USB cable MATTERS!

Quote:



You probably had a broken cable.
But at least you're happy now, which counts the most.

well, i most certainly agree with a cable not affecting quality in any way, as long as the 1's and 0's do actually get transferred from one end to the other...i had this problem of contact with the stock usb cable that came with my emu0202 - for about a month i couldn't understand what was going on, but at times the audio would just cut out, and then i'd have to switch the device off and on again to get it back...eventually i figured out that even a slight bump to the laptop/dac would cause it to momentarily loose contact and then it would reconnect at usb 1 speed. So i just replaced it with one of the hard disk (or was it a printer) cable, and noe more problems. Seems like one of the cable's connectors just fit quite loose into the socket of the emu0202, but it's fine when i use it on my hard disk...anyway - so i do think the cable matters...

Some USB cables don't work properly with full USB 2.0 spec, so some devices don't get recognized as USB 2.0.

I was probably too firm in my first post to get the point accross clearly. Fair comment about oxidation but I think it matters more when comparing with copper, most cheaper cables use silver or aluminium connectors AFAIK.

Generally I also buy the nicer cables because 1) they look good 2) you know your equipment is definately not being restricted in any way possible

Sounds like the original poster had a very poor cable to begin with, i've had similar problems before where even factory made cables have really weak joints from the wire to the connectors.

-J

Since I am an industry expert on the subject of USB audio and electrical engineer with 30 years of digital interface design experience, I though I would set the record straight.

All of you that believe that no errors take place on a USB cable that is proper length (< 5m), are absolutely correct.

On the other hand, all of you that believe that USB cables cannot color the sound of streaming playback are dead wrong, in fact they do. The effect that the cable has is called dielectric absorption and dispersion. This adds jitter to the signal as it is detected at the receiver. Cables with ferrites will actually add more jitter due to limiting bandwidth of the cable, so this is not advised. Longer cables and cables with larger than normal conductor gauge will also add jitter because of skin-effect and losses. The very best sounding USB cables will be short, use Teflon/air for dielectric and use very thin conductors for the signals, properly annealed Silver being the best.

If you dont understand jitter, then you know very little about quality digital audio. Do some searches on the forums and you will learn a lot. Getting quality results from digital audio is primarily about reducing and maintaining low jitter. Second order effects are D/A quality and preamp/analog stage noise and distortion.

Those that claim jitter is a non-issue have never heard a decent digital source IMO. We are all at different levels in our audio quest, but it is good to understand the fundamentals. This way, you know what kinds of things will actually make significant improvements.

Steve N.
Empirical Audio

Steve,
Could you explain to me further the effects of dielectric absorption and dispersion, I couldn't find good articles on them on the web =(.

Also would the cable still be an issue if the DAC had a separate clock and worked in asynchronous mode?

Thanks,
Dave

Quote:



Dielectric absorption is when the dielectric (insulation) in the cable charges due to the changing signal, but it resists discharging completely. It "holds onto" some residual charge. This can easily be measured with the right instrumentation. This absorption of charge is a kind of electrical "sponginess" of the insulator, like a sponge soaking up water, but does not release all of it when it is squeezed. The effect is that the transient signals encounter a sort of "inertial resistance" and tend to "spread" in time as a result, adding signal distortion and ergo jitter.

This also happens in capacitors. This is why a ceramic capacitor in the analog signal path sounds much worse than a polystyrene cap in the signal path.

As for the USB interface and it's rejection of incoming jitter, this depends entirely on the implementation. Even Asynch interfaces can be affected by incoming jitter. Even reclockers can be affected to some extent, although less so than typical interfaces.

Most of the digital designers, including me, that are convinced that they have finally tackled the jitter imuunity problem end up eating our hats because the problem is still there to some extent. It is one of the hardest nuts to crack IMO.

Steve N.
Empirical Audio

Steve,
Thanks for the info, that makes a lot of sense.

Feel like walking me through jitter causers or your input etc. on optical cables?

Dave

Just as Steve says.

For digital audio, don't buy a USB cable with ferrite beads. I know major cable manufacturers use ferrites -- nonetheless usb.org specifically says not to (I have posted the link about 4 times before in other threads).

I spent today with the national sales manager for a leading maker of ferrite beads and similar emission suppression products. He has an economic interest in everyone using ferrites, but he nonetheless says "not for use with digital audio, they will retard the signal".

Don't you get tired of posters who keep saying "it's just zeros and ones". No it's not, it's zeros and ones delivered in time under hard real-time constraints ... "streaming", as Steve says.

Sure the cable won't flip a bit, but the receiver will re-construct a bit stream that is NOT a perfect copy of the original if there are any significant transmission delays. Buffering and re-clocking on the receiving side help, but might not fully eliminate the problems introduced by a bad (out of spec) USB cable.

Quote:



Here are some white papers that I wrote for PFonline:

http://www.positive-feedback.com/Issue14/spdif.htm

http://www.positive-feedback.com/Issue22/nugent.htm

Steve N.

Quote:



Actually, the receiver will always reconstruct a perfect data bit-stream. It may not however create a perfectly timed bit-stream. This is what affects the D/A conversion and creates audible distortion, usually signal-correlated.

It is kind of like a timepiece that is accurately ticking-off the minutes, but each second varies as much as 20%. Some seconds may be 10% slower and some 10% faster. This is jitter. It always shows the correct time at the macro level, but at the micro-level the timing is a moving target, constantly shifting.

Steve N.

Quote:



That's a new one to me. How can a bulk mode transfer solution with reclocking be affected?

if the cable is providing power, then I would agree that it could affect sound quality.

you need inductance on that cable (the big cylindrical things often found) to prevent bit errors due to power loading, switching, ground noise, etc

Quote:



My understanding (although I have only briefly read jitter stuff) of jitter is a disagreement on when to read bits on the cable. A bit of a timing error. It is my instinct to then characterize that as infact a bit "error" and should be classified as such as opposed to coloring the audio.

In addition, I am confused as to why this is even a problem. Transfer rates of USB 2.0 is well above the 176 KB/s that is required for stereo 44.1KHz recordings. Why are the transfer clocks not slowed down to make the timing have a pretty big area to time in? or using some kind of encoding that hides the clock in it.. manchester encoding?

edit: after reading more about jitter (actually just any kind of signal integrity problems?) and usb it seems like the problems are high frequency transmission line problems which I recently took a class in! Given that USB2.0 is ridiculously fast for audio, why don't they just use a slower clocking version of USB since it is a one way transmission?

There are tons of problems with high frequencies over any kind of transmission line, as audioengr mentioned, like skin effect, proximity effect, surface roughness, impedance matching, reflections, etc. although I have not heard of dielectric absorption, but from what he says it sounds like what I've learned is crosstalk and mutual capacitance.

In the end, it seems like a new protocol for usb audio needs to be written. Something as simple as transmitting the same bit 5 times and taking the majority would seem mighty effective, but USB is not written for audio.

Steve N. Thanks a lot for the info/clarification.
As a cable believer its nice to see someone knowledgeable explain that an USB cable do indeed matter.

Quote:



LOL