Optimal lengths for USB and Ethernet cabling for digital audio

No one knows?  I must say I am pretty surprised at that!

With usb I found 1 1/2 M better than shorter.  With ethernet I did not find there was a difference.

I’ll just say I was gonna go with a shorter Ethernet cable to save $$$ but heard it’s better to keep streamer further away from DAC so I went with 1.5m.  So, not so much the cable but the potential noise from the streamer being 5he problem.  Seems like USB less susceptible to noise but look for a cable that physically separates the data and power conductors to avoid transfer of noise.  That’s all I got. 

I have directly compared in my system two pairs of USB cables of two different brands, each pair consisting of a 1m and a 2m cable. In both cases, the shorter cable had richer, more seductive midrange while the longer cable had stronger and more impactful bas, (very) slightly more emphasized highs (or, rather, the cymbals seemed to have better contour and thus were more distinct in the mix) and in general sounded tighter, more organized. I preferred the longer cable in both cases, in one of them even by a long margin. As a side note, this seems to be a typical impact of the cable length, at least in some cases (USB and power cables? shielded cables? I don’t know) because I have discovered more or less the same effect when comparing a couple of otherwise identical power cords of 1m and 2m respectively.

USB 1.5m with seperate output tails, most DACs don't need USB power so tape over pin 1 in the output power tail, or pin 1 if it's an ordinary single cable.

USB cables offer two way communication so length is an issue.

My system uses Ethernet cables between 0.4m and 1.2m, shorter seems better but I haven't done any critical experiments.

I don't recall reading anything about reflection issues with ethernet or usb cables. Makes sense ethernet cables should be shortest possible to minimize rfi, although can still get in through connectors. For usb, seems 1.5m is pretty much standard length used, I've never deviated from this length.

As far as ethernet vs cable provider coax, I've found moving cable modem close to system via longer coax/shorter ethernet far preferable to shorter coax/longer ethernet. Another advantage is one can afford much better quality ethernet cable, 1 or 1.5m AQ Vodka vs aprox. 18m price, forget about it.

it seems curious to me that it is commonly accepted that spdif rca digital cabling should not be 1 m in length or shorter, 1.25-1.5 m better to reduce chance of reflections...

but the voltages expressing 1's and 0's are still traveling down the wire pairs in usb or ethernet cabling... why are they immune from such reflections, if they exist?  

is there something about the 75 ohm nature of the spdif cabling that makes this matter?

The one digital cable where long runs are clearly bad is HDMI.  Otherwise Ethernet is the preferred cable for longer runs, and I don’t know that it matters much for usb if the receiving device does a decent job of reclocking 

steve nugent of empirical audio, who used to post sometimes here, but no longer, is one of the experts in this area (digital audio, jitter, cables etc...) -- it was his writings in Enjoy The Music some time ago which started quite a bit of useful discussion on managing jitter and read errors in digital systems data transfer, cabling etc etc...

https://www.enjoythemusic.com/magazine/manufacture/0509/

https://www.empiricalaudio.com/computer-audio/technical-papers/

https://positive-feedback.com/Issue14/spdif.htm

https://positive-feedback.com/Issue43/jitter.htm

i shot him an email this morning, asking this question - his answer is that he feels usb cables carrying digital music data are prone to the same potential reflection issues as spdif rca cables, and thus, he uses 1.5m USB cables just to be safe...

Thanks jjss49, as always your posts are very useful, I learned quite  a bit from reading those.