EtherREGEN: The long development thread. [Some Gen2 dev. pics and update starting on page 92.]

[octaviars]

7 minutes ago, JohnSwenson said:

The GND screw is on the A side gnd, so checking continuity between the BNC shell and the GND screw will show no continuity.

 

This is not what I did I checked between BNC shell and the shell on the RJ45 on B-side they are supposed to have the same potential or did I get that wrong?

[Bricki]

8 hours ago, jos said:

You can buy a half a meter here: https://www.futureshop.co.uk/supra-cat8-flame-retardant-ethernet-cable, or here: https://www.av-connection.com/?PNo=SUPCAT8-005 (quite cheap). I have this short cable from my Apple router to my ER and I just removed the golden shield at one end with a small kitchen knife quite easily.

Did you notice any audible differences after the removal?

[JohnSwenson]

6 hours ago, octaviars said:

 

This is not what I did I checked between BNC shell and the shell on the RJ45 on B-side they are supposed to have the same potential or did I get that wrong?

Aha, the RJ45 shell is not galvanically connected to the gnd plane, it is common to capacitively couple it to the plane. But high impedance leakage can still make it through.

 

John S.

[flkin]

9 hours ago, JohnSwenson said:

The shell of the BNC jack is connected to the B side gnd, NOT the A side gnd. The GND screw is on the A side gnd, so checking continuity between the BNC shell and the GND screw will show no continuity.

 

There is very high isolation between the A and B side grounds.

 

John S.

 

If I plan to use an external reference clock into the BNC jack on the B side, I take it that it would then be best to connect the A side to the endpoint (assuming the switch will be used as a one input and one output device).

 

Will the power supply to the reference clock matter? I'm guessing it shouldn't anymore but would like to confirm this. 

 

Thanks, Kin

[octaviars]

3 hours ago, JohnSwenson said:

Aha, the RJ45 shell is not galvanically connected to the gnd plane, it is common to capacitively couple it to the plane. But high impedance leakage can still make it through.

 

John S.

 

Just so I got this right.

 

If you dont use any clock connection it does not matter if the LAN cable on the B-side have the shield connected at the eR side as the connection ends via a capacitor to the groundplane on the B-side?

 

But if you use a clock connection you might form a ground loop or even disable the whole isolation if the clock is powered from something that also connects to the A-side?

 

 

[JohnSwenson]

2 hours ago, flkin said:

 

If I plan to use an external reference clock into the BNC jack on the B side, I take it that it would then be best to connect the A side to the endpoint (assuming the switch will be used as a one input and one output device).

 

Will the power supply to the reference clock matter? I'm guessing it shouldn't anymore but would like to confirm this. 

 

Thanks, Kin

If you are using an external clock and the endpoint has an RJ45, definitely hook the endpoint up to the B side.

 

I haven't done any A vs B power supply testing with external clocks, but my guess is they are quite important. Any leakage coming through the power supply will be directly modulating the clock output. Noise is also probably important. Interestingly what makes a good supply for a clock may very well be different than what makes a good supply for a switch or an endpoint.

[flkin]

8 minutes ago, JohnSwenson said:

.. Any leakage coming through the power supply will be directly modulating the clock output. ..

Would passing over the moat in the EtherRegen remove/reverse the effects of the modulated clock output in this case?

 

What I’m trying to ask is whether a B side external clock’s power supply matter on the A side?

[JohnSwenson]

2 hours ago, octaviars said:

 

Just so I got this right.

 

If you dont use any clock connection it does not matter if the LAN cable on the B-side have the shield connected at the eR side as the connection ends via a capacitor to the groundplane on the B-side?

 

But if you use a clock connection you might form a ground loop or even disable the whole isolation if the clock is powered from something that also connects to the A-side?

 

 

The fact that there is a capacitor on the B side shield may be irrelevant. High impedance leakage (the type that comes from SMPS) sails right through the cap.

 

What is super clean and what is not so clean is going to depend VERY much on the system and how things are connected, again no simple always works rule.

 

If there is a way that leakage can get from something connected to the B RJ45, through the RJ45, through the external clock, then you will have a problem. (it won't sound as good as it could, it doesn't mean it will sound bad, just not quite as good as it could). So lets look at how this can happen:

 

B side RJ45

High impedance leakage can get through the Ethernet connection whether it is shield connected or not. So if the endpoint is powered by an SMPS it can get through to the B side ground. If the endpoint is powered by an lps, the only way that can happen is through both end tied shielded cable.

 

External clock:

Any thing going through the shield of the external clock cable will be directly connected to the B side ground. So it depends on what is happening inside the external clock box. That you may not know. Again if it is powered by an SMPS there is the possibility that high impedance leakage could get through. If powered by an lps you can still have low impedance leakage getting through to the B side gnd.

 

In order for there to be a problem a loop has to occur through the endpoint and the external clock. Looking at the above it looks like if both the external clock and the endpoint are powered from a decent lps then there will be no path between them.

 

The BIG problem is if you have an external clock driving more than one thing, you could wind up shunting the ER isolation and creating a leakage loop through the external clock. If the other clock connection is to something connected to the A side (not just directly, clock to motherboard and regular switch in-between can still cause a problem) the leakage loop can cause modulation of the clock feeding the ER.

 

Personally I would avoid using an external clock for anything "upstream" (on the A side) of the ER. It MAY be possible to come up with something for clocking an upstream device, but that is going to take intimate knowledge of the internals of the external clock in order to figure out if any leakage loops will form.

 

So the general recommendation is to use lps for the endpoint and external clock, and only use external clock for ER and downstream devices. If you do this you can probably use a shield tied Ethernet cable between the B side and the endpoint.

 

John S.

[octaviars]

I think I will to be safe just tie the shield at my endpoint even if I dont have any external clock and LPSU in the endpoint. Great post by the way. @JohnSwenson 👍

[jos]

15 hours ago, Bricki said:

Did you notice any audible differences after the removal?

No, just peace of mind 😇

[nonesup]

11 hours ago, JohnSwenson said:

The fact that there is a capacitor on the B side shield may be irrelevant. High impedance leakage (the type that comes from SMPS) sails right through the cap.

 

What is super clean and what is not so clean is going to depend VERY much on the system and how things are connected, again no simple always works rule.

 

If there is a way that leakage can get from something connected to the B RJ45, through the RJ45, through the external clock, then you will have a problem. (it won't sound as good as it could, it doesn't mean it will sound bad, just not quite as good as it could). So lets look at how this can happen:

 

B side RJ45

High impedance leakage can get through the Ethernet connection whether it is shield connected or not. So if the endpoint is powered by an SMPS it can get through to the B side ground. If the endpoint is powered by an lps, the only way that can happen is through both end tied shielded cable.

 

External clock:

Any thing going through the shield of the external clock cable will be directly connected to the B side ground. So it depends on what is happening inside the external clock box. That you may not know. Again if it is powered by an SMPS there is the possibility that high impedance leakage could get through. If powered by an lps you can still have low impedance leakage getting through to the B side gnd.

 

In order for there to be a problem a loop has to occur through the endpoint and the external clock. Looking at the above it looks like if both the external clock and the endpoint are powered from a decent lps then there will be no path between them.

 

The BIG problem is if you have an external clock driving more than one thing, you could wind up shunting the ER isolation and creating a leakage loop through the external clock. If the other clock connection is to something connected to the A side (not just directly, clock to motherboard and regular switch in-between can still cause a problem) the leakage loop can cause modulation of the clock feeding the ER.

 

Personally I would avoid using an external clock for anything "upstream" (on the A side) of the ER. It MAY be possible to come up with something for clocking an upstream device, but that is going to take intimate knowledge of the internals of the external clock in order to figure out if any leakage loops will form.

 

So the general recommendation is to use lps for the endpoint and external clock, and only use external clock for ER and downstream devices. If you do this you can probably use a shield tied Ethernet cable between the B side and the endpoint.

 

John S.

Now is when I am already confused.  I thought I understood that if the cable tied at both ends was on side B, there was no problem.  After reading this post I am not so clear.  So if you have a cable tied at both ends that goes from the B side to a server that has a switched internal PSU and no external clock is used, are there problems or not?

[jcn3]

25 minutes ago, nonesup said:

Now is when I am already confused.  I thought I understood that if the cable tied at both ends was on side B, there was no problem.  After reading this post I am not so clear.  So if you have a cable tied at both ends that goes from the B side to a server that has a switched internal PSU and no external clock is used, are there problems or not?

 

John S has been consistent in saying that the shield should not be attached at both ends on the B side.

[nonesup]

27 minutes ago, jcn3 said:

 

John S has been consistent in saying that the shield should not be attached at both ends on the B side.

Very interesting.

How do you understand Alex's post 7578, which is attached?

 

“The ‘B’ side of the EtherREGEN is neither susceptible, nor can it become part of, ANY leakage loops. The ‘B’ side circuitry and port exist on their own entirely isolated power and ground domain. [Only the External clock BNC ground is common to the ‘B’-side ground-plane.] So if you have a fancy metal-shell shield-tied Ethernet cable, go ahead and use it on the ‘B’ side port for connection your DAC-attached streamer/renderer endpoint. This is the connection that matters most when using an EtherREGEN anyway: If you are going to hear a difference with an Ethernet cable, it will more likely be with the ‘B’-side cable. This is also a good one to keep short.”

[Superdad]

1 minute ago, nonesup said:

Very interesting.

How do you understand Alex's post 7578, which is attached?

 

Please do not take everything I say as gospel! 

And as John has explained, this stuff gets complicated and unless he stopped everything else to try and measure various leakage and sources of same with all kinds of configurations (NOT going to happen!) you guys are going to need to experiment and find what works best for you.  We are giving basic guidelines for use of the EtherREGEN.  It sounds terrific straight out of the box (I guess better with 100 hours as reported). Beyond that just have fun and don't break anything! 

[JohnSwenson]

17 minutes ago, nonesup said:

Very interesting.

How do you understand Alex's post 7578, which is attached?

 

“The ‘B’ side of the EtherREGEN is neither susceptible, nor can it become part of, ANY leakage loops. The ‘B’ side circuitry and port exist on their own entirely isolated power and ground domain. [Only the External clock BNC ground is common to the ‘B’-side ground-plane.] So if you have a fancy metal-shell shield-tied Ethernet cable, go ahead and use it on the ‘B’ side port for connection your DAC-attached streamer/renderer endpoint. This is the connection that matters most when using an EtherREGEN anyway: If you are going to hear a difference with an Ethernet cable, it will more likely be with the ‘B’-side cable. This is also a good one to keep short.”

If you are NOT using an external clock then using an end to end tied shielded cable should be fine. It looks like the only difference to that is if you have an external clock. The end to end tied shielded cable MAY cause a problem with that.

 

John S.

[jos]

32 minutes ago, nonesup said:

Very interesting.

How do you understand Alex's post 7578, which is attached?

 

“The ‘B’ side of the EtherREGEN is neither susceptible, nor can it become part of, ANY leakage loops. The ‘B’ side circuitry and port exist on their own entirely isolated power and ground domain. [Only the External clock BNC ground is common to the ‘B’-side ground-plane.] So if you have a fancy metal-shell shield-tied Ethernet cable, go ahead and use it on the ‘B’ side port for connection your DAC-attached streamer/renderer endpoint. This is the connection that matters most when using an EtherREGEN anyway: If you are going to hear a difference with an Ethernet cable, it will more likely be with the ‘B’-side cable. This is also a good one to keep short.”

 

 

25 minutes ago, Superdad said:

 

Please do not take everything I say as gospel! 

And as John has explained, this stuff gets complicated and unless he stopped everything else to try and measure various leakage and sources of same with all kinds of configurations (NOT going to happen!) you guys are going to need to experiment and find what works best for you.  We are giving basic guidelines for use of the EtherREGEN.  It sounds terrific straight out of the box (I guess better with 100 hours as reported). Beyond that just have fun and don't break anything! 

It’s not a gospel for sure, but very black in the manual too.

[nonesup]

26 minutes ago, Superdad said:

 

Please do not take everything I say as gospel! 

And as John has explained, this stuff gets complicated and unless he stopped everything else to try and measure various leakage and sources of same with all kinds of configurations (NOT going to happen!) you guys are going to need to experiment and find what works best for you.  We are giving basic guidelines for use of the EtherREGEN.  It sounds terrific straight out of the box (I guess better with 100 hours as reported). Beyond that just have fun and don't break anything! 

Oh that's fine, because my Göbel Ethernet cable cost me "a ball", that sounds great and it's not to be left in a corner.😁

[Ehsu]

7 minutes ago, JohnSwenson said:

If you are NOT using an external clock then using an end to end tied shielded cable should be fine. It looks like the only difference to that is if you have an external clock. The end to end tied shielded cable MAY cause a problem with that.

 

John S.

I just tested all my fancy ethernet cables with a multimeter and they are ALL tied shielded at both ends! They include Audioquest Cinnamon, Supra Cat8 and various Chord Cables. I heard Cat7 & Cat8 standards require both ends to be shielded. So are you suggesting to use Cat5 or Cat6 only? Is this shielding suggestion relevant if ER is grounded?  

[Jud]

For what some people are spending on Ethernet cables, you can have an Optical Module and inexpensive optical Ethernet cable.

 

I personally have just plugged these into the system and everything sounds great:

 

https://smile.amazon.com/Monoprice-SlimRun-Cat6A-Ethernet-Patch/dp/B01BGV2TAA/ref=mp_s_a_1_3?keywords=monoprice+slimrun+ethernet&qid=1574812177&sr=8-3
 

They replaced Blue Jeans tested and certified 6a, and everything sounded great with those too. Only reason I replaced them is that they’re heavy and stiff, and I figured the new light and flexible ones would torque the Ethernet connections less.

 

Thanks to @lmitche for turning me on to the Slimrun cables.

[JohnSwenson]

1 hour ago, Ehsu said:

I just tested all my fancy ethernet cables with a multimeter and they are ALL tied shielded at both ends! They include Audioquest Cinnamon, Supra Cat8 and various Chord Cables. I heard Cat7 & Cat8 standards require both ends to be shielded. So are you suggesting to use Cat5 or Cat6 only? Is this shielding suggestion relevant if ER is grounded?  

 Are you using an external clock? If NOT use whatever you wish. If you ARE using an external clock you MAY have an issue with those end to end tied cables.

 

I'm not TELLING you what to do. I'm stating where there MAY be an interaction that might cause a decrease in sound quality. It is entirely up to you to try different cables and listen for your self. If your expensive cables sound the best, stick with them. If something that is not shield tied from end to end sounds better, use that.

 

All the above is related to the B side, the ground screw etc is just for the A side.

 

John S.

[Ehsu]

6 minutes ago, JohnSwenson said:

 Are you using an external clock? If NOT use whatever you wish. If you ARE using an external clock you MAY have an issue with those end to end tied cables.

 

I'm not TELLING you what to do. I'm stating where there MAY be an interaction that might cause a decrease in sound quality. It is entirely up to you to try different cables and listen for your self. If your expensive cables sound the best, stick with them. If something that is not shield tied from end to end sounds better, use that.

 

All the above is related to the B side, the ground screw etc is just for the A side.

 

John S.

Great!! I am not using an external clock. It can be confusing reading people's comments on the forum 😞 Thank you John!

 

[jandersonhill]

I have to say the interaction with Uptone Audio is second to none. Far too many companies will sell you a trip to the moon and, once you've paid, say "it's up to you to figure out how to get there!". Uptone actually engage and offer helpful advice, even (or especially) in such a complex area such as grounding.

 

Obviously, it helps that they also build fantastic products!

 

 

[Bernstein]

Hehe...just connected the ER to my Keces P8 LPS with a cable for grounding. 
Idling at 0,8A/12V while my Sonictransporter i5 is idling at 0,6A/12V. Mini Computer speaking  

As I updated the FW, I could experience the overall PCB - quiet nice and a lot of chips are located on smallest room. @JohnSwenson had a hard job to Lay-out with good grounding in mind. Chapeau! 

[Superdad]

New blog post and firmware released!

https://uptoneaudio.com/blogs/news/etherregen-huge-success-new-firmware-for-november-units-reviews

[sircharles]

I just did this update. I went pretty quick and without a hitch. Thanks.