Best Ethernet Cards for Streaming

[TomJ]

Hi all,

 

Just published an article about the GigFOIL - but sorry Alex - unfortunately I had to return the Etherregen to the owner and couldn't make a direct comparison. However, I believe that the EtherRegen would perform better because it has consistent ground separation from side A to side B.

 

GigaFOIL V3 vs OLD6000: https://ethernet-sound.com/gigafoil-v3-der-bessere-old6000/

 

Best regards,

 

Eric

[TomJ]

Ok, just posted the article about the OpenWRT upgrade  of the WR902AC.

Its not „one klick do the trick“, but for more advanced users an option.

https://ethernet-sound.com/open-wrt-betriebssystem-auf-dem-tp-link-tl-wr902ac-v3/

 

Best regards,

Eric

[skipspence]

On 10/22/2022 at 10:22 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.

Can you please tell the difference between Intel I350-T2BLK and V2 you have tested if there's any? 

[TomJ]

10 hours ago, skipspence said:

Can you please tell the difference between Intel I350-T2BLK and V2 you have tested if there's any? 

Hi,

the BLK stands for "Bulkware". So as long as the BLK Version is also V2 there should be no difference, however, you get a simpler packaging and no documents such as manual etc.

 

Best regards,

Eric

 

[skipspence]

11 hours ago, TomJ said:

Hi,

the BLK stands for "Bulkware". So as long as the BLK Version is also V2 there should be no difference, however, you get a simpler packaging and no documents such as manual etc.

 

Best regards,

Eric

 

So and what's the difference between v2 and simple blk ie no v2 and what is better to buy? They are different in price.. 

Thank you! 

[TomJ]

47 minutes ago, skipspence said:

So and what's the difference between v2 and simple blk ie no v2 and what is better to buy? They are different in price.. 

Thank you! 

Hi, I would definitely go with V2 - BLK or not, doesn’t matter.

As I only measured V2 I don’t know the difference between V1 or V2.

Best regards,

 Eric 

[TomJ]

Hi all,

 

just published my measurements of 27 lan cables on my blog: https://ethernet-sound.com/draft-messung-von-lan-kabel-part-i/

Please use the language button to enable google translate if necessary.

You are also invited to participate in the small forum area: https://ethernet-sound.com/forums/forum/main/

 

Regards,

 

Eric

[TomJ]

New article with measurements about the Innuos PheonixNet Switch: 

https://ethernet-sound.com/innuos-phoenixnet/

 

[Markus8]

5 minutes ago, TomJ said:

New article with measurements about the Innuos PheonixNet Switch: 

https://ethernet-sound.com/innuos-phoenixnet/

 

Great! I have the new Network Acoustics Tempus switch and could send you for measurements. https://www.networkacoustics.com/shop/tempus-ethernet-switch/

[TomJ]

11 minutes ago, Markus8 said:

Great! I have the new Network Acoustics Tempus switch and could send you for measurements. https://www.networkacoustics.com/shop/tempus-ethernet-switch/

Hi Markus,

 

This would be really great, as I think Network Acoustics are on the right track regarding noise reduction.

Will send you a PM.

 

Best regards,

Eric

[fgribas]

Eric, the work you've been doing is fantastic!

 

I've been reading a lot about i350 T2 V2, modded TP link TL-SG105E and Netgear GS108E.

 

Based on your tests, I already purchased two i350T2V2 cards, one for the server PC and one for the endpoint PC. Connecting them directly without a switch would get the best results?

 

Or in case that a switch would be better, I guess that a modded TP-Link TL-SG105E and Netgear GS108E would be good options. I have a TP-Link TL-SG108E (stock, with all the Bob Smith termination). I guess I would benefit from removing BS termination. But I don't trust my soldering skill that much, so I would prefer to buy a Netgear GS108E 😁

 

Or maybe I will just keep my TL-SG108E and hope it will handle it. I have it powered by LPS, does it help?

 

Oh, I know you prefer the WR902AC option, but maybe another Layer-3 hop would degrade the audio in my setup (using Diretta, very sensible to small variations of latency)

 

And one last question. Supposing I put a modded TP-Link TL-SG105E and Netgear GS108E in place. If my streamer (Eversolo DMP-A6) is connected to this optimized switch, an iFi LAN iSilencer would still make sense? Lots of people recommends putting it on the LAN port of the streamer, but I don't know if it would make sense if using a good (low noise) switch.

[TomJ]

Hi fgribas?,

 

Quote

Oh, I know you prefer the WR902AC option, but maybe another Layer-3 hop would degrade the audio in my setup (using Diretta, very sensible to small variations of latency)

 

 

Before investing in all other options, I would test the Wi-Fi option (WR902AC + PowerBank or battery pack) as this is the best way to isolate your endpoint from previous network noise.
I'm not familiar with Diretta. Does it also use TCP/IP or does the communication run via UDP?

 

Kind regards,

Eric

 

 

[fgribas]

I'm almost sure it uses UDP with IPv6. It's made to run on the same subnet, with higher quantity of smaller packets.

 

As the WR902AC is very cheap, I can try it just for fun. But I don't see myself using stuff with batteries.

 

I'm full in for the i350T2V2 cards for now. In this case, what scenario would be best for me?

[TomJ]

Unfortunately, there is no clear answer to this question.

It actually depends somewhat on how much noise is produced by your server.

I tested the i350 in the last measurements with the RockPro64 board, which is maybe not comparable with your setup.

Since I am now a fan of complete isolation, if WiFi is not an option, I would insert 2 fiber optic converters with a fiber optic route. I would operate the second fiber optic converter with a power bank or with a LHY battery power supply.

 

If this isn't an option either, then I would start with a direct connection first.
The i350 itself has very low common mode noise, so I would give the setup a chance.
I would then continue with the following setup: The Netgear 108 and an isolator in front of the incoming port. You can also use the cheap Delock isolator here. The ifi is definitely not bad either. If you use this, make sure to use a shielded cable, as it sends the interference “back to sender” via the shield.

 

 

Completely equip the cable between last device and endpoint with ferrites, as shown here, in every scenario. And isolate the shielding oft the plug on one side of with scotch tape.

 

Kind regards,

 

Eric

 

[fgribas]

Thanks Eric.

 

My idea is to use i350 cards on server PC (HQPlayer) and Raspberry Pi5 endpoint. Sometimes I feed HQplayer with music from my DMP-A6 streamer -> SPDIF out -> SPDIF/USB converter -> HQPlayer input. In this case, I would put the iFi LAN iSilencer on the ethernet of the DMP-A6 streamer.

 

So this cable on the streamer ethernet port with iFi would be better to be shielded STP. Does the recommendation of scotch tape on one side of the plug + ferrites apply to this cable attached at the streamer device?

 

And for the other cable (i350 on server directly to i350 on endpoint), I believe that everything applies (STP + scotch tape + ferrites)

[TomJ]

Hi,

 

6 hours ago, fgribas said:

So this cable on the streamer ethernet port with iFi would be better to be shielded STP. Does the recommendation of scotch tape on one side of the plug + ferrites apply to this cable attached at the streamer device?

 

I would use a shielded CAT cable. Scotch tape is not helpful here. As far as I can read the layout, the ifi LAN iSilencer has a Bob Smith termination, which is routed to the shielding of the connector on the incoming port. If you now use a cable with shielding (without insulating the plug with scotch tape), then the interference can return to the transmitter - i.e. the streamer. In my opinion, this is the better option, as otherwise they can work via radiation and coupling. Ferrites never do any harm - even with this cable.

 

I recommend a Goobay CAT6a or Patch C7 cable (dont know if this is outside Europe available). These have performed best in my measurements in terms of noise. https://www.reichelt.de/1-0m-patchkabel-cat-7-rohkabel-grau-patch-c7-1-gr-p144528.html or https://www.amazon.de/dp/B00O9RMXM8

Here are the measurements without ferrite: https://ethernet-sound.com/draft-messung-von-lan-kabel-part-i/#PATCH-C7_1_GE

And here with ferrite: https://ethernet-sound.com/draft-messung-von-lan-kabel-part-i/#PATCH-C7_1_GE_Ferritiert

 

6 hours ago, fgribas said:

And for the other cable (i350 on server directly to i350 on endpoint), I believe that everything applies (STP + scotch tape + ferrites)

I would also go with the Goobay CAT6a or Patch C7 cable. Here you should insulate the plug on the server side with tape and, as always, use as much ferrite as possible.

 

Kind regards,

Eric

 

 

[fgribas]

Thank you, priceless information! 😃

[Zauurx]

@TomJ Thank you for all the information.
I've been using this Monoprice with pleasure for 2 years and now with an Ifi Silencer lan on the Gustard A26 DAC side (or Diretta I2s).

https://www.monoprice.com/product?p_id=34267
Unshielded and therefore galvanically insulated
I don't use a switch but am connected directly to my server in IPV6/UDP from my JCAT Net Femto (no need for crossover cables like years ago, the cards handle it).
I ordered this one for testing :

https://www.amazon.fr/dp/B07GVRY6LT?ref=ppx_yo2ov_dt_b_product_details&th=1
I'm wondering about the galvanic break with a shielded CAT8.
It seems difficult to insert the plug with a layer of tape.
Do you have a photo ? It's often easier to remove the metal strapping on one side.

Otherwise, I'd add a 1ft Monoprice with a DX-iso.

[TomJ]

Hi all,

 

If you look at the circuit board of the ifi LAN iSilencer, you can see that the centre taps of the transformer on the input side are routed via capacitors and to the "housing ground" of the RJ45 socket (sorry, its not a Bob Smith termination) . A connected shield can then lead the noise back to the source. In addition, a galvanic isolation is realised here so that the cable does not have to be insolated. 

 

Hope this helps.

 

Regards,

Eric

[Zauurx]

Thanks, 😉

 

Ifi audio marketing information are interesting, but.. no mesure of this galvanic isolation.
... and the diagram suggests putting it on the server or receiver side.

Quote

The LAN iSilencer boasts premium galvanic isolation technology, where
electrical circuits are separated to eliminate stray currents. Audio signals can
pass between galvanically isolated circuits to block i.e. differences in ground
potential or currents induced by AC power.

 

And the informations of DX-engineering (radio specialist) are also clear !!

Quote

RFI can creep in from switching power supplies, wall power adapters, dimmer switches, touch lamps, appliances and just about anything near Ethernet cables. The RFI generated internally by the device can be radiated over the Ethernet cable. Once RFI is in the Ethernet cable it acts as an antenna and radiates RF signals that can be picked up by your receiver. Proximity of transmitting antennas near internal wiring can cause interference that gets into Ethernet cables. Even if the RJ-45 cables are “shielded”, and especially if the shield is terminated at both ends of the cable, then the shield becomes just another conductor which can receive unwanted signals and reradiate them just as any other wire in the cable might do. ISO-PLUS RF Filters can reduce or eliminate this trouble.

Always use RF clean power supplies. Linear power supplies are usually better than switching power supplies. Some switching power supplies that are specifically designed for use with Amateur radios usually have extra internal filtering to avoid causing RF interference in the amateur radio bands. However, most wall adapters do not have proper filtering and can cause significant interference problems.

Be aware that proper equipment installations will contribute to the reduction or elimination of RF and noise interference. Proper power and RF wiring and grounding, quality coaxial cable and connector installation, and tight positive mechanical connections are all very important, as well as other good RF practices. DX Engineering offers many useful accessories for the reduction of RFI in the station including DX Engineering Ferrite and the Radio RF Ground Plane Kits

 

[One and a half]

30 minutes ago, Zauurx said:

Thanks, 😉

 

Ifi audio marketing information are interesting, but.. no mesure of this galvanic isolation.
... and the diagram suggests putting it on the server or receiver side.

 

And the informations of DX-engineering (radio specialist) are also clear !!

 

The galvanic isolation is just a few volts between input and output, so for industrial purposes this device is useless unless like Intona is 1000V or more. But they don't say, so assume zero or not going to make any difference, just fluff words.

 

For audio it's more useless as it leaks noise from input to output. 

Hah, you could use DX filter either side to isolate,  but the DX maintains the shield. So CATx UTP is still worthwhile to use for audio. 

 

I found if a shielded cable is in a path in front of the Lumin U1 the sound is gritty treble and bass drops off. The music is dull. So all cables in front are plain CAT5e unshielded of at least 15m. The rest of garbage is taken care by EtherRegen and DX filters,  that works.

 

So far have been tempted on boutique cables, but 99% of what I've found are shielded CATx which just doesn't work.

[TomJ]

Hi all,

 

all these isolators and filters are using standard transformers.

The minimum isolation for these transformers is 1500V (voltage pulse).

 

The DX Filter is just a PCB with two RJ45 jacks with integrated Transformers and CMCs.

At the beginning of my journey I built something similar:

 

You can by just 2 of these jacks at mouser and connect the pins - then you have the DX Filter : https://www.digikey.com/en/products/detail/würth-elektronik/74990101212/6598155

 

 

And as you can read in my article, you shouldn't think that this will result in fewer noise They can also increase - in my opinion due to mode conversion caused by poor design.

https://ethernet-sound.com/lan-isolatoren-sotm-iso-cat7-u-delock-isolator/

To believe that such a filter / isolator does the same job as an EtherRegen or a fibre optic cable is not a good idea.

 

Kind regards,

Eric

 

[Zauurx]

2 hours ago, One and a half said:

...

For audio it's more useless as it leaks noise from input to output. 

Hah, you could use DX filter either side to isolate,  but the DX maintains the shield. So CATx UTP is still worthwhile to use for audio. 

...

 

That's why I keep coming back to the Monoprice Microrun. 🙃

[TomJ]

On 11/20/2023 at 7:03 PM, Markus8 said:

Great! I have the new Network Acoustics Tempus switch and could send you for measurements. https://www.networkacoustics.com/shop/tempus-ethernet-switch/

Hi Markus8,

 

Is your offer regarding the Tempus switch still valid? I would be really interested in measuring it.

 

Best regards,

Eric

[One and a half]

On 11/22/2023 at 7:26 PM, TomJ said:

Hi all,

 

If you look at the circuit board of the ifi LAN iSilencer, you can see that the centre taps of the transformer on the input side are routed via capacitors and to the "housing ground" of the RJ45 socket (sorry, its not a Bob Smith termination) . A connected shield can then lead the noise back to the source. In addition, a galvanic isolation is realised here so that the cable does not have to be insolated. 

 

Hope this helps.

 

Regards,

Eric

If the noise travels back to the source, won't that stop the noise travelling to the destination? This is how standard EMC/RFI filters work. I would agree that having the housing bonded even with capacitor from input to output is not a great idea. In any case, conducted frequencies below 30MHz are affected. (added EMC guide from EPCOS)

Or, is there a chance of multiplying reflections... would have thought a single frequency could reflect, but multiple varying?

EMC Filter Application Guide EPCOS.pdf