at 44.1 kHz with a 3.5 inch difference in cable length...
So I sat down and did the math, and did the math, and did the math, and called him again and said I must be having a stroke, this can't be right.
That was almost 24 hours.
I walked him through my calculations, and he laughed and said, well, you have to be right!
ANYONE, please re-do it!
BTW, at that transfer function speed, the difference between 44.1 and 48 is no big deal.
OK, I calculate a singe positive PCM pulse (Zero to .25V and back to Zero), frozen in time, measured in length, as: THREE MILES.
So, what I heard in prototypes that were 3.5 inches different, can't possibly make a difference. But... people hear it.
I was trained as a translator of French poetry and I got a law degree and was co-author of peer-reviewed papers read at Harvard Law School, and, honestly:
I can't say whether the cables I sell sound good because of my work or DESPITE my work.
They are what they are. If they make people happy, I am happy.
John
What kind of a pulse are you talking about? What you see on the SPDIF cable? That would be at the SPDIF clock rate. Or are you talking about the analog pulses you would see at the output of an unfiltered NOS DAC. These would be at the word clock rate, e.g. 44100 Hz. The SPDIF pulses have to be at a much higher rate to allow for 20 or 24 bits of data per sample, 2 or 4 channels, and sending both the data and the clock.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Speed of light is represented by "c."
c = 1 foot per nanosecond
In one second, light travels 1 Billion feet (186,000 miles)
Given VPROP of .75c, in my cable, the electromagnetic flux travels 750,000,000 feet in one second.
To divide by fs of 44,100: 750000000/44,100
7,500,000/441
Solve to 17,006.8027
3.22 miles
A 3.5-inch difference is on the order of, 10 to -5 power.
???
JM
So what??? This is no basis for a proper cable design.
It's not the basis of the design--it was an INQUIRY (perhaps the wrong tree was barked up, so what) into why two cables otherwise identical except for length sounded differently.
NB, I discussed such issues with JA only after the cable was in distribution and use and the design frozen. These were purely theoretical and academic discussions. JA arrived at pretty much the same circa three mile figure I did, but of course he said that it was the "length" (as a function of time and drag) of the transition zone that was critical.
But Charley Hansen, for reasons of his own, said that a S/PDIF cable needs to be at least 20 feet.
I tried a few things over the course of a few years, and ended up with something I was happy with. I was ready to just make a few for family and friends when a couple of industry people, unpon hearing one used with familiar equipment, urged me to get serious with it. I made a final tweak in the design, the production units sound better than the prototype, and I have yet to find an explanation for that that to me is both understandable and plausible.
Just a matter of intellectual curiosity.
There are also aspects of the design I do not disclose or discuss publicly.
JM
S/PDIF transmits data serially in 32-bit subframes, each containing one audio sample. So for 24/96 PCM, the bit rate is 2 channels * 32 bits/sample * 96000 samples per second = 6.144 MHz (that was the spec for the original standard). S/PDIF uses bi-phase encoding, so the clock frequency is twice the bit rate = 12.288 MHz. And the width of one pulse on the wire is one half of a clock period = 4.069e-8s. If you assume your propagation velocity is 0.75c, which is relatively high (PTFE is around 0.7), then the length of a pulse on the cable is 9.15 meters. For 44.1 PCM, it would be 19.9 meters. Not 3 miles, but still a fair bit bigger than 3.5 inches.
Anyway, it's the transitions that matter, not the length of time between transitions.
BTW, in a private email, John Atkinson quoted different numbers, and again in a private email, Charley Hansen quoted different numbers.
I have always said that my cable design is what it is, it sounds as it sounds, and whether it sounds good is despite my efforts rather than because of them doesn't matter if the customer is happy.
(I know competent engineers who claim that any properly working DAC can cope with minute timing errors and therefore cables cannot matter... .)
I have never claimed that my cable is "best." I think digital cables are rather system-specific. I think that my cable is a valid option and I am happy with its sound, at the price. I could (in theory) make a hugely expensive cable but at this point I have no idea whether it would sound better.
Anyway, it's "Countdown-to-Campari Time" here at Esperanto Audio's worldwide headquarters' executive suites.
Tee hee.
John
Pretty sure I was off by a factor of two because I didn't need to double the clock rate.
Anyway, if we knew the bandwidth of a typical transceiver driving the cable, we could estimate the rise & fall times on the transitions. Or you could just measure them with a scope. It wouldn't surprise me if the length of a transition is on the order of the length of the cable, meaning the cable isn't easily characterized by either transmission line or lumped impedance models.
However, the pulses actually going down that cable are at a much faster rate, going at least 40 times faster. (stereo and 20 bits). So the individual pulses are measured in a few hundred meters. If you are worrying about the SPDIF cable, why are you worrying about word clock pulses, pulses that are generated at the opposite end of the DAC from where the cable enters?
If you are looking for an explanation as to why changing the cable length a few inches matters, well that's because the DAC gets a different signal, albeit one with the same bits. Then, given slightly different input it produces slightly different output. (This isn't logically or mathematically necessary, rather it happens due to the way the DAC is built.)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Thanks, but could you please elaborate on...
"If you are looking for an explanation as to why changing the cable length a few inches matters, well that's because the DAC gets a different signal, albeit one with the same bits."
If the bits are the same, then how is the signal different?
I am not being argumentative or provocative, I am genuinely curious. My design process was closer to a blindfolded kid with a stick's trying to thwack a pinata than anything else. But people with no need to pat me on the head have bought cables and recommended them.
ATB,
John
The same bits can be conveyed by slightly different waveforms due to noise and timing issues associated with the transmitter at the transport, the digital cable, or the digital receiver in the DAC. In an ideal world, the digital receiver in the DAC would correctly interpret the bits and would eliminate all of the noise and timing issues. In the actual world, this does not happen and some of the variations make it to the analog output of the DAC.
Here are two strings of bits that are the same, but are physically different on the screen:
0 1 0 1 0 01 0 1 1 1 01
0 101 0 0 1 0 1 1 1 0 1
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Dear Tony,
Thanks.
I think then we are using different terms for the same thing or the same constellation of phenomena... .
I believe, as apparently you may also, that suboptimal terminations (including the wiring behind the panel jacks, either RCA or BNC) can result in reflections such that when the reflection makes a round trip and comes back to the DAC, the reflection can degrade the timing accuracy of a later-dispatched signal.
(BTW, if this is the case, then should we not want higher-loss cables rather than lower-loss cables? If the cable were not capable of carrying those micro-level reflections back and forth, would not the Digital Eschaton thereby be Immanentized? Only half joking.)
I believe that cable length is an important factor in determining when the reflection arrives with respect to the zero crossings of the later-dispatched signals, which is why I do not offer cables in lengths such as half a meter and one meter.
I do have to tell you though, and I am sure that this will be no news to you, that more than one engineer has told me that if a DAC cannot cope with such small levels of induced jitter, it was either incompetently designed or is broken, and therefore all properly-fabricated, non-broken digital cables sound the same.
ATB,
John
"I do have to tell you though, and I am sure that this will be no news to you, that more than one engineer has told me that if a DAC cannot cope with such small levels of induced jitter, it was either incompetently designed or is broken, and therefore all properly-fabricated, non-broken digital cables sound the same."
This pretty much covers every DAC that has ever been made, although few designers seem to be honest enough to admit this applies to their products. There is another side to this. Even if such a DAC were to be made, there would be few subjective audiophiles honest enough to admit that they could not hear any differences caused by digital cables.
Removing noise from digital signals is well understood in the field of military electronics, where it is needed in encryption devices that must keep separate "red" signals corresponding to classified information from leaking into "black" signals being sent over cables or airwaves. If as much money flowed into high end audio as flowed into military electronics our problems with digital cables would have long been solved.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
If the Pentagon were given this problem, we'd spend a decade and many millions funding contracted R&D projects to evaluate alternatives, a couple more years and millions arguing and negotiating over requirements, a bunch of competitive prototypes demonstrated, a source selection, a protest of the result, a development contract eventually awarded to productize the winning interface, the result would be unaffordable for its original purpose, other companies wouldn't adopt it because they hadn't invented it, something would get produced if for no other reason to show some return on the investment, but by the time that happened it would already be obsolete. Don't ask me how I know this :)
nt
I don't believe any of the "variations" can me be measured in any way at the analog output of the DAC.
And if what you are saying not simply wonky theory, then the market it littered with broken digital cables, transports, and DACs.
I don't believe any of the "variations" can me be measured in any way at the analog output of the DAC.
They are routinely measured. Ever seen a jitter test?
These variations can be, and have been, measured on the analog output of DACs. Usually when this has been successfully accomplished the measurements or experimenter have been discredited or the equipment under test decreed to be defective.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"..when this has been successfully accomplished the measurements or experimenter have been discredited or the equipment under test decreed to be defective."
Can you cite specific examples of this please.
"...well that's because the DAC gets a different signal, albeit one with the same bits."
File this under "pseudo technical gibberish"....:)
I appreciate your support, but, I take Tony seriously! I am always willing to learn.
However, the other end of the equation is the scene from "Brideshead Revisited" (the novel or the ITV teledrama) where the priest asks Rex, what if the Pope had told him it was raining, but in the window behind the Pope, the day was clear?
I had an irreligious girlfriend who would razz me about that. Like clockwork.
JM
I like Tony too.
In the dictionary of internet forum personality types, he would be known as "The Professor", they are compelled to educate with every post, even when the "lesson" is unsolicited.
But I think in this case, it was a bit of a "if you can't impress them, just dazzle them with BS" type post. LOL.
All good.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: