PS Audio Lanrover USB transport discussion

[Superdad]

7 hours ago, Speed Racer said:

How the heck can the ethernet cable used make a difference in the sound quality since, no matter what, the data is the same when it gets to the receiver Lanrover. This is a case where it truly is just digital data.

 

Or am I missing something?

 

Actually, you were warm.  The receiver in the LANrover (or any Ethernet or USB device) is a PHY (physical interface), and the packets of data coming in are not fast marching little ones and zeros.  It is a high frequency analog voltage, and the job of the PHY chip is to decode that into a binary data state.  And the PHY chips are nasty, complex little buggers containing multiple PLLs, and clocking systems--and the harder they have to work decoding the incoming signal, the more perturbations they generate: bursty power draw spikes, ground-plane noise, "packet-data-noise," etc.

Ethernet cables and USB cables (and of course USB hub chips and other interfaces such as our REGEN, etc.) exhibit variations in signal integrity (as seen on an eye-pattern) which is a mix of amplitude, slew, noise, jitter, etc.

Improved SI (as well as improved impedance match in the case of USB) makes the job of the PHY easier, resulting in less device-internal generation of the issues listed above.

 

In the case of a USB (or Ethernet) interface located inside a DAC (or USB>S/PDIF converter), better signal integrity readily translates into better sound as the deleterious stuff generated by the PHY sails though most any isolation attempts, ultimately making its way to the master clock.

 

So while the Ethernet cable between the two USB-based LANrover devices is a bit removed from direct DAC/converter PHY interaction, its own PHYs may be affected by SI variation which somehow gets carried over into the SI of the LANrover's USB output--and THAT we KNOW is important.

 

That is also why a really good power supply for the LANrover has proven worthwhile.  Ideally one that preserves the galvanic isolation offered by the Ethernet connection by blocking leakage currents and keeping the "leakage loop" from being re-formed with the upstream computer, etc.

(I think you all know the sort of power I'm referring too.  Hey, my time to spell out the mechanisms of all this weird stuff is not free, so perhaps my small product plug is earned? )

[Superdad]

46 minutes ago, Speed Racer said:

Alex,

 

I get all of that. However, as a datacenter guy, I know that these interfaces can handle a lot more data than they are being sent for USB 2.0 audio. It's nothing for Gigabit Ethernet. And ethernet is nothing like USB. As I said, there are no ground, power, or clock signals. The data lines are differential. Data that doesn't make it is resent. I see no reason that a quality cable that works properly can be out done by a "better" cable. This is digital information encoded in these signals. These devices provide complete isolation from the PC ground, power, and clocking issues. The PHY is not being corrupted by anything from the sender side and as long as the data is intact, so how can there be a problem there?

 

 

It is a rabbit hole I know.  But remember two things:

1) It is not actually "data" (ones and zeros) until the PHY turns it back into such;

2) We are not talking at all about corrupted bits--we are talking about perturbations taking place inside the device (in the activity of the PHY) which are effecting the ground-plane elsewhere in the device.  And in the case of the LANrover, likely affecting its USB output SI.

 

I am not arguing that this is a major sonic variable (USB signal integrity going into a DAC is vastly more significant), but I have personally compared a pile of 25 foot Ethernet cables (mostly generics of chosen differing construction;  just going Mac to Mac for fetching of shared music files), and the BlueJeans/Belden (10GX) Cat6a came out on top.  Again, not a knock you over the hear SQ difference, but one that certainly withstood multiple swappings (I hope I don't attack the double-blind flies with that statement ).