Update on good Ethernet switch

My experience is a high quality network switch is better than no switch and that seems to be the consensus of a lot of listeners.  

My simplified view as a Mechanical Engineer is that unlike the Analog realm where less is more, the digital realm seems to work differently.  As I understand it, the digital signal is a square wave propagated through wiring.  The “ones and zeros” are created by these square wave pulses.  The 16 bit or 24 bit word length is made up of these ones and zeros as pulses which must be timed perfectly as they leave the originator and are captured by the receiver.  I think of it as a Morse Code operator.  If he misses one “dot” or “dash” thereby getting out of sync with the sender, the message becomes garbled.

I think the D/A converter needs as clean a square wave as possible with exact timing in order to produce a good analog signal.  The less well defined square wave causes the D/A converter to underperform or interpolate, ie guess at what the analog waveform should look like at that moment.  At this point the music is compromised.  

Therefore, the cleaner the square wave the better the chance for the music to get through a D/A converter.  Electrical wiring has some level of capacitance, resistance and inductance which can round off the edges of the square wave.  Electrical noise can be introduced into the wiring through EMI that can make it harder for the D/A converter to interpret the signal.  The network bridge, if my analogy is correct, redraws that square wave and sets the timing by its own internal clock passing on a sharp, clean square wave to the D/A converter.  

It’s like reading for us.  If the print is too small or too large, or the contrast is too low, our reading slows down and we might misread a few words, or have to go back and reread a sentence because it did not make sense.  We might even misread the sentence and end up completely misinterpreting it.  Newspapers and book publishers probably spent a lot of effort into researching in the past century the best print font, size and contrast for readers.   Now we are doing the same thing for music D/A converters.

@tonywinga  you have a few details correct, but you are missing a lot, like 90%.

 You are correct in that Ethernet sends a sine wave, however, depending on the speed, it could be as high as 4 simultaneously waves. Ethernet uses something called a frame to transmit over the wire. Built into the the frame is a crc check, so if there is anything that goes wrong, the frame is discarded. 
also important is that Ethernet standards, regardless of speed over what’s called an Ethernet cable (twisted pair) has building common more noise rejection, that handles most noise, certainly all noise in a residential setting, at speeds less than 2.5Gbit/s. 
 

what you are also missing is that it is a stack of protocols, called TCP/IP, which ensures bit perfect transmission, 100% of the time under normal operating conditions. And you are missing that most music services that provides per song streaming actually downloads a large part, if not the entire song (Qobuz does the entire song) as fast as possible, it doesn’t take very long to download a 24/96 FLAC, seconds, and this download is done over HTTPS, which is encrypted, so you can miss any packets at all or the decryption will fail.

so, now that you have more information, it is easy to see a switch cannot do anything to improve. A poorly design switch can detract and introduce packet loss, but no switch can improve. It is not how the protocol stack functions, in fact it is the opposite of how it works.

Just go away Fred. You are not my saviour.

Bleating on the same drivel. Reminds me of George. 

fredrik222

Ethernet sends a sine wave, however, depending on the speed, it could be as high as 4 simultaneously waves.

Huh? Word salad.

Ethernet standards, regardless of speed over what’s called an Ethernet cable (twisted pair) has building common more noise rejection, that handles most noise,

Word salad. What are you trying to say?

most music services that provides per song streaming actually downloads a large part, if not the entire song (Qobuz does the entire song)

As you know, I've previously shown an experiment that disproves that claim, so please stop repeating it.

Sir Isaac Newton made some observations about falling objects and developed a few basic equations to describe the motion of bodies.  It took some time for people to accept his equations.  Then Einstein came along and showed that Newton’s equations were not entirely accurate.  Still, the observations never changed.

Many of us have experienced and observed that audiophile grade ethernet switches improve the sound of streaming music.  I have even observed it dramatically improve the picture quality on my TV while streaming.  Therefore, it is an exercise in futility to insist that an ethernet switch can do nothing.  Try to be more like Einstein.