Master Clock for your EtherREGEN

[Argon]

After upgrading to the AfterDark Emperor Double Crown for my ER I have a BG7TBL 75 ohm clock and Ghent cable (~14") that I can send on to someone who is ready to try external clocking of their ER.  I ran this for 8 months direct to the ER w/o a minicircuts filter, I was unable to get a scope on this to confirm square of sine wave but it worked great as is.  I just ask the recipient cover shipping but else I'm happy to send it on FOC.

 

PM me and we can organize the bits.  

 

Nicholas

[aubreybobb]

Mihaylov, PYP, Richard, many thanks your info.  Will try your suggestions.

 

Best wishes

 

Aubrey

[dylanesque]

17 hours ago, Mihaylov said:

breaks not only the moat in the ER, but also makes it pointless to use the ethernet interface at all (twisted pair or optics all the same) since it deprives ethernet of its main advantage, namely galvanic device isolation. Be very careful in the network configuration so as not to break the galvanic isolation and use a separate clock for each device on the local network, then it will be sufficient one switch instead of several.

Further to this @Mihaylov, is this true (i.e compromising the moat) even if the clock in context has “Galvanically isolated outputs” to feed devices on the same network. 

e.g Ref 10

[Mihaylov]

Curious how they achieve galvanically isolation of the outputs? With some chips like Alex and John in ER?

 

 

Has anyone tried checking the resistance between the grounds of these outputs? Is that resistance equal to 0 or is there a gap (very large resistance)?

 

 

[JohnSwenson]

9 hours ago, dylanesque said:

Further to this @Mihaylov, is this true (i.e compromising the moat) even if the clock in context has “Galvanically isolated outputs” to feed devices on the same network. 

e.g Ref 10

Lets talk about the term "galvanically isolated". This means there is a high impedance (usually at least a mega-ohm) AT DC between two circuits. Note the DC, it doesn't say anything about AC signals or noise.

 

Specifically just because a device has galvanic isolation DOES NOT mean it blocks leakage current. Leakage current is primarily line frequency (50/60Hz) and lower harmonics. One of the most common galvanic isolators is a transformer. It blocks DC, but lets some frequencies through. You would think that a transformer designed to let high frequency signals through would block leakage current, but this has not been what I have found. I have measured a lot of high frequency transformers and found that leakage current from SMPS sails right through most of them.

 

The SMPS distinction is important. The leakage current from SMPS and linear supplies is quite different (I'm not going into detail here, I've written a lot about this in other places), it is the SMPS type that goes through most high frequency transformers. The leakage from linear supplies DOES get blocked by these transformers.

 

Most clock boxes that say they are galvanically isolated do so with transformers on the output. So how does using such a clock box affect the moat of an ER? Well "it depends" of course!

 

Let's take a simple case: clock box with two outputs, both with their own transformer, one output driving an ER and the other driving some network device plugged into the A side. If ANY device plugged into the A side is powered from an SMPS then leakage current from that SMPS can go through the clock line on the A side, through its transformer, through the other transformer, out the other clock line and into the B side of the ER. If everything on the A side is powered from a linear supply, including everything upstream, then this won't happen.

 

This is why optical network can be advantageous, they DO block leakage current from anything upstream. They do NOT necessarily clean up jitter on the data, but they do block leakage current.

 

Unfortunately no simple rules you have to take each configuration separately and work it out.

 

John S.

[PYP]

41 minutes ago, JohnSwenson said:

The SMPS distinction is important. The leakage current from SMPS and linear supplies is quite different (I'm not going into detail here, I've written a lot about this in other places), it is the SMPS type that goes through most high frequency transformers. The leakage from linear supplies DOES get blocked by these transformers.

 

John S.

Does a dedicated electrical line for one's setup provide leakage current protection from all the other lines that meet at the circuit breaker panel box?  I can understand why reducing/eliminating SMPS supplies on the same dedicated line would be beneficial, but wonder about whether the fact that everything joins at the panel box means the dedicated circuit is not complete protection from SMPS in other places in the home.

 

Thanks.  

[Mihaylov]

54 minutes ago, JohnSwenson said:

Most clock boxes that say they are galvanically isolated do so with transformers on the output.

Can't find these transformers on the Mutec board.

 

[JohnSwenson]

28 minutes ago, PYP said:

Does a dedicated electrical line for one's setup provide leakage current protection from all the other lines that meet at the circuit breaker panel box?  I can understand why reducing/eliminating SMPS supplies on the same dedicated line would be beneficial, but wonder about whether the fact that everything joins at the panel box means the dedicated circuit is not complete protection from SMPS in other places in the home.

 

Thanks.  

To understand this takes some background information on what leakage currents are:

 

Leakage current is a property of power supplies, it is a small portion of the line current (50/60Hz) somehow "leaking" into the output DC connection. This current flows in loops back to the power supply through the AC line. So you have leakage going to DC, through DC cables, signal cables etc until it somehow gets back to the original power supply. Frequently this happens through the safety ground of some box in the system, through the safety ground to the panel where the connection to the neutral exists and back to the power supply. If the original power supply has a safety ground the loop may be much shorter.

 

The loop cannot happen just through the AC wiring, there has to be some other path such as signal wiring, network wiring etc to form the loop. Thus the SMPS in an alarm clock cannot get into the audio system. BUT a router or switch (powered by SMPS) CAN get into the audio system on a different branch. If there is an optical segment in there then that DOES block the loop.

 

John S.

 

[JohnSwenson]

15 minutes ago, Mihaylov said:

Can't find these transformers on the Mutec board.

 

Look at the right side of the board, there is a vertical line of red cubes (those are capacitors), just to the right of each cube looks like a toroid core with wiring, I presume those are the transformers.

 

John S.

[JohnSwenson]

26 minutes ago, JohnSwenson said:

Look at the right side of the board, there is a vertical line of red cubes (those are capacitors), just to the right of each cube looks like a toroid core with wiring, I presume those are the transformers.

 

John S.

Sorry, zooming in it just looks like components on the board. They could just be using the caps, in which case the grounds are probably connects, but they might have something else in there.

 

John S.

[aubreybobb]

Hi John, 

 

I'm a newbie in this forum - 2 ER switch's and one Sotm in a cascade from router and NAS to dCS streamer.  Streamer direct to active digital speakers.  The SOTM/ER connection is with a Finnisar/glass connection and the two ER's are with copper.   Its a nuisance turning this around so that the fibre would be between the two ER's, but I will try it if you think its a good idea and can spare the time to guide me. 

 

Meantime, some counter intuitive stuff:

 

1) I have replaced a number of high end LPS's with iFi SMPS's in my system, always with great results. 

2) after reading a few of the above messages regarding breaking the moat, I decided to remove the clock connection (double emperor dCS version - 2 x word clock and 4 x 10M out) from one ER.  The drop in SQ was dismaying.  So it went right back in.  This will be tried again once everything full burned in - its only 72 hours in now.

3) Everything (except the clock itself, as yet) is grounded to Synergistic Research active grounding box.

4) One ER is A->B, but second B->A. 

 

As the great JS once said and I quote - Well "it depends" of course! 

 

But meantime, thank you for and your colleague at afterdark, from whom I have learned so much about this subject.  You guys are definitely dedicated to helping us plebes get closer to the state of the art.  And so very generous  with your guidance...I am very grateful.  Must be wonderful to be in a career you obviously love so much.

 

Best wishes

 

Aubrey

[PYP]

5 hours ago, aubreybobb said:

Hi John, 

 

I'm a newbie in this forum - 2 ER switch's and one Sotm in a cascade from router and NAS to dCS streamer.  Streamer direct to active digital speakers.  The SOTM/ER connection is with a Finnisar/glass connection and the two ER's are with copper.   Its a nuisance turning this around so that the fibre would be between the two ER's, but I will try it if you think its a good idea and can spare the time to guide me. 

 

I'm definitely not John, but it seems to me that since use already use fiber before the eRs, it is fine to use copper between the two eRs (meaning, you already have galvanic isolation implemented and there is nothing to be gained).  I no longer use fiber, but connecting the two eRs with copper works well for me.

 

Regarding grounding:  You might want to experiment with grounding.  Easier to do and makes a difference.  However, if it already sounds great, you most likely have it optimized.

 

Regarding the clock:  As you know, it takes a month to stabilize; however, if it sounds great, you probably have it right.  

 

I don't think anyone will be able to tell you exactly how to connect all this stuff -- too many variables -- and trusting your ears is the way to go.  

 

Have fun.  

[R1200CL]

8 hours ago, aubreybobb said:

4) One ER is A->B, but second B->A. 

Which one has an external clock ? And is the clock used anywhere else ?

 

Are you doing A>B>B>A og B>A>A>B ?

[Mihaylov]

10 hours ago, aubreybobb said:

after reading a few of the above messages regarding breaking the moat, I decided to remove the clock connection (double emperor dCS version - 2 x word clock and 4 x 10M out) from one ER.  The drop in SQ was dismaying. 

Should have disconnected the clock from Sotm because it stands on a "muddy" bank to the ER.

[aubreybobb]

Thanks all, will try disconnecting the clock from SOTM Mihalov,
 

[aubreybobb]

10 hours ago, R1200CL said:

Which one has an external clock ? And is the clock used anywhere else ?

 

Are you doing A>B>B>A og B>A>A>B ?

oth are connected to an external clock.  The same clock is also connected to the SOTM and dCS NB

connection is A>B>B>A

[aubreybobb]

On 10/6/2021 at 1:08 AM, Mihaylov said:

Should have disconnected the clock from Sotm because it stands on a "muddy" bank to the ER.

Did it! Increase detail and presence. Thank you.

 

I still dont understand why this helped.  Any explanation, when you got a moment would be great.

 

B rgds

 

Aubrey

 

 

[Mihaylov]

All network devices before the ER moat must not share electrical ground with the network devices after the ER moat otherwise the moat will be destroyed with all the resulting consequences for sound. The clock outputs share a common electric ground, so simultaneously connecting one output of the clock to the Sotm and the second to the ER connects the electric grounds of the Sotm and ER ("clean" side, because the clock plugs to the ER on the "clean" side after the moat).

[Mihaylov]

On 6/11/2020 at 10:27 AM, JohnSwenson said:

There ARE some ways to convert sinewaves into square waves, but the simple ways actually increase the noise on the signal making the jitter even worse.

 

Is this way simple?

 

Quote

Clipped Sine Wave - the sine wave output is manipulated so it will not reach its max high or low. By doing this you are creating a square wave output without sacrificing any of the desired phase noise performance.

 

[JohnSwenson]

On 10/9/2021 at 3:31 PM, Mihaylov said:

 

Is this way simple?

 

 

This doesn't improve anything. The rate of change (slope of the curve) at the midpoint is what matters, just simply clipping the top and bottom off doesn't change the slope at the midpoint. Compare that to a true square wave which has a much steeper slope at the midpoint.

 

Now you CAN do the clipping and then amplify the signal, which increases slope at the midpoint, BUT that amplifies whatever noise is on the signal AND adds it's own noise as well. To do this well takes multiple stages and complicated filtering between stages. This IS how the best sine to square converters are built, but I don't think anyone would call it simple! Just clipping the top and bottom doesn't help.

 

John S.

[Rasputin]

On 10/7/2021 at 7:10 PM, Mihaylov said:

All network devices before the ER moat must not share electrical ground with the network devices after the ER moat otherwise the moat will be destroyed with all the resulting consequences for sound. The clock outputs share a common electric ground, so simultaneously connecting one output of the clock to the Sotm and the second to the ER connects the electric grounds of the Sotm and ER ("clean" side, because the clock plugs to the ER on the "clean" side after the moat).

Does this mean the power supply circuits should be different before and after ER moat? Or the network devices can be powered from the same power distributor sharing the ground? 

[R1200CL]

1 hour ago, Rasputin said:

Does this mean the power supply circuits should be different before and after ER moat?

The circuits (PCB)can be equal on all power supplies You just need enough power supplies 😉

But the power supplies can share the same circuit breaker, if that’s what you’re asking. 

 

1 hour ago, Rasputin said:

the network devices can be powered from the same power distributor sharing the ground? 

Your house has same ground plane all over. Hopefully. So yes, the power supplies itself can and ought to have same ground plane. Preferably from same circuit breaker. You shouldn’t ground or connect devices together. Especially if the devices is on different side of the moat. 

 

Which is interesting if you ground dual EtherRegen in a A>B>A>B together 🤪(Don’t). 
 

If this theme isn’t clear to you, please draw an example.

[Rasputin]

Thanks, clear now;) I might try to float the ground for regen and the streamer that is connected to the dac optically and let the rest of the system be grounded, would it be theoretically better? 

[Mihaylov]

14 hours ago, Rasputin said:

Or the network devices can be powered from the same power distributor sharing the ground? 

Yes, they can. It's not so much about grounding the power grid as it is about the electrical ground (common bus) circuitry of devices on the network, including their power supplies. (I hope it's spelled correctly in English).

[R1200CL]

10 hours ago, Rasputin said:

Thanks, clear now;) I might try to float the ground for regen and the streamer that is connected to the dac optically and let the rest of the system be grounded, would it be theoretically better? 

Can you draw a simple sketch ?

What is the Farad powering ?

Please also read the user guide for EtherRegen regarding the ground screw if that’s what you’re asking. 
What is your streamer and how is it powered ? And how is your BG7TBL powered ?

 

How shall we understand that the ground is floating for the EtherRegen? (In your case).