USB cable measurements

Would be cool if these cables also had jitter measurements using identical PC and USB DAC and see if any trend comes out of it when compared to measurements you can do.
The most in-depth jitter looks to be that of Paul Miller, who also had added a new test sequence relating to a trait that is happening between DC-4hz (relates more to products-chips rather than cables but worth complete test to see if any interractions on this and also transmission noise-external interference).
Cheers
Orb

Hi

Would have liked to know more about the methodology. What were the DACs and PCs involved? Was it an asynchronous or syncchronous USB DAC? Was the test conducted in a way that would limit orremove biases? etc... Also I am confused by the units used for impedance? At what frequency were it measured?

As usual, HiFi News articles are mostly in print and not online. A quick google search shows the following post:

'Paul Miller of Hi-Fi News agrees with you.
Check his article on USB dacs ("USB if you want to") on this month's mag (December 2010).
It will continue next month on USB cables, which "can also make a difference".

And btw, about his Sony Vaio laptop, with the PSU connected (compared to battery power), "low-rate, noise-like jitter increased by a factor of 10x (~3000psec) during D/A conversion in the attached Pro-Ject box". '

I will check the dates as Amir is right on, there is also a couple of others that I will find, so bear with me while I find them.
One follows on from a contributor at Hifinews and his article that can be found online:
Paul's follow-up article in one of the print publications also explains and shows a difference in comparing results for this isochronous DAC (which Jim's DACMagic is) to an asynch one.
http://www.audiomisc.co.uk/Linux/Sound3/TimeForChange.html
The frequency is explained in the Time for change link and is related to jitter - it is worth noting that technically it is not wow-flutter but used as a description (again PM expands on this).

The cheapo standard 5m USB cable that performed better than an expensive audio marketed one was Belkin if I recall correctly.

Cheers
Orb

That was a very useful link Orb. For those who didn't chase it or read it, the measurements of USB is done over many seconds, to see what types of gross speed variations there are -- as we used to measure with analog gear. They find the standard USB to have spikes once every few seconds.



Think through why this occurs, the reasoning seems fairly straightforward. The PC is sending data out on the USB bus in a manner that approximates the clock frequency of the original sampling rate. There is a transformation of X number of samples per USB frame of data. The DAC on the other hand, has a precise audio clock that drives the DAC. A buffer exists between the USB and DAC.

What then happens is that the clock speed of the DAC is either too fast or too slow. The DAC firmware lets it run at its normal clock for a while and then adjusts it at regular intervals to compensate between the two differing clock rates. For example, if the USB clock translates to 44,098 but DAC is running at 44,100 your buffer will eventually run out of data to feed it.

In this implementation, the DAC lets the local DAC clock go for a while and then heavily adjusts it to get back on track and then lets the clock go again.

The Asynch implementation has no such tie-in. The DAC clock is the master and as the buffer gets emptier, a message is sent to the PC to get the next batch of data. Its lock is never adjusted in that manner to speed or slow it down to incoming rate (i.e. slave operation).

I suspect this type of artifact is heavily DAC dependent. Some may adjust more rapidly in smaller increments or wait longer and make larger variations.

I am interested. Keep the reports coming .

I suspect the biggest difference in how a USB cable "sounds" is more related to interaction among source, cable, and load than the cable alone, except for a very bad cable, and acknowledging that reflections are a significant part of that interaction. I commend ar-t for looking into just how far off the mark some of the cables out there are, irrespective of price... - Don

Aside: I have measured rather wide variances in the impedance of twisted pairs, more so than coax, most often due to the number of twists per inch (m, whatever) varying. Fewer twists means lower distributed C, less wire used for a given distance, and more profit; not sure that really plays into it, or it is just bad manufacturing QA.