EtherREGEN: Installation, Usage, Difficulty, Questions thread

[esmit]

5 hours ago, jcn3 said:

Can you put the Roon server and the Rendus all on the ER?

 

What kind of switch is the 8 port switch? How did you have the 8 port switch connected to the ER (what on what side)? (Most issues seem to be because of trying to put gigabit connections on the B side)

The 8 port switch is not connected to the ER, it is connected to the isp router. The ER is also connected to the isp router. 

I am reluctant to try to connect the Roon server to the ER, I prefer that it just works as it was. 

 

Powering the isp router off and on while keeping all the rest powered did restore all connections, it seems. Running for 12 hours now. 

[Superdad]

1 hour ago, esmit said:

Powering the isp router off and on while keeping all the rest powered did restore all connections, it seems. Running for 12 hours now. 

 

Glad to hear you have it running steady now! 

 

We are convinced that for the handful of users having connectivity problems, it has something to do with loss of IP address lease with DHCP server/router. 

We are looking for common denominators, and so far the most obvious one is use of OrbiterOS (with SonicTransporters or Rendus).

Also, there have been a couple of instances of Ethernet DACs which use the Conversdigital mconnect input module (Ayre and PS Audio) where it took a while before they and the router would settle down and keep a stable IP address. But both those folks have reported being good now. 

 

I’m not a networking expert so I confess that I don’t fully understand what is going on. But John is very sharp with this stuff and there is veteran network engineer (here on this forum) who has stepped forward to assist in researching the matter. Once there is a conclusion and a consistent remedy procedure, we’ll share that here as well as post it to our website. 

 

We appreciate the patience—and the detailed information reports—of those who are experiencing difficulty. None of this showed up with any of our beta testers, that’s why we are caught a bit short with this issue now. 

[esmit]

12 hours ago, Superdad said:

Glad to hear you have it running steady now! 

Had to reboot the isp router again because ultrarendu, microrendu and chromecast audio were lost agaiin, but still works 8 hours later. I left all other devices powered on by the way.

[_JL_]

External Clock

A quick note on using the newclassd Neutron Star 2 10Mhz clock with the EtherRegen. The NS2 defaults to output clock signal at 5V which is not compatible with the ER. At 5V all ER leds lights up correctly and flashing but the link became very unstable, presumably due to input clock signal clipping. I did a speedtest and only got 2Mbps.

 

Next I reconfigured the NS2 to output clock signal at 3.3V and all is well. Speedtest showed 90+Mbps up/down streams. So if one is using external clock make sure it outputs 3.3V or thereabout. This is not the section for sound impressions so I'll just say that the NS2 clock sounded very very different from the internal clock (does this ever end?). Can't say which is better at this point so will let them run-in before doing comparisons.

 

Btw, if anyone is trying NS2 you can buy ready-made coaxial cable at digikey: http://d.digikey.com/LD10KX02iATF0SzQL0N070p

[Bernstein]

3 hours ago, esmit said:

Had to reboot the isp router again because ultrarendu, microrendu and chromecast audio were lost agaiin, but still works 8 hours later. I left all other devices powered on by the way.

Did you proceed a dedicated power up sequence to get it running or just

powering on?

[esmit]

16 minutes ago, Bernstein said:

Did you proceed a dedicated power up sequence to get it running or just

powering on?

That would be a sequence of one: only the isp modem was powered off and on. Rest was not powered off. 

[lxgreen]

On 11/12/2019 at 9:17 PM, Superdad said:

 

@Nenon is correct. The copper port of the opticalModule is Gigabit only, while the 'B'-side port of the EtherREGEN is 100Mbps only.  That is why your connection setup is not working.

 

It will always be best to have the EtherREGEN as the last device--directly connected to your DAC-attached computer/streamer/renderer endpoint.  So for opticalRendu users this means running the EtherREGEN B>A, with its 'A'-side SFP/optical directly connected to your opticalRendu.  

As mentioned elsewhere, the main reason we went to the trouble and expense of making the reclocking symmetrical on both sides of the moat was so that users with optical endpoints could "turn it around" and still get top performance.

I have been away since originally installing the EtherRegen so tried again after reading replies. Yes, it does work after reversing inputs, removing optical module, connecting optical renduSE to A side optical port. I’m disappointed since after following the development discussion of EtherRegen from beginning nothing was ever mentioned about compatibility with optical module although plenty of talk about 100Mbps and gigabit. And now I’m still confused about what else I can plug into a ports since previously it was mentioned not to plug anything else into A ports if used this way. Audio vs non audio? Not sure what this means. Please understand your customers range from electrical engineering experts to users like myself just trying to get the best sq I can from my audio system.

[Superdad]

1 hour ago, lxgreen said:

I have been away since originally installing the EtherRegen so tried again after reading replies. Yes, it does work after reversing inputs, removing optical module, connecting optical renduSE to A side optical port. I’m disappointed since after following the development discussion of EtherRegen from beginning nothing was ever mentioned about compatibility with optical module although plenty of talk about 100Mbps and gigabit. And now I’m still confused about what else I can plug into a ports since previously it was mentioned not to plug anything else into A ports if used this way. Audio vs non audio? Not sure what this means. Please understand your customers range from electrical engineering experts to users like myself just trying to get the best sq I can from my audio system.

 

Hi: Sorry if this became confusing for you.  Let me see if I can make it clear.

The Sonore opticalModule is an excellent FMC (fiber media converter), probably the best on the planet. But its technology is a subset of the EtherREGEN's and so if using an EtherREGEN with a Sonore opticalRendu, the opticalModule is not needed. It is always desirable to have the EtherREGEN as the last device in the chain just before whatever is the computer/streamer/renderer endpoint--in your case the opticalRendu.

We went to the extra trouble and expense of making the EtherREGEN symmetrical specifically so that its optical SFP cage would be as good an "output" port as the 'B'-side copper port. But ideally, if going B>A for copper to optical, you would leave empty the other 'A' side (RJ45) ports. But that is not mandatory.

 

By the way, all of this is discussed in the EtherREGEN User Guide, in the same plain-speak as I always write--just as above. Please look to page 9.

 

Thanks,

--Alex C.

[Cable Monkey]

@Superdad What sort of voltages are we looking for at the clock input? 
Clock is the Teac CG-10M, output impedance 50 ohm, voltage 0.5v rms.

[vortecjr]

8 hours ago, Superdad said:

 

Hi: Sorry if this became confusing for you.  Let me see if I can make it clear.

The Sonore opticalModule is an excellent FMC (fiber media converter), probably the best on the planet. But its technology is a subset of the EtherREGEN's and so if using an EtherREGEN with a Sonore opticalRendu, the opticalModule is not needed. It is always desirable to have the EtherREGEN as the last device in the chain just before whatever is the computer/streamer/renderer endpoint--in your case the opticalRendu.

That is an overly zealous comment.

[vortecjr]

11 hours ago, jcn3 said:

I'm not clear on why you needed to post this -- he paid your product a significant compliment AND he posted in his own sponsored area.  since you are a vendor and have your own sponsored area, the comment doesn't seem appropriate to me.

I'm responding to a post about my products which under the forum rules I'm allowed to do.

 

With all do respect to Alex and John, we are industry friends and partners and the EtherREGEN is a great product, but this needs more discussion and consideration.

 

So here is the why:

1. The opticalModule, EtherREGEN, and a TP link optical FMC can all provide 100% galvanic isolation and all three will block leakage current into a opticalRendu;

2. Taking things a bit further the opticalModule and the EtherREGEN both have low noise oscillators and low noise power regulation on board. According to John's hypothesis having these upstream deminishes the clock signature. This hasn't been measured and the measurements haven't been compared between products;

3. Understanding 1 and 2, why would it ever be desirable to have the EtherREGEN as the last device in the chain just before whatever is the computer/streamer/renderer endpoint? Do computer/streamer/renderer endpoint have 100% galvanic isolation, low noise oscillators, and low noise power supplies on their network inputs....almost none do but the optical Rendu does:)    

[Superdad]

1 hour ago, vortecjr said:

I'm responding to a post about my products which under the forum rules I'm allowed to do.

Of course my friend.  I thought several of those other responses to you were harsh and deleted them much earlier.

 

1 hour ago, vortecjr said:

According to John's hypothesis having these upstream deminishes the clock signature. This hasn't been measured and the measurements haven't been compared between products;

3. Understanding 1 and 2, why would it ever be desirable to have the EtherREGEN as the last device in the chain just before whatever is the computer/streamer/renderer endpoint? Do computer/streamer/renderer endpoint have 100% galvanic isolation, low noise oscillators, and low noise power supplies on their network inputs....almost none do but the optical Rendu does:)    

 

While the galvanic isolation, low noise power supplies, and low-jitter oscillators common to the very fine Sonore and UpTone products are important and comparable, the EtherREGEN is about more than that. Its active differential isolation moat and ultra-low-jitter reclocking flops aim to erase the phase-noise fingerprint of upstream clocks--which cause ground- and power-plane noise on the board and inside the chips. This is what @JohnSwenson is gearing up to directly measure. And his explanation of the mechanisms will be part of the white paper which will accompany those measures.

Until such time as we are ready to publish that, all we can go on as proof of the concept are the many positive experiences which EtherREGEN users are having--including a number of opticalModule and opticalRendu owners who have posted in detail about the synergy between each of these Swenson designs.

[thotdoc]

On 11/16/2019 at 9:33 PM, Superdad said:

 

Glad to hear you have it running steady now! 

 

We are convinced that for the handful of users having connectivity problems, it has something to do with loss of IP address lease with DHCP server/router. 

We are looking for common denominators, and so far the most obvious one is use of OrbiterOS (with SonicTransporters or Rendus).

Also, there have been a couple of instances of Ethernet DACs which use the Conversdigital mconnect input module (Ayre and PS Audio) where it took a while before they and the router would settle down and keep a stable IP address. But both those folks have reported being good now. 

 

I’m not a networking expert so I confess that I don’t fully understand what is going on. But John is very sharp with this stuff and there is veteran network engineer (here on this forum) who has stepped forward to assist in researching the matter. Once there is a conclusion and a consistent remedy procedure, we’ll share that here as well as post it to our website. 

 

We appreciate the patience—and the detailed information reports—of those who are experiencing difficulty. None of this showed up with any of our beta testers, that’s why we are caught a bit short with this issue now. 

Anything new on this. I was listening tonite and the music just stopped.

 

[Superdad]

6 minutes ago, thotdoc said:

Anything new on this. I was listening tonite and the music just stopped.

 

We think we have it figured out. Will report more and deliver a solution as quickly as possible. So far the issue has affect only about 6-8 users.

Thanks for your patience.

--Alex C.

[thotdoc]

1 minute ago, Superdad said:

 

We think we have it figured out. Will report more and deliver a solution as quickly as possible. So far the issue has affect only about 6-8 users.

Thanks for your patience.

--Alex C.

Thanks

 

G

[k27R]

1 hour ago, Superdad said:

 

We think we have it figured out. Will report more and deliver a solution as quickly as possible. So far the issue has affect only about 6-8 users.

Thanks for your patience.

--Alex C.

I look forward to this.  It was working great all weekend without a hiccup.  Then today dropped out and it won’t reconnect 

[vortecjr]

6 hours ago, Superdad said:

Of course my friend.  I thought several of those other responses to you were harsh and deleted them much earlier.

 

 

While the galvanic isolation, low noise power supplies, and low-jitter oscillators common to the very fine Sonore and UpTone products are important and comparable, the EtherREGEN is about more than that. Its active differential isolation moat and ultra-low-jitter reclocking flops aim to erase the phase-noise fingerprint of upstream clocks--which cause ground- and power-plane noise on the board and inside the chips. This is what @JohnSwenson is gearing up to directly measure. And his explanation of the mechanisms will be part of the white paper which will accompany those measures.

Until such time as we are ready to publish that, all we can go on as proof of the concept are the many positive experiences which EtherREGEN users are having--including a number of opticalModule and opticalRendu owners who have posted in detail about the synergy between each of these Swenson designs.

No need...I welcome the descending opinions and comments. We don’t need to agree on everything and discussion is necessary to furthering our understanding. I know you and I are after the same thing and that is to understand this completely and scientifically.

 

I understand the moat, the flip flops, and the clock fingerprint concept but those measurements are needed for validation and IMHO you should wait and proceed with caution about claiming things. If John is correct even an opticalModule will diminish the effect of the clock fingerprint. However, I don’t claim this anywhere in the marketing of the product because we lack the measurements to prove it. I also don’t think it’s audible but that is another discussion.
 

Also, in regards to my comment #3 above what good will the moat, the flip flops, and the clock fingerprint concept have if you are connecting to crappy endpoints which themselves will inject the very thing you are trying to isolate into your DAC. People need to understand this if it has any chance of working as John intends. Its not possible to achieve what John is trying to do with typical computers, streaming DACs, and some endpoints. 

[LowMidHigh]

38 minutes ago, vortecjr said:

No need...I welcome the descending opinions and comments. We don’t need to agree on everything and discussion is necessary to furthering our understanding. I know you and I are after the same thing and that is to understand this completely and scientifically.

 

I understand the moat, the flip flops, and the clock fingerprint concept but those measurements are needed for validation and IMHO you should wait and proceed with caution about claiming things. If John is correct even an opticalModule will diminish the effect of the clock fingerprint. However, I don’t claim this anywhere in the marketing of the product because we lack the measurements to prove it. I also don’t think it’s audible but that is another discussion.
 

Also, in regards to my comment #3 above what good will the moat, the flip flops, and the clock fingerprint concept have if you are connecting to crappy endpoints which themselves will inject the very thing you are trying to isolate into your DAC. People need to understand this if it has any chance of working as John intends. Its not possible to achieve what John is trying to do with typical computers, streaming DACs, and some endpoints. 

 

Its not possible to achieve what John is trying to do with typical computers, streaming DACs, and some endpoints. 

 

Common sense suggests that any reduction in phase noise/voltage noise/jitter injected into the recipient would decrease the overall noise within its enclosure. Consequently, we should expect a more precise and "cleaner" output signal. 

 

Of course, the improvement may not be audible if the recipient self-generated "filth" drowns any improvement in the input signal. 

[Cable Monkey]

You cannot mitigate someone’s choice to use an endpoint that generates its own noise. But the moat works both ways so at least it shouldn’t impact anything connected on the A side. In short it may not sound (or measure) as good as other devices but at least it won’t be detrimental to other things on the network.

[vortecjr]

20 minutes ago, LowMidHigh said:

 

Its not possible to achieve what John is trying to do with typical computers, streaming DACs, and some endpoints. 

 

Common sense suggests that any reduction in phase noise/voltage noise/jitter injected into the recipient would decrease the overall noise within its enclosure. Consequently, we should expect a more precise and "cleaner" output signal. 

 

Of course, the improvement may not be audible if the recipient self-generated "filth" drowns any improvement in the input signal. 

I have spoken to John about this on several occasions and he feels that the clock closest to the DAC is the most important one because the effect of the clock fingerprint diminishes as it passes from gear to gear. Unfortunately, I think you are correct about the filth.  We are talking about devices with incredibly low noise floors and distinguishing something will be very difficult.

[vortecjr]

15 minutes ago, Cable Monkey said:

You cannot mitigate someone’s choice to use an endpoint that generates its own noise. But the moat works both ways so at least it shouldn’t impact anything connected on the A side. In short it may not sound (or measure) as good as other devices but at least it won’t be detrimental to other things on the network.

No one is mitigating anyones choice here. Also we’re not talking about the A side of the device. Clearly anything connected to the A side will have no effect on the rest of the network upstream.

[pl_svn]

1 hour ago, vortecjr said:

if you are connecting to crappy endpoints

 

please be more explicit as "crappy endpoints" doesn't help/educate/give any clue to potential customers: which endpoints are you calling "crappy"? make/model, please: purpose of this forum is helping people make good choices, isn't it? 

 

or, at the very least: what do you suggest as the bare minimum of "good enough"? again: make/model!

[rickca]

JS has explained why the ER B-side is limited to 100 Mbps.  Why was the OM designed with to be gigabit only with no autonegotiation to 100 Mbps?  @vortecjr or @JohnSwenson can you please explain?

[vortecjr]

14 minutes ago, pl_svn said:

 

please be more explicit as "crappy endpoints" doesn't help/educate/give any clue to potential customers: which endpoints are you calling "crappy"? make/model, please: purpose of this forum is helping people make good choices, isn't it? 

 

or, at the very least: what do you suggest as the bare minimum of "good enough"? again: make/model!

I didn’t want to be that direct:) 

[lmitche]

It seems to me this isn't a good time for Alex and John to be making any effort to argue design choices with "friends and partners". They are in the middle of a product rollout and major manufacturing push.

 

At a minimum this discussion, if it needs to continue, should be had in a another thead. It is way OT.