ETHERNET CABLES

@dpcoffey

Listen to what DPCoffey is telling you.  Research Ethernet communications.  I have worked with the technology.  There is full error correction built into the receiver and error correction codes are sent with the data.  The received signal is quantized to a 1 or a 0 by a slicer.  At these short distance you are not getting errors and it will not improve your sound quality.  Spend you money on stuff that will improve sound quality.  

@kennyc

What you are saying is technically incorrect.  The data is encoded, such as using PAM constellations, meaning the digital signal (1 and 0) are converted to a multilevel signal such that each transmitted symbol represents more than one bit.  As the signal is transmitted over the channel, it starts to appear as an analog signal due to the nature of high speed transmission and the channel.  This can be seen as eye diagrams.   The data does not ride or sit on an analog signal.  Just the opposite, the data is the signal.  Upon receipt at a receiver, the voltage of the signal is detected at specific time and that voltage is compared to numerous thresholds.  The signal is then re-created based on the voltage value in relation to the threshold to re-create the PAM signal.  For example, if the threshold are 0, 1, 2, and 3 in a PAM4 system, and if the received value is 1.1, then the signal is re-created and a 1 value is output.  The noise or timing error that causes the received signal to be 1.1 instead of 1 does not have any impact after slicing (quantization).  If the received signal is 1.4, it is sliced to 1.  If the received signal is 2.7, it is output at a 3 from the slicer.  Again, that noise or timing error does not have any impact.  The signal is recreated and the noise discarded.  The PAM4 signals are then decoded back to 1s and 0.  This is a good link to read.  

Best of all, the bit error rate of the home ethernet networks is amazingly good.  There are CTLE, FFE, DFE that all clean up the signal to correct channel degradation.  Plus, there is error correction such that each packet includes a error correction data that is used to detect and correct errors. And if, in the very rare instance, an error is not correctable, the packet will be resent and filled into the buffer of the streamer.  A typical bit error rate is 10 to the minus 13, which means that for every 10,000,000,000,000 bit sent, there is 1 error.  People do not hear this.  

Even if there were issues (which there are not) between the steaming source, which could may be an another city or state, and your streaming box, the last 3 feet inside your house is the least of your worries after the likely awful path the data has already taken for the miles or hundreds of miles to get to you.  And, over ultra short 3 foot link, there is so little signal degradation that we should not even be having this conversation.  I encourage anyone to correct any technical errors I have made to advance the conversation and improve the accuracy of this forum.  

I wanted to provide a few more details that may be helpful. What is important to remember is while there is signal degradation that occurs in the channel (cable) the processing inside the network receiver and transmitter (pre-equalization) overcomes the signal degradation. Plus, we are talking about a very short run between audio components. The engineers who designed the Ethernet system know what they are doing.

Ethernet cable consists of 4 twisted pair conductors, so 8 conductors total. Each twisted pair carries a differential signal meaning the voltage on the pair is not referenced to ground, but is the difference between the voltage on the two conductors that form the pair. The audio world understands this as being very similar to a balanced audio cable. Therefore, if noise or interference does couple into one or more of the pairs it does not matter because it will for the most part couple into both conductors, which will not change the value of the voltage difference between the two conductors. This takes care of noise, EMI, RFI that may couple into only conductor, and any small amount that couples into only one conductor is accounted for by the slicer that I described above. This is a good link.

As to shielding/grounding, it usually unnecessary unless there is a specific reason and can actually harm the audio outcome. One reason it is not needed is because the signal is a differential signal as discussed above, and the signals values are not referenced to a ground, but to the other voltage value of the two twisted pair. Second, the shielding with grounded will couple one device chassis ground to another. If/when once devices ground is at a different voltage than the ground value of a second device, then couple these two together with a grounded end to end shielding will couple the bad or non-zero ground into the other device(s). In my opinion, if this finds its way into analog devices such as the DAC, where the analog signal exists, that could spell trouble. One might think ground is the same everywhere, however each device may have leakage, coupling, interference, or a bad ground. We don’t what that device’s poor or different ground to leak into other device chassis.  We don't want that coupled from one device to the next.  Trust the designers of the Ethernet system. Grounded/shielded cables can be needed for industrial applications or where specific issues exist, but not for typical 3 foot runs in an audio system.

This is a good link that explains the issued better than me, with multiple posters each providing the information slightly different with pictures. I find this help me understand.

Finally, as to retiming, the data gets retimed at every network device, such as at the switch or router or streamer. This can be performed by retimers, or other hardware in the network interface This is all taken care of so we don’t need to worry about it. This link is a good screen shot because it shows the eye diagram I reference above, with and without re-timing. Notice the improvement due to the retiming module Keep in mind, the digital signal is completely regenerated within the network interface after retiming anyway.

This thread is an example of personal truth and absolute truth. One group really believes that the 3 foot network cable affects the sound and the other group greatly believes that it does not affect the sound. @fredrik222 is spot on, however at the end of the day, if a person perceives a better sound in their brain, perhaps that is all that matters. We all just want something that sounds (based on our brain’s interpretation) pleasing. This is a very interesting thread.

@jerrybj   

I don't find his posts be snarky.  A forum should welcome diversity in thought and input.  Every aspect of music reproduction is influenced by the technology and it would unfortunate to lose that perspective.  

Everyone keeps focusing on the cable and noise that may couple into the signal on the cable, but the key here is what occurs after the signal leaves the cable in the Ethernet receiver. In a 3 or even 10 foot run, a tiny amount of noise can couple into the signal. However, in the receiver, the received signal is compared to reference or threshold values and based on this comparison an entirely new signal is created. The received signal (and any noise) is effectively discarded. The newly created signal is output at the correct voltage level without the noise. This new signal could be transmitted for another 300 feet (or some other distance based on data rate) or passed on to the internals of the streamer for further decoding. At each hop in the communication path, the signal is re-created and retimed thereby creating an entirely new and clean signal.

While analogies are often imperfect (and I am told, mine are often terrible), think of a piece of blank 8.5x11 inch paper sent through the mail. The post office may crumple it or bend it such that when received at your house, it is crumpled (noisy). However, the person receiving the paper can tell that it is blank piece of 8.5x11 inch paper (albeit crumpled) and they get a brand new piece of paper out of the drawer and toss the crumpled paper in the trash. The received paper was just used as a reference to know what size of new paper to pull from the drawer. Same with an Ethernet receiver. It compares the voltage value of the received signal to a threshold value, to create an entirely new signal (without noise) and the received signal (with noise) is discarded. This is unique to the digital domain and does not occur in the analog domain.

In a PAM4 system, typical of Ethernet, the signal is transmitted at one of four voltage levels, such as 0, 1, 2, or 3 volts. The received signal will vary some what due to noise and effects of the channel. For example a signal originally at 2 voltes, that is transmitted over a long cable run, could be received at 2.2 volts or even 1.7 volts. It would be compared to the four voltage levels and an entirely new clean signal at 2 volts outputted, which is the closest voltage value. The original signal is effectively discarded. This is the benefit of digital communication over analog communication over long distances (and short distances).

Based on this method of operation, the tiny improvement a better cable provides will not yield a different outcome when the signal is re-created.  A signal transmitted at 2 volts might be received with a shitty cable at 2.1 volts and with a amazing cable at 2.08 volts (small improvement).  Both will be compared to the threshold values (0 volt, 1 volt, 2 volts, 3 volts) and the receiver will output a clean 2 volt signal.  

Peace.

@soix 

Mostly what I am doing is sharing the technical details of what occurs inside Ethernet receiver.  There were some earlier posts that were providing technically incorrect information regarding Ethernet receivers and the signal.  I have always loved technology and discussing it with others.  I am not blinding walling off any suggestion of trying a cable.  Sure people can try.  If they perceive a difference then they should keep the cable.   I don't know if the sound improvement is due to improved sound waves or their perception and IMO maybe it doesn't matter as long as they perceive an improvement.  As you mentioned with CD's, there may be technical aspects we don't understand that could affect sound quality.  If there are, I can not identify them.  As I mentioned above, at the heart of this hobby is the enjoyment so if someone enjoys an expensive cable then they should buy it.  It never hurts to better understand the technology that makes our systems work and be a more informed buyer.  

@soix

I will definitely try an expensive ethernet cable to see if I can hear a difference after I buy a separate DAC. I will only do it with a blind test though because otherwise it would be meaningless. I will put a blindfold on or put up a barrier and my wife can swap the cables. It will be easy to swap out a 3 foot cable between my switch and steamer. I personally don’t think I will hear a difference because I know how Ethernet receivers work. If it was not a blind test, I would introduce that bias into my listening. I can have my son try it as well. He is younger and plays acoustic guitar so he has a good ear.

In my mind, I believe it will be a waste of time because I know what occurs in an Ethernet receiver and my test won’t change anyone’s else mind because I will be told I have shitty ears or a shitty system, both of which may be true. It will still be fun to try it and who knows maybe I am wrong. We do blind tests with tequila as well to see if we can guess the brand or pick the expensive one.