Jump to content
IGNORED

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


Recommended Posts

On 7/29/2019 at 3:03 AM, JohnSwenson said:

The power supply for the EtherREGEN goes into the A side, there is an isolating DC/DC converter between the A side and the B side. This converter has extremely small capacitance across the moat. There are a bunch of 3045 and 3042 regulators that take isolated output from the isolating converter and drive all the circuitry on the B side. You have a VERY high degree of isolation between the A side and B side for both data and power.

 

So you have an audio PS system and a non-audio PS system. If you power the EtherREGEN from the audio PS side and use RJ45 connection  to the A side, B side to DAC, you MIGHT get some leakage current through an electrical A side connection and the power supply on the audio PS side. IF the A side is connected with optical, that cannot happen.

 

IF the EtherREGEN is powered from the non-audio PS any leakage coming in through the PS or A side input will be blocked from the getting to the B side RJ45.

 

So it seems to me powering the EtherREGEN from the non-audio PS side is the safest no matter how you hook up the A side. There is no path for any leakage in the non-audio PS domain to get into the audio PS domain. With EtherREGEN power from the audio PS domain there is a small possibility of getting leakage from the non-audio PS domain, through an A side RJ45 through the EtherREGEN power supply and into the audio PS domain. If the A side connection is optical that cannot happen either.

 

I hope that makes sense.

 

John S.

 

 

Hi Mr. S.🙂

 I built a six outlet power strip and have it connected to an ultra-isolation transformer. I have two dedicated circuits and have  bonded the grounds together. 

 I currently have everything plugged into the power strip, except for the power supplies for my LPS-1 and 1.2. 

 Which methodology is more important, the everything in one strip or if you have a non-audio side,  like above,  to separate out the power supplies. 

 I have a 300W HDPlex ATX LPS that I would like to separate out with my other power supplies. 

 Thank you 

 

Link to comment

I have long time believed it was as easy as just use the EtherRegen “reverse way” in order to connect the EtherRegen to the opticalRendu. 
And maybe add my SonicTransporter as well.

As I expect none of the A side ports would affect the SFP port. 

 

Now I’m wondering if I should put a opticalModule between the EtherRegen and the opticalRendu  in order to have the signal going “right” way. Or even a second EtherRegen. 
 

Or maybe add a RJ45 SPF module on opticalRendu in order to use the output B side of the EtherRegen ?

 

Maybe I’m complicating this to much.....

Any thoughts ?

 

 

(Hope you have enough bourbon 😀)

 

 

Link to comment

You

5 minutes ago, R1200CL said:

I have long time believed it was as easy as just use the EtherRegen “reverse way” in order to connect the EtherRegen to the opticalRendu. 
And maybe add my SonicTransporter as well.

As I expect none of the A side ports would affect the SFP port. 

 

Now I’m wondering if I should put a opticalModule between the EtherRegen and the opticalRendu  in order to have the signal going “right” way. Or even a second EtherRegen. 
 

Or maybe add a RJ45 SPF module on opticalRendu in order to use the output B side of the EtherRegen ?

 

Maybe I’m complicating this to much.....

Any thoughts ?

 

 

 

 

 

You can certainly use the SFP port to connect optically to a opticalRendu. Connect the B RJ45 to the network. This will work great as long as you don't connect anything else to the A ports. If you do connect other devices then their clock signatures wind up on the A side ground plane and will affect what is going out on the SFP port. The clock signatures from the B side are blocked but anything else on the A side will affect the SFP port.

 

So yes you certainly can use the EtherREGEN to optically connect to an opticalRendu, don't use any of the other A side ports, just B side RJ45 to A side SFP.

 

John S.

Link to comment
On 7/30/2019 at 3:07 PM, JohnSwenson said:

The audio Endpoint should be on the opposite side from anything else. The way you have it here is probably the worst possible way to hook things up. The Endpoint is now susceptible to both leakage and clock effects from the NAS and external network.

 

@JohnSwenson and @Superdad. Maybe you guys missed my question earlier? I was wondering what would you suggest if I plan to plug in more than 1 end-points? Is the performance on the Gb sides inferior to the 10/100 side? My plan is to have Roon ROCK, home network, bedroom audio and Apple TV on the Gb side and my main music room streamer on the lone side to maximise main audio system performance. I will also use an audio linear PSU to power ER within a separate dedicated power circuit which is again separated from my main music system power circuit.   

Link to comment
2 hours ago, Ehsu said:

@JohnSwenson and @Superdad. Maybe you guys missed my question earlier? I was wondering what would you suggest if I plan to plug in more than 1 end-points? Is the performance on the Gb sides inferior to the 10/100 side? My plan is to have Roon ROCK, home network, bedroom audio and Apple TV on the Gb side and my main music room streamer on the lone side to maximise main audio system performance. I will also use an audio linear PSU to power ER within a separate dedicated power circuit which is again separated from my main music system power circuit.   

The really good leakage and clock isolation only happens when going across the moat. If you have your network connection and other audio endpoints on the A side you don't get the really good isolation between them. Thus leakage and clocking issues from the network will very likely have some effects on the other devices on the A side.

 

Even though an A side jack to A side jack doesn't have the full isolation of crossing the moat to the B side, it is still a REALLY good switch. Using the A side as a switch will probably give better sound than any other switch you can get, but not as good as going to the B side.

 

Your scheme sounds like a perfectly workable setup.

 

John S.

Link to comment

@JohnSwenson

If technical possible, can an optical version of the EtherRegen be made. 
Meaning exchange the RJ45 B side with a SFP cage ?

 

I think not, as the present design requires the B side to be 100 MHz, and since the reason for this request is to connect the opticalRendu, a 100 mH optical interface won’t work. (OpticalRendu requires 1000 MHz) 
 

So I ask again using a gigabit only RJ-45 SFP transceiver on the opticalRendu https://amzn.to/2XbBAJA be a better solution than using the EtherRegen “the reverse way”. 
Edit wrong link. Has to be this one 10/100/1000 MHz version. So at least the 100 MHz will talk to the B side. 
 

And will actually such a RJ-45 SFP transceiver work when B side is 100 MHz  and the opticalRendu is 1000 MHz. Normally not of cause, but maybe the use of a RJ-45 SFP transceiver can overcome that issue. 
 

Again, if expect working, is it still a better solution, and possible to explain why if it isn’t. 
 

 

 
Link to comment
8 hours ago, R1200CL said:

@JohnSwenson

If technical possible, can an optical version of the EtherRegen be made. 
Meaning exchange the RJ45 B side with a SFP cage ?

 

I think not, as the present design requires the B side to be 100 MHz, and since the reason for this request is to connect the opticalRendu, a 100 mH optical interface won’t work. (OpticalRendu requires 1000 MHz) 
 

So I ask again using a gigabit only RJ-45 SFP transceiver on the opticalRendu https://amzn.to/2XbBAJA be a better solution than using the EtherRegen “the reverse way”. 
Edit wrong link. Has to be this one 10/100/1000 MHz version. So at least the 100 MHz will talk to the B side. 
 

And will actually such a RJ-45 SFP transceiver work when B side is 100 MHz  and the opticalRendu is 1000 MHz. Normally not of cause, but maybe the use of a RJ-45 SFP transceiver can overcome that issue. 
 

Again, if expect working, is it still a better solution, and possible to explain why if it isn’t. 
 

 

 

Nope unfortunately that will not work. The SFP port on the opticalRendu is fixed for gigabit fiber mode. The 10/100/1000 RJ45 SFP modules use a completely different protocol which will not work with what is in the opticalRendu. The protocol chip in the opticalRendu has control pins that determine which way it operates and those are soldered to the board in the appropriate combination for gigabit fiber mode.

 

There ARE RJ45 SFP modules that use gigbit fiber mode, but they are just gigbit. There is a 100Mbit fiber protocol, but it is a different protocol, again both sides of the link have to be programmed for that protocol.

 

Again I don't see the problem of hooking the opticalRendu up to the SFP port and the network to the B side RJ45. It was designed specifically to do that.

 

John S.

Link to comment

Powering the EtherRegen with the LPS-1 with the original Meanwell supplied that has the ground shunt implemented. (Modified Meanwell)


Shall that ground pin in EtherRegen be used or not ?

 

Will your answer be same for LPS-1.2 ?

 

(Yes I know this has been answered, but sometimes an answer that requires a YES / NO, just gets to confusing 😀)

Link to comment
3 minutes ago, JohnSwenson said:

So if powering the EtherREGEN from an LPS-1.2 here is the table:

 

Connection                                                                             should use ground screw

 

If you use a linear PSU to supply it will it be the same result regarding recomendation of the use of grounding? 

 

My system will be SFP in from router, one rj45 on A-side to server, B-side to streamer and a linear PSU powering it (negative side not grounded by default).

Main system
TAD D1000mk2, TAD M2500mk2, TAD CE-1, Ansuz Mainz 8 C2, Ansuz Darkz D-TC, 
Qobuz Studio -> Roon ROCK on NUC -> Uptone etherREGEN -> dCS Network Bridge -> AES/EBU -> DAC
HD Plex 200W PSU (4 rail for ISP fiber, router, etherREGEN and NUC)
 
Second system
Qobuz Studio -> Devialet Silver Phantom, Devialet Tree
Link to comment
1 minute ago, octaviars said:

 

If you use a linear PSU to supply it will it be the same result regarding recomendation of the use of grounding? 

 

My system will be SFP in from router, one rj45 on A-side to server, B-side to streamer and a linear PSU powering it (negative side not grounded by default).

It depends on the linear supply. If the negative pin of the supply is NOT connected to the safety ground of the mains connection then follow the table. If the supply DOES have the negative output pin connected to the mains safety ground then you should never need the ground screw.

 

John S.

Link to comment
1 hour ago, JohnSwenson said:

The original LPS-1 will not power the EtherREGEN, it doesn't have enough power. The LPS-1.2 when set to 12V WILL power the EtherREGEN.

 

So if powering the EtherREGEN from an LPS-1.2 here is the table:

 

Connection                                                                             should use ground screw

 

SFP optical to B RJ45                                                           NO

Single A RJ45 to B RJ45                                                      NO

Multiple A RJ45          no audio endpoint                           NO

Multiple A RJ45         at least one audio endpoint            YES

 

The ground screw is to prevent leakage between A side RJ45 jacks. So the only time you should probably be using it is when you have multiple connections on A side RJ45 jacks and at least one is an audio endpoint. (I presume you are using the B side RJ45 for your best audio endpoint). Note this does not take care of any clocking issues, just the leakage. Both leakage and clocking are taken care of when crossing the moat.

 

John S.

Just to make it clear, when you say "connection should use the ground screw" you are referring to connecting the etherregen ground screw to the mains ground. Am I right?

Link to comment
4 hours ago, JAB said:

John S,

During development did you ever use a LPS to see if it made a sonic difference ?

Would you consider a package of the EtherRegen and lps 1.2 ,like you do with the iso Regen?


This is for Alex  at UpTone Audio to answer. John is the developed or designer of several nice products. 
 

I’ve noticed this is your first post. And I can’t answer for his (Alex) @Superdad, bundles options, but don’t expect it to happen with is low margins explained many times. But you can save some shipping cost probably purchasing both at the same time. 
 

If you’re after a LPS-1.2, look at the “Novel thread’ as it so happens austinpop is selling his, and just PM him. 
 

Welcome to the forum 😀

 

 

Link to comment
8 hours ago, JAB said:

During development did you ever use a LPS to see if it made a sonic difference ?

So far listening has been done only with early prototypes (where full isolation and all the good local regulator stuff was not even in place), but as it happens one of John's bench supplies is a JS-2 LPS.  So the real answer is same as it has been to-date: We still do not know if or by how much a nice LPS will make a sonic difference EtherREGEN users--versus the decent SMPS it will ship with at $625.

 

Quote

Would you consider a package of the EtherRegen and lps 1.2, like you do with the iso Regen?

I have certainly been thinking about offering a bundle. Probably not right at the launch though.

 

Would also need to figure out a different packaging/shipping scheme for an EtherREGEN/UltraCap LPS-1.2 bundle as they won't both fit into the blue box I designed (which is max size for Priority Mail Flat Rate Padded Poly Pouch and what the LPS1.2 and bundles with ISO REGEN ship in). 

Link to comment
4 hours ago, Superdad said:

PROGRESS UPDATE:

 

Well as pretty and near-done the pre-production boards looked (see https://audiophilestyle.com/forums/topic/38968-etherregen-we-are-getting-much-closer/?do=findComment&comment=973852), there are just too many hours under-the-microscope changes to make them viable for use in beta test or for us to risk going straight to production.  

 

Plus, John has decided to implement a complete new scheme for power-on sequencing of each main logic chip and the clocks. Certain parts need to be receiving a clock before they start, and some parts are best turned on after others to assure function in all situations.

We already have two very small microprocessors on the board. We will separately flash them during production with code for the clock synth and the switch chip on each side, so they will program those parts each time power is applied (then they turn themselves off to keep quiet).  There is plenty of extra code room in these processors, so John ran traces to reset pins on various parts and will write and tweak code to turn them on at just the right time. Quite elegant!

 

In addition, since this next (hopefully final) round of pre-production/beta boards are also for bench test, he decided to bring out to small via pads a variety of chip pins to make it easier for him to watch exactly what is going on between them. Most of the main logic parts we use have the tiniest of pins (a lot of them sort of tucked under), and getting a probe wire soldered to one--and then having it not break off--is an exercise in hours of frustration.

 

I found a few mechanical I/O parts movements and some interference stuff, so those got adjusted as well.  

Sent off the new board files and the very slightly revised BoM (Bill of Materials) to our board house this afternoon.  Should be able to get a very quick turn as all the parts have already been purchased (for more than 250 first units).  

Well, all but one. John realized he had left off (from the last boards) the terminating resistor for the external BNC clock. It only came up when I went to identify it for when we have a few boards made in production where the BNC jack and resistor are left off so we can accommodate special orders for 50 ohm clock input termination. I am glad he had left the resistor off because it gave me a chance to ask him to put it in an accessible spot and to make it larger than the grain-of-salt 0402 size we use for most resistors on the board.  I hate soldering 0402s by hand!

 

So while we are sad that the first pre-production/beta boards will just go in a scrap heap (a few $K investment), a lot of things, small and large, were learned from them, and implementing a proper, programmable chip reset sequence mechanism--versus having reset controllers everywhere--makes so much sense.  All the magic stuff--data and clocks flowing in the loops across the moat/isolators and into key logic parts and through the flip-flops--is working beautifully. (It should, because many prior months of work was already done!)

 

So really, aside from EtherREGEN being far more advanced than anything else out there, folks will really be getting a refined, second generation product right out of the gate! :D

 

 

Will keep you all posted.

Have a great weekend,

--Alex C.

------------------

 

P.S. For those following my stupid bike-fall saga:

I finally had X-rays done yesterday, and yes it turns out I did fracture three ribs (#4,5,6 if I recall; not too much of a surprise as I heard the cracks when I landed!).  Then the doc insisted on a CT scan to be sure I did not puncture or partially collapse a lung. (Why they felt they had to pump me with heavy duty pain drugs to do it I have no idea. I hate that stuff and it makes me sick.) Anyway, after 4 hours in the emergency room until 11:00 p.m. last night, they gave me the news that nothing is serious--beyond the serious pain I have when I move around.  Amazing how much we use or torso for everything! They kept asking me my pain "level" on 1-10 scale.  It's zero, and then I make some normal movement and its an instant 11! :o

I'm figuring it out (sleeping on a pad on the floor is the only bearable thing), and I know it will get better in time.  

So many thanks for your words of sympathy--but no more. It's music--and EtherREGEN time! B|

 

Bunch of lovely freaks and brains! 

Does the Beta test involve actual user testing? 

Link to comment
2 hours ago, Bernstein said:

Bunch of lovely freaks and brains! 

x-D

 

2 hours ago, Bernstein said:

Does the Beta test involve actual user testing? 

It will once we have a small batch of boards which are fully functional.  User testing will be mostly to look for weird corner issues with people who have somewhat complex networks or applications—in case we need to adjust our programming of the main Ethernet switch chip (though it is field updatable via the microUSB port on the board).

Link to comment
7 hours ago, Superdad said:

x-D

 

It will once we have a small batch of boards which are fully functional.  User testing will be mostly to look for weird corner issues with people who have somewhat complex networks or applications—in case we need to adjust our programming of the main Ethernet switch chip (though it is field updatable via the microUSB port on the board).


Hmmm...my network is not that complex I think, but maybe having the EtherRegen in Germany could lead to some dizziness in its behaviour (due to beer and weisswurst spilled over it :) ) which could lead  eventually to better SW and waterproofing...

Link to comment
On 8/1/2019 at 7:05 PM, JohnSwenson said:

You

You can certainly use the SFP port to connect optically to a opticalRendu. Connect the B RJ45 to the network. This will work great as long as you don't connect anything else to the A ports. If you do connect other devices then their clock signatures wind up on the A side ground plane and will affect what is going out on the SFP port. The clock signatures from the B side are blocked but anything else on the A side will affect the SFP port.

 

So yes you certainly can use the EtherREGEN to optically connect to an opticalRendu, don't use any of the other A side ports, just B side RJ45 to A side SFP.

 

John S.

 

I have the opticalRendu with the generic optical switch they sold as a package in the beginning. It was my intention to replace that switch with the EtherREGEN but I need to plug my NAS and Mac Mini into it. I think that means I would need to have one of them plugged into the A side along with the opticalRendu in the SFP port and you don’t recommend that?

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...