The EtherREGEN thread for various network, cable, power experiences and experiments

[R1200CL]

42 minutes ago, kostas6a3 said:

The power supply of HDPLEX 300W is expected, which will power the two EtherREGEN, the Melco N100 and the NUCK for the final hearings.

This isn’t a good solution. Use a separate PS for second EtherRegen. 
 

42 minutes ago, kostas6a3 said:

So far the two EtherREGENs in the series have not performed as expected.

Are you doing A>B>A>B ?

What ethernet cable ?

 

What connect to B side second EtherRegen ?

 

Are you using both A sides for other equipment ? If yes, why ?

If no, are you using the first or second EtherRegen to connect other equipment ?

[kostas6a3]

8 hours ago, R1200CL said:

This isn’t a good solution. Use a separate PS for second EtherRegen. 

The initial thought is the HDplex, since the second EtherREGEN remains in the device, to power the 4 devices (Melco. NUC, both EtherREGEN) after the manufacturer reports:

"True Separate Rail Design. No shared Ground. Guarantee No Cross RFI/EMI Interference. This is a must for multi-rail Linear Power Supply. All output can work at the same time with no loss of performance and no cross-talk with each other".

The final form will be made after testing.

[kostas6a3]

8 hours ago, R1200CL said:

Are you doing A>B>A>B ?

What ethernet cable ?

What connect to B side second EtherRegen ?

Are you using both A sides for other equipment ? If yes, why ?

If no, are you using the first or second EtherRegen to connect other equipment ?

The final hearings - comparisons were made as follows:

The router, iMac and NAS are connected to the first Ubiquiti Edgerouter x switch.

Then with Supra CAT8 on the first EtherREGEN, A side as well as the NUC with ethernet cable Neotech NEET 1008 silver.

On the B side of it with Supra CAT8 on the A side of the second EtherREGEN and on the B side of it with Audioquest cat 700 Carbon in Melco.

[dbastin]

17 hours ago, kostas6a3 said:

So far the two EtherREGENs in the series have not performed as expected.

The sound became softer, lost in sharpness and the bass looser. Everything is powered by the power supplies that come with them. The power supply of HDPLEX 300W is expected, which will power the two EtherREGEN, the Melco N100 and the NUCK for the final hearings.

Some observatiins to ponder ...

 

Sounds to me like the things you are describing as missing could well be the adverse impacts of noise from the network.

 

Noise can impose a hardness and unpleasant edge, and that may be preferred in the context of some systems, or people become accustomed to it as being normal.  I am curious about the loose bass though.

 

In my experience, bass changes a lot in different ways as the network is improved and generally becomes more well defined, accurate impulse.  Bass parts in music are far more in sync with the rhythm.  In musician terms, the bass playing sounds 'tighter' with the band.

 

But there is also a more extended/wider frequency range, and stronger bass response, as if the bass has been compressed (whereas with the affects of noise the bass is more spread out over longer duration).

 

It is possible the 2 ERs are now highlighting problems elsewhere in your network, room or system.

 

Consider that the new sound could now be exciting your room differently and creating a new perception??

 

Also, take care with how ethernet shields are connected.

 

And, be careful using one power supply for 2 ERs, and server.  It may be possible for noise to transmit from one device to another via the power supply, which could negate the MOAT isolation.  It probably depends on the design of the power supply.

 

Finally, the N100 built in switch might not be providing you maximum benefit if the N100 is your endpoint.  Instead of using NUC as server, you might get better results using N100 as server and having a DAC with an integrated  player.

 

Router > ER1 > N100 > ER2 > endpoint DAC

[kostas6a3]

29 minutes ago, dbastin said:

Some observatiins to ponder ...

 

 

Maybe I did not put it correctly. The sound was lost in clarity.

The two EtherREGENs currently use their own pulse power supply and do not have a common power supply.

The Melco N100 is a music server and end point and not a network switch so I do not see how to EtherREGEN could be connected to the DAC.

 

[barrows]

Other possibilities to consider here:  Note, that the etherRegen, while excellent, does not provide "absolute" isolation from A to B.  This is the same with all isolation devices I am aware of: while the moat mostly works, and the isolator(s) them selves mostly work, there will always be some coupling of noise across the isolation (the same is true of isolated USB receivers inside DACs as well).  Considering that the isolation is not perfect, then also consider this possibility when thinking that using more switches in series might offer an improvement (lower noise transmission): every switch you add, also adds its own high frequency noise component.  This added noise is unavoidable, as it is inherent to any switch-these are high spreed processing devices and sources of noise.  Then the question becomes: does adding another switch produce a net reduction in noise on the Network which reaches the endpoint, or a net increase?  It is very possible that the additional switch produces a net increase in noise, so that is something to consider as a possibility.  Without the ability to do spectrum analysis on the Network, one is a bit lost here.  My personal approach would be to keep it simple.

I also wonder a bit about how "isolated" the HDPlex power supplies outputs are from each other?  For example, is there a separate transformer for each output?  Or, is there just a separate winding on a single transformer?  Separate windings creating different outputs will couple some noise, as they are in close proximity to each other.  Even separate transformers, if in close quarters inside the same chassis, will couple some noise to each other.  Again, without doing some complex measurements it is difficult to know how much noise coupling there is.

If my goal is to have isolation from one area to another, I am always going to have completely separate power supplies-this means separate chassis, separate power cables, and hopefully the supplies themselves separated by some distance.  In my system, I prefer to have the audio system as isolated from with only an optical Ethernet fiber running to the system and all the computer gears are in another part of the home, powered from other AC line feeds.    

 

[audiobomber]

I have an EtherREGEN in my main audio system, feeding a Playpoint renderer from the B side, via Furutech LAN8 cable and LAN Isolator. I was using a Netgear GS-105 with LPS upstream, near my network gear, but I acquired more gear and ran out of ports. Adding a second ER was not feasible, so I upgraded the GS-105 to a Silent Angel N8. I was dissatisfied with the sound until I replaced the N8 Meanwell SMPS with a grounded iPower2. Very happy now, more clarity and detail, improved bass power. 

[dbastin]

52 minutes ago, kostas6a3 said:

Maybe I did not put it correctly. The sound was lost in clarity.

The two EtherREGENs currently use their own pulse power supply and do not have a common power supply.

The Melco N100 is a music server and end point and not a network switch so I do not see how to EtherREGEN could be connected to the DAC.

 

Hmm, I suspect something is not quite correct, perhaps a connection.

 

N100 has at least a network bridge and possibly a full switch.  It seems to have 2 PHY chips.  The 'player' port can be used to connect to a DAC with an integrated player/streamer.(eg. LUMIN D2, Aurender A10, lots more).

 

Off loading player functions from N100 will reduce load on the cpu, in short most likely gain improvement in SQ.  Having a player integrated with the DAC reduces complications if converting formats, cables etc.  Again, probably benefits SQ.

 

[R1200CL]

14 hours ago, kostas6a3 said:

Audioquest cat 700 Carbon

Can you confirm there is absolutely no connectivity between the plugs ? (Outer metal). 

If there is, please try another cable. 

[kostas6a3]

35 minutes ago, R1200CL said:

Can you confirm there is absolutely no connectivity between the plugs ? (Outer metal). 

If there is, please try another cable. 

Everything is connected to the plug with the shielding cable as suggested by the manufacturer.

Both Supra and Audioquest and Neotech. They are factory made.

[kostas6a3]

3 hours ago, dbastin said:

Off loading player functions from N100 will reduce load on the cpu, in short most likely gain improvement in SQ.  Having a player integrated with the DAC reduces complications if converting formats, cables etc.  Again, probably benefits SQ.

 

Unfortunately I do not have a DAC with such input and I do not plan to get one in the near future.

[kostas6a3]

4 hours ago, barrows said:

This added noise is unavoidable, as it is inherent to any switch-these are high spreed processing devices and sources of noise.  Then the question becomes: does adding another switch produce a net reduction in noise on the Network which reaches the endpoint, or a net increase?  It is very possible that the additional switch produces a net increase in noise, so that is something to consider as a possibility.  Without the ability to do spectrum analysis on the Network, one is a bit lost here. 

 

I think you are absolutely right.

I'm waiting for the HDplex next week for the final auditions as well as some ethernet cables.

There is currently only one EtherREGEN left.

[R1200CL]

1 hour ago, kostas6a3 said:

Everything is connected to the plug with the shielding cable as suggested by the manufacturer.

Both Supra and Audioquest and Neotech. They are factory made.

I think you didn’t understand my question. Do you have an ohm meter ? You need one in order verify if the shield is connected to the Telegartner metal plug. 
 

If your cables has shield connected as the standards require, this may cause a problem. 
Please test with simple standard RJ45 plastic plug cables. 
 

You can open the plug and isolate the shield. 
 

 

[kostas6a3]

47 minutes ago, R1200CL said:

I think you didn’t understand my question. Do you have an ohm meter ? You need one in order verify if the shield is connected to the Telegartner metal plug. 
 

If your cables has shield connected as the standards require, this may cause a problem. 
Please test with simple standard RJ45 plastic plug cables. 
 

 

All cables are connected to the shield with the metal plugs.
They have been measured.
Maybe try it with simple cables.

[R1200CL]

39 minutes ago, kostas6a3 said:

All cables are connected to the shield with the metal plugs.

This is the recipe to create earth loops. 
 

41 minutes ago, kostas6a3 said:

Maybe try it with simple cables.

Looking forward to learn if you notice any differences. 

[Superdad]

1 hour ago, kostas6a3 said:

All cables are connected to the shield with the metal plugs.
They have been measured.

Multiple people have verified that Audioquest connects the cable shield drain wire to the metal end shell at only one end. 
Other brands will vary and can easily be tested with a meter. 

[dbastin]

I am afraid it is trial and error.   Perhaps best to use a decent UTP to replace every one of your cables one at a time to find out which connection is causing the problem.  Don't worry about the SQ until you work out which cable connection is causing the problem, and then work out the best sounding cable for that connection that does not cause the problem.

 

In one situation in my network, I have a relatively low cost commercial grade Cat 6a UTP performing better than a $800 audiophile Cat 8 cable that connects shields at both ends with metal connectors.  But the Cat 6a is trumped by audiophile cables elsewhere downstream.

 

This is because:

• there does not appear to be a standard way to connect shields for ethernet cables used for audio and
• there may not be a standard way audio gear deals with the ground of their ethernet ports, and metal connectors. 

In my liaison with the makers of the audiophile ethernet cables I have, they use what ever method achieves the best sound from their cables, and they use different tech and methods in the cable construction. 

[barrows]

25 minutes ago, dbastin said:

there may not be a standard way audio gear deals with the ground of their ethernet ports, and metal connectors. 

I am not aware of any product which does not connect the shell of the Ethernet port to the ground of the circuit.  This is standard practice, and I would be surprised to find any product which does not do this.  Therefore, if one has an Ethernet cable with a shield terminated at both ends to metal bodied connectors, there is the potential for a ground loop and ground current induced noise.  I would always advise against using Ethernet cables with shields tied at both ends to metal bodied connectors.  Note that just because an audiophile Ethernet cable has metal bodied connectors it may not tie the shield at both ends, so check with a multi meter.  For example, the (excellent and recommended) Sablon Ethernet cable does not tie the shields to the connectors at both ends of the cable.

[Superdad]

1 hour ago, dbastin said:

 Perhaps best to use a decent UTP to replace every one of your cables one at a time to find out which connection is causing the problem. 

I don't think @kostas6a3 has reported any functional problem.  I just reread all his posts going back a few pages and it seems his is just not liking the sound he is getting with with his multi-switch setup and the Melco.  Plus he has a whole mix of cables, power supplies, and a DIY USB cable.  So it would seem that he just needs to experiment (and maybe simplify) to find what suits his taste. 

[Johnnydev]

17 hours ago, Superdad said:

Multiple people have verified that Audioquest connects the cable shield drain wire to the metal end shell at only one end. 
Other brands will vary and can easily be tested with a meter. 

Yes, audioquest connects the cable shield only on the designation site

if you make a choice for supra cat 8 DIY you can make it easily on which end you want connect or not.

[R1200CL]

13 hours ago, barrows said:

I am not aware of any product which does not connect the shell of the Ethernet port to the ground of the circuit.

Audioquest Vodka has been verified not to. Can’t answer for other AudioQuest Ethernet cable’s. 
 

All Gent Audio ethernet cables is built this way. No connection to shield. In addition he applies a Faradays Cage to the cable. The idea of using metal plugs is better EMI/RFI protection. The manufacturer Telegartner has more information of the claimed benefits. 
 

[Superdad]

@R1200CL:

@barrows was referring to Ethernet equipment—not to cables—as most always connecting their jack shields to the ground-plane of their circuit board. 
However, not all RJ45 jacks (with built-in magnetics or not) have jack shields.  Most do but such is not required.

[dbastin]

7 hours ago, Superdad said:

@R1200CL:

@barrows was referring to Ethernet equipment—not to cables—as most always connecting their jack shields to the ground-plane of their circuit board. 
However, not all RJ45 jacks (with built-in magnetics or not) have jack shields.  Most do but such is not required.

That is what I was implying,, so sometimes trial and error might be needed.

 

21 hours ago, Superdad said:

I don't think @kostas6a3 has reported any functional problem.  I just reread all his posts going back a few pages and it seems his is just not liking the sound he is getting with with his multi-switch setup and the Melco.  Plus he has a whole mix of cables, power supplies, and a DIY USB cable.  So it would seem that he just needs to experiment (and maybe simplify) to find what suits his taste. 

What I was suggesting is that the outcome @kostas6a3is hearing (taste, preference, perception or otherwise) might be caused by something that has not been identified or has been overlooked.  So I agree that experimenting and trial and error might help identify if something is causing the outcome.

 

Given the approach to ethernet shield grounding is somewhat unpredictable, I figured that would be a good place to start, hence use UTP to focus on the effect of the shielding and leave everything else as a constant.

 

Previously I have inserted new devices into my network that have caused such a fundamental change that it prompted me to strip back my network and basically reassemble it based on a new paradigm created by experiencing the new device, such as adding a 2nd ER.

 

For instance, it is possible using the NUC as server and N100 as player might be better than using the N100 as both server and player.

[kostas6a3]

13 minutes ago, dbastin said:

For instance, it is possible using the NUC as server and N100 as player might be better than using the N100 as both server and player.

Dear @dbastin, initially it was the first test I did.

The N100 with Twonky and Linn Kazoo Server sounds a bit better than the NUC Roon Rock (about 5%).

But I sacrifice this loss for what Roon provides.

Now the question is whether I should keep the second EtherREGEN in order.

At the moment I agree with the @barrows "My personal approach would be to keep it simple."

I am waiting for HDplex to see if it will change the SQ.

[R1200CL]

On 6/11/2022 at 4:58 AM, kostas6a3 said:

Now the question is whether I should keep the second EtherREGEN in order.

PM me If it’s for sale. If a quick agreement it shall be shipped to Kreta to be smuggled to Norway 😀