The inaudibility of Ethernet

Jinjuku

New Member
Apr 18, 2011
228
0
0
#1
I wanted to start a thread on the topic of Ethernet and some of the presumed audibility. Siemons and T.I. have some great technical papers that bring some light to bare on the robustness of twisted pair data cabling.

With the understanding that often real-time adjectives are used to describe decidedly non-realtime systems and events I want to dive right in.

First is the susceptibility of Ethernet to gross error. Siemons Paper: http://www.siemon.com/us/white_papers/97-10-02-presentation.asp

Worst Case Channel Model.PNG

So I think we can all agree that a home environment comes no where close to the above.

measured field strength.PNG

Several items were used to generate massive amounts of EMI.

As their conclusion:

"No packet errors were detected for either the ‘generic' or ‘enhanced' category 5 channel configuration regardless of EMI source type, source location, or duration of exposure."

Keep in mind the cabling is UTP. Not STP.
 

Jinjuku

New Member
Apr 18, 2011
228
0
0
#2
Siemons has another great paper on the 'Antenna Myth':


Screened and Shielded Cabling - Noise Immunity, Grounding, and the Antenna Myth


In this paper Siemons looked at three contributors:


Differential noise (Vd): Noise induced from an adjacent twisted-pair or balanced cable
Environmental noise (Ve): Noise induced by an external electromagnetic field
Ground loop noise (Vg): Noise induced by a difference in potential between conductor ends


Environmental noise disturbers: Environmental noise is electromagnetic noise that is comprised of magnetic fields (H) generated by inductive coupling (expressed in A/m) and electric fields (E) generated by capacitive coupling (expressed in V/m). Magnetic field coupling occurs at low frequencies (i.e. 50Hz or 60 Hz) where the balance of the cabling system is more than sufficient to ensure immunity, which means that its impact can be ignored for all types of balanced cabling. Electric fields, however, can produce common mode voltages on balanced cables depending on their frequency. The magnitude of the voltage induced can be modeled assuming that the cabling system is susceptible to interference in the same manner as a loop antenna [1]. For ease of analysis, equation (1) represents a simplified loop antenna model that is appropriate for evaluating the impact on the electric field generated due to various interfering noise source bandwidths as well as the distance relationship of the twisted-pairs to the ground plane. Note that a more detailed model, which specially includes the incidence angle of the electric fields, is required to accurately calculate actual coupled noise voltage.

Where: is the wavelength of the interfering noise source

A = the area of the loop formed by the disturbed length of the cabling conductor (l) suspended an average height (h) above the ground plane
E = the electric field intensity of the interfering source

The wavelength, , of the interfering source can range anywhere from 5,000,000m for a 60 Hz signal to shorter than 1m for RF signals in the 100 MHz and higher band. The electric field strength density varies depending upon the disturber, is dependent upon proximity to the source, and is normally reduced to null levels at a distance of .3m from the source. The equation demonstrates that a 60 Hz signal results in an electric field disturbance that can only be measured in the thousandths of mV range, while sources operating in the MHz range can generate a fairly large electric field disturbance. For reference, 3V/m is considered to be a reasonable approximation of the average electric field present in a light industrial/ commercial environment and 10V/m is considered to be a reasonable approximation of the average electric field present in an industrial environment.


The one variable that impacts the magnitude of the voltage coupled by the electric field is the loop area, A, that is calculated by multiplying the disturbed length of the cabling (l) by the average height (h) from the ground plane. The cross-sectional view in figure 3 depicts the common mode currents that are generated by an electric field. It is these currents that induce unwanted signals on the outermost conductive element of the cabling (i.e. the conductors themselves in a UTP environment or the overall screen/shield in a screened/fully-shielded environment). What becomes readily apparent is that the common mode impedance, as determined by the distance (h) to the ground plane, is not very well controlled in UTP environments. This impedance is dependent upon factors such as distance from metallic raceways, metallic structures surrounding the pairs, the use of non-metallic raceways, and termination location. Conversely, this common mode impedance is well defined and controlled in screened/fully-shielded cabling environments since the screen and/or shield acts as the ground plane. Average approximations for (h) can range anywhere from 0.1 to 1 meter for UTP cabling, but are significantly more constrained (i.e. less than 0.001m) for screened and fully-shielded cabling. This means that screened and fully-shielded cabling theoretically offers 100 to 1,000 times the immunity protection from electric field disturbances than UTP cabling does!

And finally:

Well balanced (i.e. category 6 and above) cables should be immune to electromagnetic interference up to 30 MHz.
 

c1ferrari

Member Sponsor & WBF Founding Member
May 15, 2010
2,115
33
485
#3
Hi Jinjuku,

Is my understanding accurate that in post #1, UTP cable acquits itself satisfactorily wrt EMI; however, in post #2, a theoretical advantage is conferred by STP cable :confused:
Thanks.
 

Jinjuku

New Member
Apr 18, 2011
228
0
0
#4
Hi Jinjuku,

Is my understanding accurate that in post #1, UTP cable acquits itself satisfactorily wrt EMI; however, in post #2, a theoretical advantage is conferred by STP cable :confused:
Thanks.
In the 1st paper the factor is introduced noise not effecting the ability for data to be reliably delivered.

The 2nd paper is talking about noise immunity in both UTP and STP cable.

So you have between the two write up's really good BER characteristics in very EMI noisy environments and just phenomenal noise immunity.

The take away is a standard home environment shouldn't come anywhere near this level of interference.
 

c1ferrari

Member Sponsor & WBF Founding Member
May 15, 2010
2,115
33
485
#5
Thank you, Jinjuku, for the clarification. Should the price differential between UTP and STP be non-prohibitive, I would likely opt for the latter.
 

Jinjuku

New Member
Apr 18, 2011
228
0
0
#6
STP CAT cabling is generally considered affordable. I can't speak for your wallet :)

STP isn't needed in the home environment. But if you decide to implement it: it requires more thought and planning plus it's more of a pain to install.

Glad you found the information useful.
 
Last edited:
Apr 3, 2010
15,814
9
0
Seattle, WA
#7
Great info indeed. Ethernet noise is not an issue anyway unless we can show that it bleeds into the rest of the system. The challenged there is that the rest of the system that receives packet traffic is so noisy as it executes the TCP/IP protocol that such noise becomes well, noise :). The whole system needs isolation from audio circuits and that two for one would fix any crosstalk.

That's my theory and I am sticking to it. :D Seriously, I plan to one day test the effects of these cables.
 

Jinjuku

New Member
Apr 18, 2011
228
0
0
#8
Great info indeed. Ethernet noise is not an issue anyway unless we can show that it bleeds into the rest of the system. The challenged there is that the rest of the system that receives packet traffic is so noisy as it executes the TCP/IP protocol that such noise becomes well, noise :). The whole system needs isolation from audio circuits and that two for one would fix any crosstalk.

That's my theory and I am sticking to it. :D Seriously, I plan to one day test the effects of these cables.
Archimago has some great measurements various Ethernet cables (all the same within the margin of testing variance). T.I. also has a white paper showing, if I recall correctly, nominal BER even when they midwifed an over 500 foot run of CAT5.
 
Last edited:
Apr 3, 2010
15,814
9
0
Seattle, WA
#10
Archimago has some great measurements various Ethernet cables (all the same within the margin of testing variance). T.I. also has a white paper showing, if I recall correctly, nominal BER even when they midwifed an over 500 foot run of CAT5.
Thanks. I am not worried about bit errors. Error correction in TCP/IP takes care of that and any glitches there would be plainly audible in audio pauses and pops/clicks.

Love to see the Archimago tests. Do you have a link? If not, I can search for them.
 

Jinjuku

New Member
Apr 18, 2011
228
0
0
#13
I don't know about that. The unicorn Ethernet cables Michael Lavorgna hears a readily apparent difference in weren't in the mix.
 

Jinjuku

New Member
Apr 18, 2011
228
0
0
#15
So far I'm only aware of 6 people that can hear a difference in Ethernet cables. You aren't one of them (neither am I apparently):D
 
Feb 25, 2016
462
2
18
Zhejiang
#16
I'd hazard you're barking up the wrong tree if you're interested in uncovering audible differences between ethernet cables. If there are differences other than placebo it'll be down to 2nd order effects, just as with USB cables. The differences won't be due to bit errors.
 

c1ferrari

Member Sponsor & WBF Founding Member
May 15, 2010
2,115
33
485
#18
Thanks. I am not worried about bit errors. Error correction in TCP/IP takes care of that and any glitches there would be plainly audible in audio pauses and pops/clicks.
Amir, why aren't you worried about bit errors...doesn't bit error reflect jitter? Thanks.
 
Apr 3, 2010
15,814
9
0
Seattle, WA
#19
Amir, why aren't you worried about bit errors...doesn't bit error reflect jitter? Thanks.
No they don't in this system. TCP packets have checksums. That checksum allows the TCP layer in the operating system to detect bit errors. If that happens, it discards the packet and the sender will retransmit it. So unless the link is broken solid in which case you get a glitch or playback stops, you are good on bit errors.

The player software and DAC are downstream of the above. They have no ability to see or know about packet bit errors. They either get reliable bits or nothing.

Now if there is noise that is riding on Ethernet connection that gets coupled to an internal DAC in a streamer, then it could cause jitter and other distortions. The measurements provided before in that test fixture shows that such noise does not bleed. Of course broader tests are necessary to make sure that test was representative.

Let me know if I answered the question adequately :).
 

Jinjuku

New Member
Apr 18, 2011
228
0
0
#20
I'd hazard you're barking up the wrong tree if you're interested in uncovering audible differences between ethernet cables. If there are differences other than placebo it'll be down to 2nd order effects, just as with USB cables. The differences won't be due to bit errors.
It won't be down to even or odd order effects. It's a data cable not an audio cable. There can't be differences.
 

About us

  • What’s Best Forum is THE forum for high-end audio, product reviews, advice and sharing experiences on the best of everything else. A place where audiophiles and audio companies discuss existing and new audio products, music servers, music streamers and computer audio, digital to audio converters, turntables, phono stages, cartridges, reel to reel, speakers, headphones, tube amplifiers and solid state amplification. Founded in 2010 What's Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals we enjoy learning about new things and meeting new people and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing