Does Ethernet induce Jitter?

[ToddC]

Simple enough question, I just have never seen this topic addressed anywhere....  I'm using a network file share to host my music and I access it via Foobar on another computer that's hooked into my 2 channel system.  My ethernet connection between the server and the player computers is Gigabit ethernet.  My assumption is that the receiving computer's ethernet would buffer the data coming in, so that no jitter would be introduced by the transmission, but it would be introduced after the packets are put back together and the digital signal is processed (just guessing though)

Do you think that  Ethernet (between 2 computers) is a better solution than a hard disk or SSD in a single computer setup?  Would the hard disk on the Server cause jitter that would be fed through the system, or is it somehow removed during the conversion to the IP packets?  I have an Async USB dac, would that clean up any jitter that might get introduced in this process?  Your thoughts are appreciated.

[richidoo]

There are a lot of data transfers from the original file stored on HDD until the data reaches DAC chip. Each media has it's own "buss frequency." 1's and 0's must always be moved in a stream at fixed frequency. The sender and receiver must speaker the same language and speed. Consider the data moves through the PCs main processor at 3GHz, but PCI buss is about 1GHz, RAM is another speed, as is USB, Firewire, etc.

Ethernet packetizes the data into chunks and applies data integrity rules so the journey through the "great unknown" network is safe. Packet 100 can arrive after packet 1000 and it still works. If it doesn't work, there's a program to fix it. After the packets are reassembled, the resume their clocked journey through various hardware parts until they reach the DAC at 44.1kHz, the slowest leg of the journey by far.

Each leg except networking is clocked. Network is clocked too, but packet arrival order is not guaranteed so clock is not really relevant. So all these different "sample rates" and no data loss. Jitter is irrelevant there, because there is no conversion to analog. No brains to notice the jitter. The computer parts don't care about jitter, they only care about data integrity and speed.

So the "dejittering" is only needed at the DAC. A clock does the dejittering when connected to the transport and DAC it synchs them to the same speed. Various ways of doing this and varying quality of clock result in different types and amounts of jitter.

[ToddC]

So in an async dac the clock used to sync the a-d conversion would be critical to keeping the jitter low, but not much else in the previous digital chain.  This would imply that other differences observed in audio quality caused by equipment (prior to the digital to analog conversion) such as usb cables, or SDD drives, would have their effect by changing some other non digital parameter, such as RFI, vibration or Power supply interaction. 

[Carlman]

My thought is that with anything in the digital chain, the less corrections that need to be made along the way, the better the audio output will be.

Why a hard drive makes a difference, especially one used remotely on a network, is a tough argument to make and I wouldn't get into it with anyone unless they've been with me to witness the same experiment.  I had a defective hard drive on my server that I replaced and I definitely noticed a nice little improvement in sound quality while using a desktop in my sound room that accessed it.  I have since learned that others have experienced the phenomenon that hard drives generally do make a difference. 

When I did my networking I used cat6 and sprayed all the connectors with Pro-Gold.  (ridiculous I know but I figured, why not?) 

In any case, to the original point, 'jitter' specifically is introduced with clock issues in the analog conversion process and Rich has explained this well.  However, there are other factors besides jitter that will make a difference in ultimate output quality.

-C