Jump to content
IGNORED

Best Ethernet Cards for Streaming


Recommended Posts

3 minutes ago, Cogito said:

The noise in the network is what causes problem to the audio. Ethernet cables act line antennas. They pickup all kinds of noise which enters the computer and eventually the DAC thru’ USB interface. Noise filtration/elimination is the key to good audio.

 

That is common mode noise that gets cancelled in the reception of differential signal. In addition, all the ethernet signals are transformer isolated and you don't have galvanic connection unless you spoil it with shielded ethernet cables...

 

Of course you can use optical ethernet where fibers cannot pick up any such noise on the way in first place, nor have any galvanic connection.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
28 minutes ago, Miska said:

 

Of course you can use optical ethernet where fibers cannot pick up any such noise on the way in first place

 


That’s the most effective way to get rid noise on Ethernet. You can setup a Ethernet to optical converter, fiber cables and fiber optic NIC at fraction of the cost of audiophile NICs.

Link to comment

It would also be really cool to see measurements of the fiber converters or SFPs. Sometime the cure to an ailment also has side effects. I’ve always wondered if putting an SFP slot and module inside a HiFi component also causes noise issues itself, even though the fiber has isolated anything upstream. 
 

All interesting really stuff. 

Founder of Audiophile Style | My Audio Systems AudiophileStyleStickerWhite2.0.png AudiophileStyleStickerWhite7.1.4.png

Link to comment
1 minute ago, The Computer Audiophile said:

It would also be really cool to see measurements of the fiber converters or SFPs. Sometime the cure to an ailment also has side effects. I’ve always wondered if putting an SFP slot and module inside a HiFi component also causes noise issues itself, even though the fiber has isolated anything upstream. 
 

All interesting really stuff. 


Holo Audio is already doing Electrical to Optical Conversion and back to Electrical in its USB input to get rid of noise (galvanic isolation).  

Link to comment
10 minutes ago, Cogito said:


Holo Audio is already doing Electrical to Optical Conversion and back to Electrical in its USB input to get rid of noise (galvanic isolation).  

Yes, Ayre has done that for a long time. Some of the optical conversion tech used in this capacity is less than ideal. The cure can be worse than the illness. 

Founder of Audiophile Style | My Audio Systems AudiophileStyleStickerWhite2.0.png AudiophileStyleStickerWhite7.1.4.png

Link to comment
2 hours ago, Cogito said:


That’s the most effective way to get rid noise on Ethernet. You can setup a Ethernet to optical converter, fiber cables and fiber optic NIC at fraction of the cost of audiophile NICs.

I disagree here, and explain.

 

To remain faithful to the transmission of data on the Ethernet source  side, the media converter should convert signal + unwanted (noise) as a correct representation, bits is bits......... right? @plissken goes on about this forever.

 

So all the fibre cable does is to transmit that same signal + unwanted through the pipe and it does this without adding extra interference from transformers, spike ridden cables adjacent to the run of the fibre cable, and that's all cool. 

At the load end, the media converter faithfully decodes the data as well as unwanted, exactly as from the source. So if there's garbage already on the electrical Ethernet, there will be at the destination Ethernet conversion.

 

The fibre cable does *not* remove noise per se, but it will certainly not *add* any either.

 

Personally, I've been reluctant to use fibre for audio, since I don't possess any Ethernet fibre renderers, like the Lumin X1. Media converters add hash and garbage at the electrical sides, read also unwanted crud from PSU, as they are another unwanted in the chain. Ideally, a fibre NIC at the source and the load (like a renderer)  already has a fibre receiver, then the need of media converters is eliminated and a potential source of garbage is also gone. There's also some conjecture as to what type of NIC and SFP is suitable, go ask @wklie, some work and some don't.

 

 

AS Profile Equipment List        Say NO to MQA

Link to comment
19 minutes ago, One and a half said:

I disagree here, and explain.

 

To remain faithful to the transmission of data on the Ethernet source  side, the media converter should convert signal + unwanted (noise) as a correct representation, bits is bits......... right? @plissken goes on about this forever.

 

So all the fibre cable does is to transmit that same signal + unwanted through the pipe and it does this without adding extra interference from transformers, spike ridden cables adjacent to the run of the fibre cable, and that's all cool. 

At the load end, the media converter faithfully decodes the data as well as unwanted, exactly as from the source. So if there's garbage already on the electrical Ethernet, there will be at the destination Ethernet conversion.

 

The fibre cable does *not* remove noise per se, but it will certainly not *add* any either.

 

Personally, I've been reluctant to use fibre for audio, since I don't possess any Ethernet fibre renderers, like the Lumin X1. Media converters add hash and garbage at the electrical sides, read also unwanted crud from PSU, as they are another unwanted in the chain. Ideally, a fibre NIC at the source and the load (like a renderer)  already has a fibre receiver, then the need of media converters is eliminated and a potential source of garbage is also gone. There's also some conjecture as to what type of NIC and SFP is suitable, go ask @wklie, some work and some don't.

 

 

 

What you're describing seems analog. Is the noise from the origin transmitted through the conversion, or only an optical, then electrical, representation of the bits (plus any noise inherent in the destination conversion)?

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
20 minutes ago, One and a half said:

I disagree here, and explain.

 

To remain faithful to the transmission of data on the Ethernet source  side, the media converter should convert signal + unwanted (noise) as a correct representation, bits is bits......... right? @plissken goes on about this forever.

 

So all the fibre cable does is to transmit that same signal + unwanted through the pipe and it does this without adding extra interference from transformers, spike ridden cables adjacent to the run of the fibre cable, and that's all cool. 

At the load end, the media converter faithfully decodes the data as well as unwanted, exactly as from the source. So if there's garbage already on the electrical Ethernet, there will be at the destination Ethernet conversion.

 

The fibre cable does *not* remove noise per se, but it will certainly not *add* any either.

 

 

 


‘Only the detected bits are converted to light and transmitted over the fiber optic cable. Electrical noise is so low,  it is not detected by the converter and hence not transmitted over the fiber.

Link to comment
2 minutes ago, Cogito said:


‘Only the detected bits are converted to light and transmitted over the fiber optic cable. Electrical noise is so low,  it is not detected by the converter and hence not transmitted over the fiber.

That's the problem, how low is low electrical noise to be ignored and not be transmitted?  There's also offset in PHY and deterioration of LED receivers to worry about over time.. another issue.

 

8 minutes ago, Jud said:

 

What you're describing seems analog. Is the noise from the origin transmitted through the conversion, or only an optical, then electrical, representation of the bits (plus any noise inherent in the destination conversion)?

Electrical signals contain wanted and unwanted (parasites) along digital transmissions, the DXE filter provide some attenuation and avoid adding additional noise for Ethernet cables. 

AS Profile Equipment List        Say NO to MQA

Link to comment
4 minutes ago, One and a half said:

That's the problem, how low is low electrical noise to be ignored and not be transmitted?

 

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.

 

I think however that you're rightly concerned about PSU noise getting into the system, and Chris is also right in being concerned about the noise of the destination converter itself.

 

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

In looking at fiber transmissions, losses are due to the distortion apart from the photons vanishing, s'pose, but there's a device that perks up the losses and regnerates the signal. Sounds familiar to audio... the application is for long distances, but how much distortion does the cable exert for a distance of less than 50m?

 

FIBER-BASED 2R REGENERATORS3R optical regeneration: An all-optical solution with BER improvement

 

 

AS Profile Equipment List        Say NO to MQA

Link to comment
5 hours ago, The Computer Audiophile said:

It would also be really cool to see measurements of the fiber converters or SFPs. Sometime the cure to an ailment also has side effects. I’ve always wondered if putting an SFP slot and module inside a HiFi component also causes noise issues itself, even though the fiber has isolated anything upstream. 
 

All interesting really stuff. 

You are right. As SFP converters have increased jitter, but also their power supplies are hard to optimize because the laser needs large currents to switch very fast, which is an EMI problem.

 

As described before, Ethernet is a differential signal and therefore quite immune to ambient EMI.

If EMI were a problem, then I'm surprised that different cables with similar construction sound different.

 

But the signal itself can be considered noise at 31.25 MHz at 100base Tx.  This penetrates unhindered via the transformer into the end device. Therefore, I am currently concentrating on the signal.

Link to comment
6 hours ago, Cogito said:

You can setup a Ethernet to optical converter, fiber cables and fiber optic NIC at fraction of the cost of audiophile NICs.

 

I don't use any audiophile networking gear. I get great results from regular ones, also from regular copper gear and CAT6 UTP cabling.

 

6 hours ago, The Computer Audiophile said:

I’ve always wondered if putting an SFP slot and module inside a HiFi component also causes noise issues itself, even though the fiber has isolated anything upstream.

 

I think SFP slot is good option because it keeps all the electronics related to physical link in a separate shielded metal cage. It also gives you flexibility to choose between optical and copper connections.

 

58 minutes ago, TomJ said:

As SFP converters have increased jitter, but also their power supplies are hard to optimize because the laser needs large currents to switch very fast, which is an EMI problem.

 

Have you measured this increased jitter? I don't think there's such. Also ethernet at default power has large currents, that's why I use 802.3az links where equipment measures cable length and uses reduced power for shorter links and also automatically goes into inactive idle state.

 

1 hour ago, TomJ said:

But the signal itself can be considered noise at 31.25 MHz at 100base Tx.

 

But who would use 100baseTX? It's a bit too old spec typically lacking modern features. And not enough speed. 1000baseTX is pretty much requirement these days.

 

 

What I've measured DAC outputs so far. I've mostly seen ground link, PSU and USB related problems. With USB related problems I mean 8 kHz packet noise leaking through which is problem of the USB receiver in DAC.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
1 hour ago, Miska said:

Have you measured this increased jitter? I don't think there's such. Also ethernet at default power has large currents, that's why I use 802.3az links where equipment measures cable length and uses reduced power for shorter links and also automatically goes into inactive idle state.

 

Not myself. This was the info I get from an embedded ethernet engineer. But if you imagine that the signal has to be converted 2 x more, this statement does not seem outrageous to me. Regardless of the power consumption, it is an additional component that can produce noise.

 

1 hour ago, Miska said:

But who would use 100baseTX? It's a bit too old spec typically lacking modern features. And not enough speed. 1000baseTX is pretty much requirement these days.

 

I don't know what features and what speed I should miss with 100base tx for music streaming. Less is more in my opinion - Fewer crosstalk, through less signal wire-pairs for example.

Link to comment
13 minutes ago, Miska said:

There's no "2x more" or power consumption. You need same amount of components with or without SFP. SFP just moves PHY to a module.

I did not write that there is a doubling of the power consumption. Only that an additional active component with SGMII connection is added, which is in addition to the controller, which has the phy on board for ethernet. In addition, there is a media conversion from power to optics that does not occur with Ethernet.

 

22 minutes ago, Miska said:

802.3az for example (cable length detection and idle sleep). And with 100baseTX you have worse transfer/sleep ratio, so every packet transfer takes 10x longer and produces 10x more noise. WIth 10 Gbps you have only tiny transfer blip every now and then, 100x less transfer noise.

 

In many cases 100baseTX begins to fail at DSD1024 or 1.5M PCM. Not to even mention 8 channels of DSD256 or 384/32 PCM.

 

IEEE 802.3az is not a feature of 1000base TX but is also specified for 100base tx.

With regard to your calculation for the amount of noise, this is probably a bit simplistic. How do you measure the amount of noise? Status changes of the signal over time?
Even if no stream is transmitted, the line is not silent.

 

Link to comment
12 minutes ago, TomJ said:

In addition, there is a media conversion from power to optics that does not occur with Ethernet.

 

That's of course the optical isolation we are looking for? This is very low power, compared to the power blasted on magnetics for full cable length Ethernet (which you have unless you have cable length detection through 802.3az).

 

12 minutes ago, TomJ said:

IEEE 802.3az is not a feature of 1000base TX but is also specified for 100base tx.

 

It practically is. If you look at networking gear specs, 802.3az is practically never included in 100baseTX hardware, and falls off if you drop 1000baseTX controller to 100baseTX backwards-compliance.

 

12 minutes ago, TomJ said:

With regard to your calculation for the amount of noise, this is probably a bit simplistic. How do you measure the amount of noise? Status changes of the signal over time?

 

Amount of low level network activity as function of time. When people talk about audio "streaming" they think there is some constant traffic going on, which is not the case. All the network traffic is packet based. Just like USB too.

 

Higher the network speed, less time it takes to transfer a packet. Which improves active/idle ratio.

 

12 minutes ago, TomJ said:

Even if no stream is transmitted, the line is not silent.

 

This is specifically feature of 802.3az that the line can go silent.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
44 minutes ago, The Computer Audiophile said:

Comments like this don’t help anyone. 


True. 
We all have our subjective biases and limitations to objective knowledge. We struggle with those all the time. We constantly are seeking more information, more knowledge, to enhance our experience of this hobby. Some of the time we come across good info, most of the useless info. 
But when someone is posting blatantly false information, it needs to be pointed out, IMO. 
If I post some wrong info and you point it out, I will take a step back and try to digest it and correct myself. If OTOH, I double down and make more wrong assertions, I would hope someone points it out. That’s exactly what happened here.

Link to comment
45 minutes ago, Cogito said:


True. 
We all have our subjective biases and limitations to objective knowledge. We struggle with those all the time. We constantly are seeking more information, more knowledge, to enhance our experience of this hobby. Some of the time we come across good info, most of the useless info. 
But when someone is posting blatantly false information, it needs to be pointed out, IMO. 
If I post some wrong info and you point it out, I will take a step back and try to digest it and correct myself. If OTOH, I double down and make more wrong assertions, I would hope someone points it out. That’s exactly what happened here.

 

Posting correct info is great. Telling someone to use Google isn't. 

Founder of Audiophile Style | My Audio Systems AudiophileStyleStickerWhite2.0.png AudiophileStyleStickerWhite7.1.4.png

Link to comment
9 hours ago, One and a half said:

That's the problem, how low is low electrical noise to be ignored and not be transmitted?

 

Electrical noise is very low level. When converting electrical signals to optical, we look for signals with certain properties of our interest, like voltage and frequency. Think about it this way, if I am interested in catching butterflies, the net would have relatively large gaps between the fibers. Those gaps are designed to trap insects of certain size. Trillions of airborne insects will pass thru it and we would not even realize. 

 

Its a well established technology. 

 

9 hours ago, One and a half said:

There's also offset in PHY and deterioration of LED receivers to worry about over time.. another issue.

 

Dont know what that means and how it is related to electrical to optical conversion.

 

8 hours ago, One and a half said:

In looking at fiber transmissions, losses are due to the distortion apart from the photons vanishing, s'pose, but there's a device that perks up the losses and regnerates the signal. Sounds familiar to audio... the application is for long distances, but how much distortion does the cable exert for a distance of less than 50m?

 

There are different uses.  Virtually in every datacenter, all communications occur over fiber optic cables, often the distances are within a few feet.  

 

7 hours ago, TomJ said:

You are right. As SFP converters have increased jitter, but also their power supplies are hard to optimize because the laser needs large currents to switch very fast, which is an EMI problem

 

Jitter (timing errors) are part and parcel of ALL digital telecommunications.  All telecommunication protocols like Internet Protocal (IP) have methods built in to hand timing errors, data corruption, missing data etc. etc. in real time.   Jitter is NOT an issue that one worries about in telecommunications, in audio world yes, jitter is a big deal.

Link to comment
5 hours ago, TomJ said:

 

Not myself. This was the info I get from an embedded ethernet engineer. But if you imagine that the signal has to be converted 2 x more, this statement does not seem outrageous to me. Regardless of the power consumption, it is an additional component that can produce noise.

 

You are specifically talking about streaming. If you've 15 hops, as indicated by a traceroute.... Using your argument they are all additional components that can add noise. The link at the destination counts.

 

5 hours ago, TomJ said:

I don't know what features and what speed I should miss with 100base tx for music streaming. Less is more in my opinion - Fewer crosstalk, through less signal wire-pairs for example.

 

802.3az for starts. 802.3af/at/bt for seconds. 1000Base T is still CMNR. Math is math. You have 10X more possible wire time with 100 vs 1000.

 

Would you rather take an ambulance at 100, 1000, 10000 KPH to get you to the emergency room?

Link to comment
14 hours ago, The Computer Audiophile said:

It would also be really cool to see measurements of the fiber converters or SFPs. Sometime the cure to an ailment also has side effects. I’ve always wondered if putting an SFP slot and module inside a HiFi component also causes noise issues itself, even though the fiber has isolated anything upstream. 
 

All interesting really stuff. 


i get the feeling you are conflating two different technologies. 
 

I was referring to the network side of the PC based music server.  Let me explain.  In our homes we have copper based and wireless networks. Wireless network is not recommended for streaming as it has high levels of radio freq noise. Even in the wired network there is lot of noise present. Since the goal is to reduce the noise getting into the DAC, it needs to be eliminated before it enters the music server PC.  
 

The solution I am talking about works like this. Just before the music server convert the electrical signals to optical and pass them on to the PC. The PC will have a Optical NIC where the optical signals are converted back to electrical. The SFP are installed on either end of the short optical cable and the data is still in the IP domain where all the error detection and error correction protocols are embedded, I am not sure what parameter you would measure which are relevant to audio. If the parts are Of decent quality, there should be any issues.


I am not aware of any use of SFP modules between music server and DAC.

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...