Best Ethernet Cards for Streaming

[Superdad]

16 hours ago, Cogito said:

Network signals are at 2.5v which is divided into 2 states, 1 and 0.

That is not correct.  100Mbps Ethernet is 3 voltage levels and Gigabit Ethernet signaling is 5 voltage levels.

 

For convenience I’ll quote from Wikipedia:

10BASE-T transmitter sends two differential voltages, +2.5 V or −2.5 V. A 100BASE-TX transmitter sends three differential voltages, +1 V, 0 V, or −1 V.^[13] Unlike earlier Ethernet standards using broadband and coaxial cable, such as 10BASE5 (thicknet) and 10BASE2 (thinnet), 10BASE-T does not specify the exact type of wiring to be used but instead specifies certain characteristics that a cable must meet. This was done in anticipation of using 10BASE-T in existing twisted-pair wiring systems that did not conform to any specified wiring standard. Some of the specified characteristics are attenuation, characteristic impedance, timing jitter,^{[citation needed]} propagation delay, and several types of noise and crosstalk. Cable testers are widely available to check these parameters to determine if a cable can be used with 10BASE-T. These characteristics are expected to be met by 100 meters of 24-gauge unshielded twisted-pair cable. However, with high-quality cabling, reliable cable runs of 150 meters or longer are often achievable and are considered viable by technicians familiar with the 10BASE-T specification.^{[citation needed]}

100BASE-TX follows the same wiring patterns as 10BASE-T, but is more sensitive to wire quality and length, due to the higher bit rates.

1000BASE-T uses all four pairs bi-directionally using hybrid circuits and cancellers.^[14] Data is encoded using 4D-PAM5; four dimensions using pulse-amplitude modulation (PAM) with five voltages, −2 V, −1 V, 0 V, +1 V, and +2 V.^[15] While +2 V to −2 V may appear at the pins of the line driver, the voltage on the cable is nominally +1 V, +0.5 V, 0 V, −0.5 V and −1 V.^[16]

[Superdad]

12 hours ago, Jud said:

My non-expert layperson's impression is that the answer is anything that doesn't flip a bit isn't "transmitted" through the conversions in the sense of being present on the other side of those conversions.

Yet the ground-plane noise generated by the chips themselves (especially as a byproduct of their phase-noise/jitter) is transmitted.  As always, this is not about the data itself (the decoding to “ones and zeros” is always fine), it is about second order effects. We have wrote about this in our paper—and have performed some measurements of the effects of such “clock-threshold jitter” and leakage currents on transmitted ground-plane noise.

 

Some here will argue that all the above is irrelevant but I no longer have time, energy, or interest in convincing them otherwise. 2,850 very happy EtherREGEN users is convincing enough for me. 

[Superdad]

1 minute ago, TomJ said:

 

Please read, what I wrote. The PHY for copper is mostly integrated in the ethernet chipset of the nic. So for fiber you need an extra active component on the NIC.

Not necessarily—and not typically. SFP/SFP+ cages are simply two differential pairs (one transmit, one receive) and Ethernet signaling to them is SGMII. Many NIC chips that have built-in PHYs also have an SGMII port—which could be directly connected to the SFP cage.
[Due to my firm’s focus on switches I admit much greater familiarity with the port/topology options of switch chips/transceivers for such than with NIC chipset options.]

[Superdad]

23 minutes ago, One and a half said:

You may also have a look at the connector from Halo in the DXE Filter. This is a combo transformer and CM choke in the one package, probably very common.... DX Engineering have two of these back to back, and a pair of (dual) filters is very effective in removing induced noise along a CATx cable. The DXE does not claim HV isolation, only for filtering.

Of course very high frequency AC leakage (say from SMPS somewhere in the system) sails right on through all these devices... 

[Superdad]

7 minutes ago, Markus8 said:

But they are also using shielding in that cable - or did I get this wrong?

Audioquest ties the shield drain wire to the metal shell of the RJ45 connector at only one end.  So use of their cables does not join the component grounds of the two pieces of gear being connected.

[Superdad]

1 hour ago, jabbr said:

it would be really great, and at this point we could discuss the intricacies of common mode noise testing.

 

@JohnSwensonwill be including a bunch of information on this very subject as part of something we are preparing to help guide @TomJ's investigations.  

The place the CM noise matters most is at the differential between the DAC clock's input pin and the ground plane. And the frequencies examined are important also (so far Tom/Eric has been looking much too broadband; 1MHz and far below--especially way down at AC line frequency and multiples--matter more).

That's all I'll comment on at the moment. Will leave the rest to JS.

[Superdad]

On 9/16/2022 at 2:25 AM, Markus8 said:

Many are eagerly awaiting any findings, insights that could be helpful to the overall HiFi community. Could you give as a short update if @TomJhas received/is working with you on that?

I know.  I’m traveling on a 3-week vacation this month and @JohnSwenson is still hard at work on the 3 test boards for EtherREGEN Gen2–and I am loath to distract him from that. Still, we have several times discussed providing at least some of the promised input soon. We really appreciate everyone’s patience in this. 

[Superdad]

10 hours ago, Markus8 said:

Will the Gen.2 apart from functional upgrades like 1GB on B port also provide audible improvements?

 

Although this is not the appropriate thread to discuss EtherREGEN Gen2, I can confirm that our new model will incorporate a number of technical improvements. These include:

--Much lower jitter clock synthesizers (moving away from the SilLabs/Skyworks parts);

--The world's best sine>square wave converter chip (benefiting those using an external reference clock; the first EtherREGEN did not have such a chip at all);

--All different Ethernet switch chip and PHY transceiver set

--Different pre-regulators for all the LT3042/3045 LDOs

--Simplified clock buffer architecture.

 

Development is proceeding apace.  Yet it will be a little while before the whole design comes together onto one board (with isolators and moat) as we have learned from the past that it is more efficient to first create separate test boards for each new section--and to first get the wiring and programming right for each.  So right now we have separate clock synth test board, 'A' side switch chip and PHY test board, and the beginnings of the 'B' side test board.

 

It is impossible to answer the question of "audible improvements" just yet. We will know for sure once the design is unified onto one board--and that will be quite close to actual production; then all you users will get to decide for yourselves.  

It is important to understand that UpTone products are developed with careful technical analysis (and John's 3.5+ decades of engineering expertise) and are not just a tossing together of off-the-shelf chips and designs sprinkled with fairy dust and wishful thinking. And while we try to be a little cost conscious, there are an awful lot of expensive parts of the board--it really adds up.  

So yes, since the internal measurements (ground-plane-noise, jitter, signal integrity, isolation, etc.) will most all be improved, we expect this to translate to better SQ.  We will see hear. That's all we can do. 

[Superdad]

On 10/23/2022 at 1:07 AM, TomJ said:

Hi Markus,
After I had a really open and friendly exchange with @Superdad, it has unfortunately currently become somewhat quiet. I have no idea whether this is due to her current workload or whether they have changed their mind and does not want to give any assistance. If that were the case, I would find it a great pity, because the conversation was very promising not only for me, but also for many silent readers of the threads.
The intension of my investigations is not audiophile snake oil bashing and I would be really happy if manufacturers also engage here to get some light in the dark regarding their products. But if this is not the case, I will continue based on my own findings. However, it would not be pleasant if this then degenerates into a justification discussion as it progresses. Currently would be really a good time to enter the discussion.

Regards,

Eric

 

Hi Eric:

Sorry for the long delay. I just now sent you a lengthy PM with some explanation of the situation and some ideas for proceeding in a positive direction.

We shall continue privately for a bit--and perhaps you will find from the exchanges some worthwhile paths to share on your forum(s) or on this one.

Thanks,

--Alex C.

 

[Superdad]

11 hours ago, TomJ said:

And this is where the TP Link WR902AC came into play.

Hi Eric:

Any particular reason for choosing for test the 2.4GHz/5GHz WR902AC model instead of the simpler and possibly quieter 2.4GHz-only WR802N?