What's the consensus on ethernet cables?

[marce]

12 hours ago, Axial said:

Anyone here who have measured them to see any graphical differences, plus participating in blind listening tests to check the auditive similarities, or slight variances?

 

Forget the money, I'm talking strictly sound quality between brands, construction, connectors, metallurgy, science, mathematics, physics, concrete factual studies. 

 

I'm interested in higher knowledge, in the-matter-of-fact truth.

Your starter:

http://www.x2y.com/filters/TechDay09kr_hpa_Track2_1_Precision_Analog_Designs_Demand_GoodPCBLayouts _JohnWu.pdf

 

Some good stuff on here:

https://www.bethesignal.com/bogatin/

 

This ain't bad:

http://www.sigcon.com/

 

Then there's Ralph Morrison, move on fields of electronics:

https://www.amazon.com/Fields-Electronics-Understanding-Using-Physics/dp/0471222909

 

Henry Ott for the EMC side:

http://www.hottconsultants.com/

 

All the IC manufacturers sites are full of papers, design guides, technical notes on every interface you can think of, including from the likes of Intel, complete guides to lay out motherboards. Funnily reading all this stuff and playing with it you realise bits are periodic current spikes and "periodic current spikes are periodic current spikes" by any other name.

An example of the detail we often have to work to today, especially with arrays is signal traces for critical timing matched to within 1.3nS (0.2mm) and down to pS (0) in some instances. With microwave it gets even sillier as the following guide illustrates, you will notice a lot of the structures (components) don't have direct electrical conductivity, here youn have to think of the signal as a propagating wave: 

https://www.qsl.net/va3iul/Files/RF-Microwave_PCB_Design_and_Layout.pdf

 

Its fun.

 

[marce]

17 hours ago, KingRex said:

 

 

 

 

FWIW all my digital audio is routed through a Linear Solution Switch.  You can clearly hear the upgraded switch is better than a decent Netgear switch. 

This:

https://audiobacon.net/2018/09/28/the-linear-solution-ocxo-audiophile-switch-reference-ethernet-cable-the-missing-pieces-of-digital-audio/

Not the best modification of a clock I have seen, in fact it could be the worse, where is the return for the clock, why aren't the supply lines twisted pair and the wire soldered to the chassis has no mechanical relief. Its a good example of how not to connect a clock module.

[marce]

Maybe they don't understand how Ethernet cables work.

[marce]

19 hours ago, KingRex said:

 

I went with non shielded.  If the shield is not terminated on one end, it can act as a choke.

 

I have experimented with CAT7 cable used as a spdif cable.  There is a big sonic difference between leaving the shield on and stripping it off.  It has a very choked and congested with loss of high frequency with the shield on.  That was on a 22 foot run.

 

There was also a big difference in performance based upon the connector I used.  I have a Furutec Connector that is screw down and that was way better than some other solder on brand I tried.  I wonder how much of it was the solder as opposed to the metal in the body.

Wrong cable for the job... Also 22 foot for SPDIF is a bit extreme to say the least.

I am intrigued as to how the cable preformed DSP functions and reduced the treble, this could be a major discovery....

[marce]

17 hours ago, KingRex said:

Ok, I guess this is DC, not AC.  I don't know much about DC.  Does that mean there is no way for the voltage and current to induce to the metal shield.  

 

Every signal has a return path, for low frequency signals this path is the path of least resistance, as signal frequency increases the return current follows the path of least inductance, this is the main path for return current once we get to 1MHz and above... A signal always has two conductors the signal wire and its return path, the signal travels in the space between these two wires. For co-ax cable the shield is also the return path and the EM fields are within the shield. Other signals can be single ended or differential and can be transmitted by either shielded or unshielded cable, the shield is generally connected to the chassis and should be separate from the signals return path, usually the system 0Vs. 

[marce]

11 hours ago, mansr said:

I wish people would stop using that phrase. As you well know, electric current flows through all possible paths in inverse proportion to their resistances.

I am specifically talking about return current path :

http://www.sigcon.com/Pubs/news/8_08.htm

Even Dr Howard Johnson uses that terminology as do many others involved in signal integrity/high speed design.

So I would say that the terminology used in this situation is correct and is to illustrate how return path current flows, a critical part of understanding signal flow.

[marce]

17 minutes ago, marce said:

I am specifically talking about return current path :

http://www.sigcon.com/Pubs/news/8_08.htm

Even Dr Howard Johnson uses that terminology as do many others involved in signal integrity/high speed design.

So I would say that the terminology used in this situation is correct and is to illustrate how return path current flows, a critical part of understanding signal flow.

To be totally correct the terminology should be the return current path is always the path of least impedance, at low frequencies (<low kHz) the path of least resistance dominates as frequency increases (low kHz +) the path of least inductance dominates...

Happier😀

[marce]

As Paul has said and as frequency increases the signal will follow the path of least inductance, that means it will follow the path with the minimal loop area and thus track the signal as closely as possible. Also the harmonics are always the base frequency and greater, so once you get to 1MHz or above the return current density will be directly under the trace where possible. Audio frequency range is a pain because ALL the return path choices are used as we have very low frequency signals that want to follow the path of least resistive impedance and higher frequencies that want to follow the least inductive impedance path, understanding this is a great way of understanding ground loop problems and mains hum. Rane have some good papers, mostly though referencing balanced audio and the pin 1 problem... But this thread is Ethernet signals, so high speed rules apply....

Cables often have simpler paths than PCB's, in the case of Ethernet signals the twisted pair for a signal and return path for each other (the beauty of balanced signals) you don't have to worry about a continuous 0V etc. On a PCB the return path will be in the case of signals from a few hundred Hz will be the path of least inductive impedance, generally the plane adjacent to the signal, be it a 0V plane or a voltage plane.

Again the terminology I used is used quite extensively in the signal integrity world and again I believe is correct in this context, hense why I used Howard Johnson as an example of the use of this terminology. As the majority of people reading this thread have far less hands on experience of signal transmission, keeping things simple is the best bet, even down to explaining signal return paths as one path.

Lets get funky...

http://www.hottconsultants.com/pdf_files/Digital Logic Current Flow.pdf

 

As said signal integrity does not end at the PCB, we have to deiced of how we get the signal from board a to board b, and what type of cable is best suited to the job and also what connector to use...

[marce]

3 hours ago, mansr said:

Current always flows in a closed loop, and the same rules apply in all parts of said loop.

 

That's talking specifically about a high-frequency signal in a trace over a ground plane on a PCB. As it happens, the inductance increases considerably with distance from the signal trace, confining most of the return current to a small area directly beneath it. This is contrasted with the DC case wherein the return current is rather spread out with the highest density around the most direct path between the two points, regardless of the "forward" conductor routing.

 

 

Better have a go at this guy and all, I can't believe he's using the same terminology I used...😀 As punishment I shall read some Synergistic Reasearch White papers ...

tonight.

http://www.emcs.org/acstrial/newsletters/fall08/tips.pdf

 

[marce]

1 hour ago, mansr said:

I have come across people on this very forum believing that if you connect two resistors, 1 Ω and  2 Ω, in parallel, all the current will flow through the smaller one because it provides the "path of least resistance." That's what I'm trying to avoid. The sources you're citing are addressing an audience who already knows this isn't how electricity works, and such shorthand terminology is thus unlikely to cause any confusion.

I understand your point now, as said I was using the terminology only in reference to return current path, exclusively. The thing is with digital signals (and analogue) is understanding the return path for the signal, it is as critical as the signal and often some audiophile advice on grounding, signal returns etc. are at odds with what is required for a low noise system with maximum signal integrity. Often signal integrity is discussed, often with minimal reference to all the issues that have an effect on the signal, return path being a critical one. Trying to get across all the issues and develop a better understanding of signal propagation would IMO especially with digital get the point that "Bits are Bits" as well as the understanding that EMC (noise) and signal integrity are two sides of the same coin.

It's been an interesting little discourse.

One of the things that has bugged me regarding a lack of understanding of return paths with digital audio is clock modifications, both here and on DIY audio... To illustrate my frustration have a look at this link...

https://audiobacon.net/2018/09/28/the-linear-solution-ocxo-audiophile-switch-reference-ethernet-cable-the-missing-pieces-of-digital-audio/

The clock signal is connected by a single wire, the return path is going to have to meander round the board and return by one of the supply leads. Whilst a wiring issue, the problem and issues are best illustrated by a slot in a PCB ground plane. The lack of an intimate return wire for the clock creates a huge discontinuity in the return path, not only creating ground noise (at the fundamental clock frequency and its harmonics, to be avoided at all costs) but the clocks signal integrity will be compromised...

[marce]

16 hours ago, Ralf11 said:

I just want to make sure people see at least one important thing in marce's post above.

 

Once you get past DC resistance and the "Mansr Problem" of least resistance, you get to Level 2 of Misunderstanding.

 

Level 2 is that you have to use impedance for all AC signals.

 

But "Wait!  There's more!  The Word Salad Shooter of circuits shows us that is just Level 3.

 

You next need to realize that impedance has to be deconstructed* by freq. bands.

 

very low frequency signals that tend to follow the path of least resistive impedance and higher frequencies tend to follow the least inductive impedance path  (I removed the motivational "want to" and replaced it with "tend to") in marce's quote

 

 

* just a nod to Derrida for the liberal arts majors... (or as we say down in NoLa, Jock-Imo)

Sorry but you get an F. My reference in to return currents only, so there is no tend to or maybes they do follow the said paths, its a wavefront travelling between the hot conductor and the conductor that forms the return current path with those nice squiggly little lines that Maxwell mentioned, the E and H fields.

Understand return paths is a big part of understanding signal integrity, I can't honestly believe the conversation has got so derailed, I am talking exclusively about the return current path...

[marce]

11 hours ago, barrows said:

Isn't there a phase noise tradeoff here though?  (Believe me, I am not condoning the nonsense with the LS switch!).  For example, how far a distance does a clock trace need to travel (considering proper layout) before a buffer with differential output will be an advantage?  As the buffer is always going to add some noise/inaccuracy of its own.  I could see this approach being sensible for a USB DAC, where one might have the oscillators directly adjacent to the DAC (chip or otherwise), and then their output is also sent back to the USB receiver (and through isolators).

This depends on many factors, layout, is the clock point to point or distributed, distance ...

Many interfaces now have differential clocks as standard, or the clock is generated as a differential signal from say an FPGA, this is useful where the return paths cant be guaranteed, such as when the adjacent layer is a power plane with multiple supplies as the clock is less affected by crossing splits in planes if its a differential signal routed using differential pair routing.

[marce]

2 minutes ago, mansr said:

Every current is a return for something.

Return current paths are hammered home in EMC and Signal Integrity training and information, for noise free design and signal integrity you have to understand it and the implications of not considering it when routing a signal...

[marce]

FFS I have mentioned cables and explained in an earlier post that it applies to cables as well, signal Integrity and EMC don't end at the PCB and biether does the signal, you have to use cables... For someone so intelligent I would have presumed you could understand my posts instead of trying to be deliberately obtrusive, though I have seen this superior put down attitude before and was not impressed.

I shall not waste anymore of anyone's time as I have far better things to do with my time.

 

[marce]

6 hours ago, Blake said:

I think Crenca gets probed on the daily.

Wow, impressive reply, gets my vote for reply of the year!