Best Ethernet Cards for Streaming

[TomJ]

Hello guys,  

 

as already announced, I made a comparison measurement between the Intel i350 V2 T2, the Intel i350 V2 T4 and the JCAT NET Card XE in an RockPro64 SBC. The measurements were made according to the state of my research so far. All values can only be considered relative to each other, since many marginal parameters are relevant for the measurements, which can change when measurements are made by other people or in other setups.

 

 

The results are published in the german open-end-music forum: https://www.open-end-music.com/forum/privatforen/thomas-michael-rudolph-tmr/651284-messungen-von-ethernet-infrastruktur-switches-nur-lesen?p=680071#post680071.

 

Since I don't want to duplicate the report, if you don't speak German, please use Google translate:
https://www-open--end--music-com.translate.goog/forum/privatforen/thomas-michael-rudolph-tmr/651284-messungen-von-ethernet-infrastruktur-switches-nur-lesen/page3?_x_tr_sl=de&_x_tr_tl=en&_x_tr_hl=de&_x_tr_pto=wapp#post680071

If some topics are not understandable by the translation, please write to me.

 

The results of the measurements will be discussed in the following thread:
https://www.open-end-music.com/forum/privatforen/thomas-michael-rudolph-tmr/651298-diskussion-ueber-messungen-an-ethernet-infrastruktur#post679797

But of course also here.

 

Best regards,

 

Eric

[The Computer Audiophile]

Interesting results. 
 

I’ve been using a Ravenna based 12 channel system for 24/48 - 24/352.8 music. The network card for this system is incredibly important, as are the correct settings within the card config. A system can go from not working, to working pretty well, to working perfectly, based on the config. 
 

I wonder if network card config could change your results or even the content that’s sent through the card could change the results. 12 channels of DXD can be pretty demanding vs two channel 16/44.1. 
 

Here are some network card recommendations from Merging - https://merging.atlassian.net/wiki/spaces/PUBLICDOC/pages/4817173/RAVENNA+ASIO+recommended+Network+Adapter

 

 

I’ve been considering this card - https://daleproaudio.com/products/merging-technologies-net-msc-gbex1-gigabit-aes67-ravenna-pcie-card-for-horus-system#description

[TomJ]

From my measurement experience the most driver settings doesn’t change much regarding Jitter and CM noise. But I only do measurements with 100Base tx.

The only setting, but only available on Linux, which has an impact on CM noise is the MDI/X setting. On Linux you can set the mode to MDI for the outgoing traffic and MDIX for the incoming, So it is secured that the data will only be sent on the 1/2 pair, not on the impedance critical 3/6 pair. This impacts CM noise.

 

I friend of mine works with 8 channel stuff and he also reports, that the settings of the driver are difficult, but I think it has more to do with throughput of the card itself than with Jitter and noise.

[The Computer Audiophile]

3 minutes ago, TomJ said:

From my measurement experience the most driver settings doesn’t change much regarding Jitter and CM noise. But I only do measurements with 100Base tx.

The only setting, but only available on Linux, which has an impact on CM noise is the MDI/X setting. On Linux you can set the mode to MDI for the outgoing traffic and MDIX for the incoming, So it is secured that the data will only be sent on the 1/2 pair, not on the impedance critical 3/6 pair. This impacts CM noise.

 

I friend of mine works with 8 channel stuff and he also reports, that the settings of the driver are difficult, but I think it has more to do with throughput of the card itself than with Jitter and noise.

It would be interesting to see how these devices change under loads. If the noise and jitter are effected by 12 channels of DXD vs 2 of CD quality.  I think heat certain goes up on the devices with this much data, but I’m not sure if that effects the measurements you’re doing. 
 

I really don’t know, but am curious. 

[TomJ]

I have no 12 channel setup to test this.

I will do measurements with maximal load for 100Base TX and report.

[TomJ]

5 hours ago, The Computer Audiophile said:

 

I’ve been considering this card - https://daleproaudio.com/products/merging-technologies-net-msc-gbex1-gigabit-aes67-ravenna-pcie-card-for-horus-system#description

This card seems to me to be an Intel EXPI9301CTBLK 50$ card with the 82574L PHY : https://www.bechtle.com/de-en/shop/intel-gigabit-ct-desktop-adapter-pcie--538196--p

At the start of this thread I posted measurements (However, not with the measuring equipment I now have) of the Intel Pro 1000PT with the 82571EB PHY, which is similar to the  82574L.

 

Maybe they have some special souce driver, or driver settings. But if it only comes to settings, the i350 should have at least the same setting parameters.

[TomJ]

On 10/22/2022 at 9:48 AM, Markus8 said:

 

Hope @Superdad you have had a good vacation time and Gen2 is also progressing nicely. 
And did you also provide @TomJ some feedback so he is able to continue his very passionate work?

I online forums worldwide flooded with the „data is data / everything is perfect, there cannot be any audible difference on Ethernet cables and switches“ and findings or measurements would be a real game changer here. 

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

 

[itomonje]

On 10/22/2022 at 4:22 AM, TomJ said:

To give a preview on the measurements report:

The Intel i350 V2 is one of the best devices I have measured so far regarding jitter and CM noise.

Anyone looking for a network card should seriously consider this one.

 

[itomonje]

i think you men the intel i350 t2 card. did you compared with rhe jcat also? is As i can read the best choice up till now is the i350 t2 so no other choice to get that so low values in jitter and CM noise departments. any other reccomendation?

[TomJ]

3 hours ago, itomonje said:

i think you men the intel i350 t2 card. did you compared with rhe jcat also? is As i can read the best choice up till now is the i350 t2 so no other choice to get that so low values in jitter and CM noise departments. any other reccomendation?

You can read my comparison with the JCAT Card here: https://www.open-end-music.com/forum/privatforen/thomas-michael-rudolph-tmr/651284-messungen-von-ethernet-infrastruktur-switches-nur-lesen?p=680071#post680071.

 

I have once compared different cards here at the beginning of this thread: 

 

The Intel i350 cards do so much right that I don't currently look around for alternatives. You can get a used card for round about 50$.

But be careful on ebay as there are counterfeits offered by mainly Chinese sellers.

 

[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.

 

[Jgwtriode]

Thanks Tom:

 

I installed an Intel I 350 V4 T2 today at lunch.  Brand new, never used or opened for 95$.   After 10 hours of burn in this card is clobering my onboard NIC.  Lower noise floor, you can immediately hear, that has even improved a bit with burn.  Stage is about 10% larger at this point in all dimensions, more detail, PRAT and intelligibility.  More relaxed but more dynamic analog presentation.  More emotional and more vibrant.  Focus and layering and tangibility are also about the same level of improvement.  I remember my JCAT XE USB started to sound right at about 150 hours and kept improving a bit after that.  Wondering if this might be the same.  It sound a bit better every couple hours right now.  Everything just breathes and emotes in a far more lifelike fashion.  My ZMF VC headphones sound even more like the best speakers I have owned, but even better in many ways.  This card also really lets my OTL triode Amp open up and I am thoroughly enjoying what this Card is doing.  Want a JCAT XE Network Card at some point still at least I think I do, but I can certainly wait to see just how far this little beauty will take me fully burned.

 

Truly thanks for your comments and measurements,

 

Happy listening,

 

jgwtriode

[TomJ]

Hi Jgwtriode,

 

Glad to hear you're happy with it and can relate to my recommendation!

The Intel i350 is really best bang for the buck. Very good CM noise performance and best jitter value.

 

Best regards,

 

Tom/eric

[Jgwtriode]

Indeed I am....sharing on some of the other audiophile threads I frequent.  It keeps improving.  

 

Happy listening,

 

jgwtriode

[Rovo]

I have installed also an Intel I350T2V2 instead of an Intel Gigabit CT desktop adapter. I clear improvement, especially the lower noise floor. 

 

Thank you Tom/eric for suggesting this network adapter. 

[itomonje]

Hello 👋 I also installed the same Intel i350 t2 on my server and was a great improvement in Quality on all respect's. Thankful  for such a good recommendation!!

[TomJ]

Hi all,

 

Nice to see that my work is not in vain and is also appreciated by you - not like in the German Open End Music Forum, where someone misused my work misunderstood only to improve his own position.

 

Best regards,
Eric

[Rovo]

17 minutes ago, TomJ said:

Hi all,

 

Nice to see that my work is not in vain and is also appreciated by you - not like in the German Open End Music Forum, where someone misused my work misunderstood only to improve his own position.

 

Best regards,
Eric

Eric, 

no your work is definitely appreciated. Hardware wise this was a clear improvement. I do not know if you are familiar with the work of Taiko on their new network interface card? There they also use the software to optimize the performance (for audio) of the NIC. Are you familiar with the software/settings for the NIC and how to optimize?

 

Any progress made in your information exchange with Uptone? I also use an EtherRegen and can appreciate this product and the way they work. 

[TomJ]

On 5/6/2023 at 1:27 PM, Rovo said:

Eric, 

no your work is definitely appreciated. Hardware wise this was a clear improvement. I do not know if you are familiar with the work of Taiko on their new network interface card? There they also use the software to optimize the performance (for audio) of the NIC. Are you familiar with the software/settings for the NIC and how to optimize?

 

Any progress made in your information exchange with Uptone? I also use an EtherRegen and can appreciate this product and the way they work. 

Hello Rovo,

 

sorry for the delay - but I've been quite busy.

 

No, I haven't really looked into the settings of this Taiko devices yet. I've been experimenting with the Auto MDI/X settings as they can mitigate the RJ45 connector pin3/6 problem - but these settings are only available on Linux.

 

I am still in contact with Uptone, but nothing concrete has come up yet regarding the measurement of phase noise.

 

In the meantime I started my own blog: ethernet-sound.com

 

As already reported, I was grossly disappointed in the www.open-end-music.com, since the admin of the TMR sub-forum claims my results and now the whole topic of Ethernet sound for himself, so that there is no longer any perspective for me there. I've been making his forum with my work interesting for over a year and posted all my work there for free and now those  admin guys on this forum don't even acknowledge copyright. Unbelievable. I can only recommend everyone to be careful what you share on this forum.

 

I also wrote an article on my blog about this "Poor Man's Etherregen" OLD6000, which is currently getting a lot of hype in the open-end-music forum after I introduced it there. https://ethernet-sound.com/old6000-gamechanger/

 

All older articles in the blog are still in German - so please use Google translate.

 

Best regards,

 

Eric

 

[Rovo]

On 10/22/2022 at 12:35 PM, TomJ said:

as already announced, I made a comparison measurement between the Intel i350 V2 T2, the Intel i350 V2 T4 and the JCAT NET Card XE in an RockPro64 SBC. The measurements were made according to the state of my research so far. All values can only be considered relative to each other, since many marginal parameters are relevant for the measurements, which can change when measurements are made by other people or in other setups.

 

@TomJ

 

For two and a half years I have been using a JCAT NET XE in my HQPlayer NAA endpoint. The last months I have had problems with random dropouts and the music stopping and occuring more frequently. After a lot of testing it turned out it could be the JCAT NET XE. I decided to replace the JCAT NET XE with an Intel I350-T2V2 for further testing. The sound quality of the Intel network card was not up to the level I was used to with the JCAT NET XE. This was immediately after the replacement and after approximately 20 hours of playing time. But the system was stable again, no random dropouts or music stopping. I left the Intel network card in the system to make sure the system would stay stable. After approximately 50 hours of playing time I noticed that the music was sounding very good. I sat down and started listening to some of my reference numbers and supprise the Intel network card was sounding better than the JCAT NET XE, specifically more control of the bass and in the higher frequencies more natural sounding instruments. 

 

Tom/Eric have you done your tests or repeated your tests after some 50 hours of playing time? Do you know an address where they sell genuine Intel I350-T2V2 network cards? Most of the time they are selling compatible third party cards. 

 

[TomJ]

Hi Robert,

 

please excuse my delay - but you have since found my blog ethernet-sound.com where I have already answered you. - I repeat this again for the fellow readers!

I have a current article regarding the Bob Smith termination published, which may be of interest: 

https://ethernet-sound.com/bob-smith-terminierung-hintergrund-und-modifikation-am-beispiel-tp-tl-sg105e-und-tp-wr902ac/

 

I bought the i350 I tested used, so I assume that I could not have measured any burn in effects.
But nice to hear that the i350 is doing so well for you in the meantime!

I do not have a supply address for the i350. You really have to be careful that you do not buy China Fakes, which happens quite often at eBay.

 

But I am now convinced of the connection of the streamer via a WLAn client and have been operating this now for some time with great success- the TP Link WR902AC.

This has so far the best measurements regarding common mode noise but also very good jitter. I can warmly recommend it with battery PS to everyone.

 

Here are some articles about it from me:

https://ethernet-sound.com/tp-link-tl-wr902ac-im-akku-modus-100ige-isolierung/

https://ethernet-sound.com/modifikationen-am-tp-link-tl-wr902ac-part-i-software/

https://ethernet-sound.com/tp-link-tl-wr902ac-modifikationen-part-ii-hardware/

 

Best Regards,

 

Eric

 

 

[TomJ]

 

———————————

[TomJ]

Hi all,

 

To make it a little clearer, here are a few more explanations:
In the course of my investigations it has become clear that the main cause of sound differences over Ethernet is common mode noise. Reducing this should be the goal if you don't want to manipulate your sound with HF noise.

The least chance of noise coming from the home network into the audio chain is WLAN. Of course, you can also use fibre optics, but this is even more complex.
The task now was to find a WLAN client that itself only produces a small amount of Cm noise - unlike the OLD6000 isolator, for example, which turned out to produce immense common mode interference - which is actually clear if you have a bit of knowledge and look at the fly-back converter on the circuit board.
And this is where the TP Link WR902AC came into play. This has the lowest common mode interference to date - also better than all fibre optic FMCs measured to date.

The device performs incredibly well, especially if you power it isolated from the mains using a PowerBank or, as I explain in my blog, a LiFePo4 battery.

I can only recommend this to everyone. This eliminates all the fiddling around on the Ethernet line and you can finally concentrate on listening to the music.

The last cable to the streamer should then be fully equipped with ferrite cores to reduce the rest of the CM noise - done.

 

Kind regards,

Eric
 

 

 

[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?

[TomJ]

Hello Alex,

 

my WR902AC is now running the OpenWRT operating system and 5GHz is disabled (I will explain this modification in my blog soon). I have not tested the WR802N yet - but the WR902AC is so good that I am not currently looking for an alternative. I had made comparisons with 3 alternative 2.4GHz only WLAN clients (TP Link, D-Link and GL.iNet), which were all worse in measurements, but also in the listening tests.

 

Kind regards,

Eric