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

[matthias]

On 11/30/2018 at 6:12 AM, Superdad said:

 

100Mbps is more than twice the data rate of even DSD1024!  So no problem. Know of any DACs that run that?  

 

The new Holo Audio Spring2 should be able.

 

Matt

[matthias]

4 hours ago, vortecjr said:

I prepared this diagram (approved by the heads of state) to help explain how to properly hook up a Rendu to an EtherREGEN. 

 

 

IMO, you can also connect an opticalRendu via the B side with copper SFP.

 

Matt

[matthias]

1 hour ago, jos said:

In my humble opinion you don’t need optical anymore with the promises of the new ER. Don’t make your chain over complicated. 

 

Yes, from the Uptone HP:

 

"Use of optical into—or out of—EtherREGEN is not required to obtain best performance. The whole point of all the isolation and re-clocking techniques we are using (in dual domains) is to make most all of the upstream stuff not matter. Other than the complete galvanic isolation that optical offers going into the EtherREGEN (which we duplicate going across the moat), there is no advantage to running optical into it.

If you happen to have a router with an optical port and have already run many feet of optical cable to your music system rack, then sure, use the optical port of the EtherREGEN.  But don't go out of your way to run optical if you do not already have it. And of course for the few people who have an endpoint with optical input (the excellent Sonore opticalRendu comes to mind), one can “turn around” the EtherREGEN and feed that DAC-connected endpoint from the optical cage."

 

Matt

[matthias]

On 3/24/2019 at 8:42 PM, Superdad said:

 

While fully addressing the technical issues of what you envision would take this EtherREGEN thread off-track, I'll point out a couple of factors, because what might sound conceptually simple is anything but :

 

a) USB is a bi-directional protocol--data goes back and forth on a single pair of wires (and no, it's not one wire one way, the other wire the other).  All digital isolator chips (regardless of their speed or technology--optical, capacitive, RF, magneto-resistive, etc.) are ONE-WAY devices for each bit (yes, they are often packaged with multiple bit-channels, but you choose which channels are going which way and that's set.  

So the only way to put a USB signal across an isolator is to process it into a format where data going upstream is not on the same line as data going downstream.  The only integrated chip capable of doing this at USB2.0 high-speed is the Silanna ICE08USB--the part we use in the ISO REGEN.  Beyond that one can accomplish this with two FPGAs and a bunch of code (as done in the Intona-brand isolator box).  [There is another, much more elegant method than either of the above, but I'm sure as heck not going to discuss it here.]

 

b) Since you brought up optical in the context of this Ethernet-oriented thread, I will assume your suggestion is in the realm of optical transmission as produced by Ethernet optical transceivers, such as what is in an SFP module..  (Of course if you simply meant simple opto-isolator chips, then see "a)" above--and BTW, opto-isolators add much more jitter than some of the better digital isolator methods such as GMR.)

Optical Ethernet SFP transceivers are interesting pieces, and there are some possibilities to use them for transmitting other than Ethernet--with remarkably low jitter (in fact I am pretty sure that's what MSB is doing with their new ProUSB modules--proprietary to their DACs--though they are making the USB side appear as a full UAC2 audio endpoint device to the computer).  

John and I recently spoke about using optical SFP transceivers in a non-conventional way to produce a fantastically well-clocked, isolated USB extender--your DAC could be miles away!--but there are issues about protocol at each end and also about clock synchronization.  Suffice it to say it would not be a simple affair--and it would not be bus-powered!

 

So you can see there are no snap-together solutions to high performance USB isolation.  Heck there weren't any for Ethernet--at least not the sort of active isolation/separated clock-domain we are going for with the EtherREGEN--so we have had to get really creative.

 

Hope I have answered your questions.

Cheers,

--Alex C.

 

Hi Superdad,

now more than four years later what do you think about releasing such a device similar to MSB ProUSB where the first stage is USB bus powered.

 

I think it could be a killer product.

 

Matt

 

[matthias]

48 minutes ago, MarkusBarkus said:

Emile Bok at Taiko has been talking about new protocol he is working on the eliminate USB completely, which he asserts is the primary limiting factor to truly improved performance. 

I am a great admirer of Emile's work but this upcoming protocol should not be limited to a Taiko Extreme.

It should be possible to use it with an adapter from a Thunderbolt output of a MBP for example and for the DAC you would need a receiver board inside the DAC. I really hope that Emile offers this technique to other manufacturers.

 

Matt

[matthias]

11 hours ago, JohnSwenson said:

Strangely enough I was just thinking about it this morning. We need a differential interface and there are very few connectors for this. The best thing I could come up with is the SFP cage, good quality SFP+ modules and cables are not very expensive. Even if not needed I would still go with SFP+ due to significantly lower jitter in the optical/electrical conversions. The SFP standard on the electrical side is fully differential already. Since the SFP includes two independent opposite direction lanes we can do a format where the source sends digital data and the receiver sends a clock back to the source. The good clock is at the receiver side and the source is slaved to it. Just build this concept into the design from the beginning so we don't have to deal with PLLs etc at the receiver side. Modern ECC techniques could be used for extremely high data integrity. We have to make sure that this stays simple and doesn't turn into USB, it is just digital audio from one box to another, no control or complex protocols. Control can be done with traditional networking techniques but the actual digital audio goes over this simple yet very high quality link.

 

John S.

Great post, John, thank you.

I think there is a bigger market for USB devices than for Ethernet devices.

If you go via WiFi from your ISP router to your endpoint you do not need an EtherRegen, but you need an USB solution between endpoint and DAC. So hoping for a new Uptone Audio USB device.

 

Matt