[audioengr]

Quote:

Originally Posted by AudioCats /img/forum/go_quote.gif Interesting. So jitter reduction is not just a lower-ppm clock module, it has a lot to do with everything else (power quality, wiring, signal reflection in the cable, optical sending/receiving modules, etc)?

You got it. Read this white-paper:
jitter

Quote:

which means the digital signal from different transports' digital-out are not exactly the same, even if the data stored on the CD is identical and the disk read-out is 100% correct?

Correct. The timing and jitter is different, even if the data is identical.

Quote:

If that is the case, then a flash/harddrive system will still have the same problem, different digital-out will sound different.

Correct also. Depends on many factors in the implementation and the design (these are different things) including the jitter of the master clock (PPM accuracy has little to do with jitter).

This is why customers buy more expensive USB converters and DACs. They sound better than inexpensive ones. Almost everything in the design matters.

Steve N.
Empirical Audio

 

[Prog Rock Man]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Me too, but there really isn't any scientific evidence proving Jitter is the issue. It has kind of become accepted by ruling anything else out, whats left is the ghost in the machine.

I say that if you have various possible causes to something and you rule all out bar one, however unlikely, that one becomes your cause.

 

[chesebert]

to be honest if we can make processor run at 8+ ghz we can make perfect transmission at 44.1Kzh - 192Khz. Most of the jitter/clock skew problems have already been solved by the VLSI community (you can find most solutions in IEEE papers). The only thing left to do is adopt some of those techniques to audio design. It has taken the longest time for manufacturers to do the buffer + reclocking; but that's just a very basic mean of taming jitter/clock skew. I think most manufacturers are just taking their sweet time to squeeze more money out of us regular folks.

 

[audioengr]

Quote:

Originally Posted by chesebert /img/forum/go_quote.gif to be honest if we can make processor run at 8+ ghz we can make perfect transmission at 44.1Kzh - 192Khz. Most of the jitter/clock skew problems have already been solved by the VLSI community (you can find most solutions in IEEE papers). The only thing left to do is adopt some of those techniques to audio design. It has taken the longest time for manufacturers to do the buffer + reclocking; but that's just a very basic mean of taming jitter/clock skew. I think most manufacturers are just taking their sweet time to squeeze more money out of us regular folks.

Strange thinking. Are you an engineer?

It turns-out that highly integrated devices like FPGA and gate-arrays actually aggravate the jitter problem. Because there is so much switching and power supply noise on the typical highly integrated die, the jitter is actually worse than the same circuit designed with simpler discrete parts. There are usually only a few power pins on such parts being shared by maybe a hundred signal pins, when there needs to be dozens. You can only apply a few power decoupling caps to these pins. Output drivers on large parts are usually crippled as well.

It's synonymous with the op-amp versus discrete transistor amplifer performance gap. Same deficiencies.

This is precisely why my Pace-Car USB design is such a complicated thing with so many discrete parts. I could have done the whole thing in a programmable gate-array, but the jitter performance would suck and the isolation from input to output would also suck. This is precisely why the jitter game is such an artform.

I was a design team lead on the Pentium 2, so I understand these tradeoffs. I have done many VLSI semi and full-custom chip designs in my 30 years as a digital designer and engineering manager.

Steve N.
Empirical Audio

 

[RickEC]

.

 

[IPodPJ]

Quote:

Originally Posted by audioengr /img/forum/go_quote.gif This is precisely why my Pace-Car USB design is such a complicated thing with so many discrete parts. I could have done the whole thing in a programmable gate-array, but the jitter performance would suck and the isolation from input to output would also suck. This is precisely why the jitter game is such an artform.

So what are the jitter ratings of your device? You have been asked repeatedly to state the measurements of your gear and have refused to answer. How can anyone take seriously an engineer selling equipment that supposedly reduces jitter when he says he can't be bothered with taking jitter measurements? Yet you seem to have no problem posting the 2ps jitter measurements of the Superclock you offer with other devices.

Much of what you say makes sense and based on my experience with audio equipment it seems to be logical, but without any proof to back up what you say it is more or less useless.

 

[chesebert]

Quote:

Originally Posted by audioengr /img/forum/go_quote.gif Strange thinking. Are you an engineer? It turns-out that highly integrated devices like FPGA and gate-arrays actually aggravate the jitter problem. Because there is so much switching and power supply noise on the typical highly integrated die, the jitter is actually worse than the same circuit designed with simpler discrete parts. There are usually only a few power pins on such parts being shared by maybe a hundred signal pins, when there needs to be dozens. You can only apply a few power decoupling caps to these pins. Output drivers on large parts are usually crippled as well. It's synonymous with the op-amp versus discrete transistor amplifer performance gap. Same deficiencies. This is precisely why my Pace-Car USB design is such a complicated thing with so many discrete parts. I could have done the whole thing in a programmable gate-array, but the jitter performance would suck and the isolation from input to output would also suck. This is precisely why the jitter game is such an artform. I was a design team lead on the Pentium 2, so I understand these tradeoffs. I have done many VLSI semi and full-custom chip designs in my 30 years as a digital designer and engineering manager. Steve N. Empirical Audio

was an engineer; was swinging VLSI with the PowerPC (Power 4) team. Now I am just another paper pusher.

however your answer still does not explain why manufacturers are not using knwon techniques to reduce jitter. I do agree clocking in FPGA is a nightmare; ASIC is the way to go for clock sensitive logics.

 

[audioengr]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif So what are the jitter ratings of your device? You have been asked repeatedly to state the measurements of your gear and have refused to answer. How can anyone take seriously an engineer selling equipment that supposedly reduces jitter when he says he can't be bothered with taking jitter measurements? Yet you seem to have no problem posting the 2ps jitter measurements of the Superclock you offer with other devices.

It's not that I refuse to answer. I just dont have the gear here to make these measurements, and I cannot afford it. It's over $2K just to rent it for one month, and then you need to learn it and develop measurement suites. Much better to have a measurement professional do the work, and cheaper. Like most folks in the high-end audio business, I'm not getting rich, in fact I'm still in the red after many years. I only do this because its my passion, not because I expect to get rich.

My plan is to send a Pace-Car USB off to my professional measurement guy that has the latest audio precision gear as soon as I have funds for this. This will be coincident with some high-profile reviews.

Just found out this morning that I have been scammed by a cybercriminal in Indonesia for over $5K, so this has to go on the back burner.

Steve N.
Empirical Audio

 

[audioengr]

Quote:

Originally Posted by chesebert /img/forum/go_quote.gif was an engineer; was swinging VLSI with the PowerPC (Power 4) team. Now I am just another paper pusher. however your answer still does not explain why manufacturers are not using knwon techniques to reduce jitter. I do agree clocking in FPGA is a nightmare; ASIC is the way to go for clock sensitive logics.

As for myself, ASIC is a big investment in money and time, and it delays time to market, which is probably even more important in this fast moving market. ASIC still does not solve the issues with highly integrated chips that I stated. Discrete solutions always perform better. I do use Xilinx for control functions.

As for others: based on the large number of devices that I have modded, I have to conclude that the depth of digital design experience in high-end audio is very limited. Designers with EE degrees and lots of years designing synchronous and self-timed logic are scarce as hens teeth.

And I am using lots of known techniques and some new ones to reduce jitter. Have you even read what my Pace-Car is?

Steve N.
Empirical Audio

 

[AudioCats]

ok, maybe we have reached a sensitive topic here:

if jitter has everything to with everything inside the digital source, including the power supply quality, does that mean the "computer as source" route is kind of a dead-end (SQ-wise)? since the computer's internal power supply quality is gonna be noisy by default, and it is not likely somebody will design a seperate more "pure" supply to put into the PC (not to mention most of us are thinking about using laptop/notebooks).

 

[IPodPJ]

If you guys really want to get serious about low jitter and the best clock performance you can get ---

Antelope Audio

 

[doctorcilantro]

Excluding all other causes or sources of jitter, Toslink connections are important, that's where the transmitters are converting from voltage impulse to light and vice-versa. Coax "should" be better theoretically but Toslink is immune to RFI & EMI. With BNC or AES (RCA coax) the signal doesn't go through this conversion process.

The difference is probably how the differing terminations are interfacing with the transmitters.

I have personally found Toslink unreliable, you often bump a cable and cause sync loss. With well built Toslink cables and devices this is less so I guess. I know someone who had to go through 3 Toslink cables to get their Airport to work to feed their Peachtree Nova.

I steer clear of Toslink these days due to personal bad experiences. It will work but I'm leery of it for the aforementioned reasons, and assuming the initial posters observations are correct, or reflect some actually, measurable, varyig amounts of jitter exacerbated by cable construction and quality of Tos. transmitters.

DC



Quote:

Originally Posted by AudioCats /img/forum/go_quote.gif So I modded the D2C today (power-supply cap replacement; shorted DC blocking caps; took out the balance output caps; got rid of the headphone amp; and finally, added a 8.5V EI core transformer for the digital power). Then I compared a few more transport/cables. gear: -Sony X555ES CDP -Sony C601ES changer -Sony CA70ES changer -Sony blueRay player (the cheap $130 stuff) -Thick and flexy optical cable (no marking, was supposedly made by BJC) -Thin generica optical cable (Sharp brand, originally costed me $2.5) -AudioQuest cinemaquest video/digital coax cable (6'). The Dac is, of course, the modded Zhaolu; Headphone amp is a Stax SRM1 with output cap upgrade, driving a pair of HeAudio Jade's. Some results: *with the thick optical cable: X555ES has the softest sound, the C601ES has the sharpest. CA70ES is somewhere in between. The X555ES is way soft, can't be just placebo. *Difference between thick and thin optical cable (using C601ES as transport): the thick cable is mellower; the thin Sharp cable has more sparkles and a little more "metalic". Difference is not huge but noticiable with a quick A/B. *Somehow "C601ES -->thick optical" combo sounds the same as "Blueray -->digital Coax". So, the new question: why would optical cables sound different ?! They are just light pipes.....

 

[doctorcilantro]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif If you guys really want to get serious about low jitter and the best clock performance you can get --- Antelope Audio

Black Lion makes clocks as good, in shootouts go head to head with them; boxes just don't look as pretty.

 

[doctorcilantro]

Quote:

Originally Posted by AudioCats /img/forum/go_quote.gif ok, maybe we have reached a sensitive topic here: if jitter has everything to with everything inside the digital source, including the power supply quality, does that mean the "computer as source" route is kind of a dead-end (SQ-wise)? since the computer's internal power supply quality is gonna be noisy by default, and it is not likely somebody will design a seperate more "pure" supply to put into the PC (not to mention most of us are thinking about using laptop/notebooks).

I don't think so because A. you can use a low power cpu and PSU like PicoPSU and even buy a linear to power it (if you felt the need - I don't), and B. Lynx cards, for example, are in use in very high-end systems and they have designed the card to be in the PC environment. Other cards may suffer from poor design or implementation.

 

[doctorcilantro]

Quote:

Originally Posted by chesebert /img/forum/go_quote.gif to be honest if we can make processor run at 8+ ghz we can make perfect transmission at 44.1Kzh - 192Khz. Most of the jitter/clock skew problems have already been solved by the VLSI community (you can find most solutions in IEEE papers). The only thing left to do is adopt some of those techniques to audio design. It has taken the longest time for manufacturers to do the buffer + reclocking; but that's just a very basic mean of taming jitter/clock skew. I think most manufacturers are just taking their sweet time to squeeze more money out of us regular folks.

fascinating, thanks!