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

[Superdad]

[EDIT: The original first post of this long thread had outdated information and so is deleted.]

 Development of EtherREGEN is complete.  The web page for it is up (UpTone Audio EtherREGEN), and a new thread for the launch has been started. Please see:

 

 

[Superdad]

2 minutes ago, rickca said:

I guess you mean 10/100 Mbps.  I hope it's OK to ask for a clarification on this point.  Are the SFP fiber ethernet cage and the isolated ethernet output jack two separate ports?  If so, I guess I don't understand the role of the SFP cage.

 

Sorry, yes, 10/100 Mbps.

And yes, the SFP cage is separate from the single special isolated port.

It is there because we could do it.

And some people like to run fiber from their other switch or gear.  But no DAC or streamer--that I am aware of--currently has optical fiber Ethernet input.

[Superdad]

On 3/25/2018 at 8:47 AM, austinpop said:

Just curious - did John and you consider using XO or even OCXO? Was that a cost tradeoff?

Of course, I can wait until you're ready to say more about the clock design.

 

Well the clocking in our switch is much more sophisticated than is typical as:

a) we have to pass the clock back across the isolation moat;

b) more than one frequency is used.

That is all I want to say about the clocking architecture at this time.

 

As for use of an OCXO: As John has written about this, cheap OCXOs are no better (often worse) than a good XO--one needs to spend hundreds (at OEM pricing) to get a really good OCXO.  And for data clocks (for USB, Ethernet, or processors), such is total overkill.  Save really great clocks for the DAC audio clock(s).  Even the clock boards in the SOtM units don't use a OCXO as a reference for the SiLabs synthesizer chip. 

Besides, this switch--and the issues we are tackling--is about far more than just clocking...

 

3 minutes ago, Albrecht said:

I so can't wait for this!!!

 

That makes quite a few of us! 

[Superdad]

33 minutes ago, austinpop said:

If you guys can deliver that or better at a fraction of the cost, you'll have a major winner!

 

Well you already know that great "bang-for-the-buck" and delivering real improvements are what UpTone is all about! 

 

So that's the goal with the new-thinking switch as well.  

We have a philosophy of doing all the important things well in a circuit--and avoiding excessive (expensive) focus on any one aspect.  (We see that sort of thing with a lot of digital products.)  John would rather get to the root of the issues and do things right from the ground up.

[Superdad]

13 minutes ago, Celts88 said:

Any chance of the 2 x Superclean ports being a goer, or is that a no go?

 

I decline to state why, but that is a no-go.  Would require a second Ethernet switch chip.  If I say anything more we would begin to give away competitive details. Sorry.

Since the price of the product will be as reasonable as we can make it, so folks who really need 2 of the super-clean outputs may simply purchase two units.

[Superdad]

On 4/2/2018 at 9:37 PM, charlesphoto said:

Will be curious if this switch negates the need for a fiber optical bridge like many of us have pre-renderer/dac. Would be nice if it did. 

 

Well that is what we are expecting and why we are doing this.  And we expect the clocking, signal integrity, and isolation to do more.

 

As far as I have been able to tell, the only thing the FMCs (fiber media converters) have ever brought to the table is blockage of leakage (assuming one uses a good PS on the downstream unit to avoid reintroduction of leakage).  Those FMCs are dirt cheap and not in any other way improving the signal.

 

I do want to tread lightly here as John is still determining which of the key aspects of our design are going to be most relevant (hence the measurement/research work he is now doing on upstream clock markers finding their way through to the DAC output).

 

We are only going to go through with EtherREGEN if it provides worthwhile benefits.  As much as our loyal clients would love to throw more money at us, our reputation is based on delivering products that produce real and audible improvements.  Don't take this as skepticism or lack of enthusiasm on my part, I just don't intend to put the cart before the horse with claims in advance of tests.  One or two steps at a time! 

 

--Alex C.

 

[Superdad]

51 minutes ago, R1200CL said:

 

Shall this be an active or passive device ?

 

If active, you purchase two EtherRegen. If passive, I’m not sure if it requires a gigabyte port in with 8 wires active, as a spitter can only use 4 wires and thus is either 10 or 100. 

 

 

Ø

 

Just to be clear gents, those so-called "splitters" do not create any new ports.  They just allow the use of one cable for two separate ports.  You have to put one of these at both ends of the cable--and have two Ethernet ports to connect to.  

Remember, just one IP address per port. 

[Superdad]

6 minutes ago, R1200CL said:

Some old stuff......

 

Any particular reason you are bringing up that discontinued project from 4 years ago?  

It would have worked and been great, but there is an undue amount of software coding involved, and our thinking--for other future projects--has evolved.

But that project has zero connection/relevance to our Ethernet switch.  

[Superdad]

2 hours ago, Theobetley said:

So Superdad going to have anything at Axpona on this?

 

I did CES (and sometimes the Stereophile shows and Munich) for 10 years with Hovland Company.  Those were semi-worthwhile back then because we met with dealers and distributors there.

 

But with UpTone’s direct sales model—to people all over the world—small regional end-user shows like Axpona can be a colossal waste of time and money.  What, would we A/B a network switch in a hotel room?  Or sit at a booth display so people can walk up and touch a simple box that looks just like our UltraCap units?

 

While I’d enjoy meeting a bunch of our Midwest customers and chatting, it’s not like I’d have a staff of people to help out and trade off with all weekend.  And I would not even get to see the show myself.  So I’d rather stay home and keep the $10K of travel and show cost, and keep the prices of our products as low as we can.

 

Our products are sold with a 30-day, money-back guarantee (with no restock fee), and always the best way for someone to decide is to try in their own system!

 

All that said, I frequently will loan or sell products to other companies who wish to use them at a show—to improve the SQ in their room.  However, AXPONA is next weekend and EtherREGEN is not going to enter production until sometime this summer, so don’t expect to spot one in Chicago next week. 

[Superdad]

6 hours ago, R1200CL said:

Personally I rather woukd like to see first number start with the digit 2.

 

What, so you’d like it to be $2,000? 

Seriously though, $200 would not even cover our parts costs (and that’s with a 250 unit run).

[Superdad]

3 hours ago, Solstice380 said:

There is a lot of room below the mentioned hope of <$500.  If it’s too close to that I might be better off getting an AQVOX or other.  Possibly not the nth degree of SQ as the single clean port but I really want to feed the NAAs for both the Lampi and the dCS.

 

Well you certainly will be able to feed both NAAs from our switch.  ALL six ports on the EtherREGEN will have the same ultra-low phase noise clocking.  And all ports will be blocked from external leakage and EMI. So right there all ports will be equal to what you get from something like the $500 AQVOX modified D-Link switch.

It is just that the is only one super-special port (where a lot of extra chips and $ are going) which will be highly isolated—using techniques that have not previously be applied to Ethernet.

So you get more with our piece, not less.  But the proof will be in the pudding as they say.  It is too early to make claims, but we do want the concept to be understood.  We are doing something very different—starting from scratch. 

[Superdad]

15 minutes ago, thyname said:

John: are you saying that we don't need to use your ground plug "device" with the new etherRegen?

But we will need to use it if using an LPS-1.2 to power it instead of the UASMPS that comes with it?

 

Well, as he and I both indicated, it remains to be seen if a high-quality LPS is even needed for our new switch. (Wow, why should I say that? Don't I want to sell more UltraCap supplies? ).

 

But in case people do--or if they use some other PS that is not ground-shunted--then yes, one will need to ground the EtherREGEN.  However, as John mentioned, we plan for the unit to have a ground screw (or socket)--so that John's double-barrel-connector doohickey is not required. 

 

[Superdad]

55 minutes ago, thatguy said:

Thanks for the reply, to confirm: you are saying that any clock that connects to the circuit board of the DAC chip is creating "phase noise" of the DAC's clock.

 

Thanks your interest.

I mean no offense, but since you are new around here, and based on the phrasing of your questions, please allow me to suggest you do a bit more reading here at CA on the topic of clocks, phase noise, and DAC architecture.  Some of this stuff may then start to make a bit more sense.

And then you will also see where John is saying we are currently limited in the proofs of how upstream clock signatures can migrate into the workings of the DAC itself.  

 

This is an area of intense research for us (I speak about it a bit in the very first post of this thread), and if the project (a series of boards, a high-speed 32-bit ADC, a jitter marker injected, etc.) is successful, it will likely call into question a number of conventional assumptions and attract a lot of interest from other digital engineers.

 

But we really don’t want to get into any of this here and now.  

In fact, I think this thread has run its course in that there really are no more questions to answer until much closer to the release of the EtherREGEN.

And when John’s upstream clock signature/blocking measurement project begins to bear fruit, we will most certainly start a new topic about that.

 

—Alex C.

[Superdad]

Five petty, personal argumentative posts just deleted.  Knock it off or I’ll simply lock this thread. 

[Superdad]

1 hour ago, hopkins said:

Anyway, the part I am curious about now is how more findings (experimentation) by John will change the solution you design in your products. 

 

It may affect the particular isolation technique we us.

Perhaps in a few weeks, when John has finished assembling his elaborate, mult-sectioned “clock-blocking/jitter-marker injection” test board set up, he might post a photo of it (not too hi-res as we don’t want to give away the farm) so you folks can see the lengths we are going to to develop a truly efficacious solution.

[Superdad]

5 hours ago, Blake said:

It seems that the EtherRegen may not benefit from an upgraded LPS.

 

That is not a certainty. 

 

5 hours ago, Blake said:

Therefore, I no longer have a place for my  LPS-1.  Before I put the LPS-1 up for sale, are you at liberty to say whether there might be some other upcoming Uptone products in the pipeline where the LPS-1 could be used?  If so, I'll keep the LPS-1 rather than sell it.

 

Sorry, can't say...

 

5 hours ago, Blake said:

I am not asking for any details of any upcoming device, simply if any upcoming device will benefit from a quality LPS and if so, if the LPS-1 could be used to power such device? 

 

A great power supply is always a handy thing to have about.  You never know what use you might find for it!

[Superdad]

18 minutes ago, Bricki said:

Just wondering if the new EtherRegen will be able to be powered by the lps1?

 

7 volts may be the lower limit for the switch, but we are trying to insure that an original LPS-1 could power the EtherREGEN.

We know that there are a lot of LPS-1 units out there (as so many of you have upgraded to the LPS-1.2, many thanks!), so your question has been on our minds...  

[Superdad]

14 minutes ago, audiojerry said:

Yes, the PS Audio DirectStream Jr has an ethenet/LAN input to its built-in network bridge. The network bridge also has an hdmi input for i2s, which supposedly is superior because it separates the clock from the audio signal.  I don't no why, but PS recommends against using wireless ethernet.

 

Hi Jerry:

 

Well the Ethernet input is just one of the 6 inputs available on your DirectStream Jr.:

 

The I2S (via LVDS on HDMI) input is not technically part of the "Bridge" circuity at all.  Without going too far into the complexities of formats and such, I'll mention that I2S is in some sense the most "native" input--in that ultimately all other inputs are producing an I2S output to feed the DAC circuitry itself.  So a USB input is really a USB>I2S circuit, and the Ethernet input is Ethernet>I2S, and so on (though it's a bit more complicated for AES, S/PDIF, and Toslink)--and I am oversimplifying in any case.

 

The point being that yes, depending upon how the input circuits--and all that follows--are executed, different inputs may deliver slightly different sonic results.  Even use of the I2S input introduces the vagaries of whatever I2S source one uses (typically an outboard USB>I2S converter), and then the audio-rate clocks in that source become rather important (since few DACs feed their master clock out their HDMI connector to slave the source).

 

14 minutes ago, audiojerry said:

I'm dubious that sound quality using ethernet to the bridge would improve what I'm already getting from USB with the IR and LPS-1 (or LPS-1.2). If ethernet to a bridge is the superior approach, then I'm hesitant to get the LPS 1.2. 

 

One can not generalize.  You just need to try the various inputs and see hear! 

Of course current Ethernet inputs on DACs generally impose restrictions on how you get the music to them.  It is not as if our computer operating systems have built-in support to see DACs connected on our LANs.  That's where DLNA, Roon, etc. come in.

 

 

Have a great weekend everyone.  I about done with the flu, and UltraCap LPS-1.2 boards are at last going into their cases and being prepared for shipments around the world!  So I am in MUCH better sprits that a week ago! 

[Superdad]

3 minutes ago, R1200CL said:

@JohnSwenson

How is the progress ? ?

 

Just fine, thanks.  We are not going to share details of exactly where we are in development.  Still looking forward to a summer release as planned.  

[Superdad]

2 minutes ago, rickca said:

@JohnSwenson have you built the measurement equipment you need for your research yet?

 

He is getting closer.  The bare PCBs for the "clock cascade/blocking test/validation board" are on hand, but there are so many parts for the board that he decided to finish building his parts placing robot.  Below is an early pic.  It is MUCH further along now; control board wiring, camera mounting, vacuum pick-up, and computer connections are all being finished now.  I can't wait until it is done so he can post a video of it in action.

 

 

[Superdad]

1 hour ago, mansr said:

The question in my mind is, what's wrong with commercially available pick and place machines?

 

Other than price ($50K+ when fully configured) there is nothing wrong with small, short run commercial placers (which are nothing compared the giant full production units at our board house) like the one in your video.  John just needs a unit for prototype boards.  

 

It can take many hours to hand-place a 100+ tiny parts—with high risk of error.  And while we often have simply given the files and BoM to our board house for prototypes, that typically will run us $2-3K—for pieces that will be tossed on the heap in a few weeks time.

 

This DIY kit placer (LitePlacer.com) from Finland is just $2K, and the support John has received from them (including updated parts, spare part sent FedEx, etc.) has been extraordinary.

 

Plus, John already has a countertop vapor-phase reflow oven that does beautiful soldering—even with QFN and BGA parts. 

[Superdad]

We know that the summer and fall has passed with not much update on the EtherREGEN.  John got sidetracked a bit with another engineering project, though work continued on this one and we are at last getting very close.

 

Development has been proceeding nicely. This week we receive our third round of PCBs to conduct the next set of tests on. John did make a topology change a few weeks ago—for greater overall performance of all ports—and this necessitated addition of a second special switch chip (and it took me two weeks to get through the NDA process to obtain the full datasheet for it).

So yes, we are a some months behind schedule—I was originally expecting we would begin shipments this month. But given how every technical aspect of EtherREGEN will be so completely different from any other switch presently offered, we thing it will be worth the wait. We intend to present a block diagram and some explanations closer to the launch (which we are pretty confident will be February). 

[Superdad]

2 hours ago, Bruce Orr said:

Sorry, if I missed this, Alex. Will your EtherRegen have a Reference clock input port?

 

Yes, it will, though it will already have built in what is about the lowest phase-noise production XO on the market, the Crystek CCHD-575.

As important is how clocking is handled.  Because EtherREGEN will have a true digital isolation "moat"--with expensive high-speed digital isolators and differential flip-flops (as opposed to just Ethernet transformers modules for isolation), there has to be separate clocking on each side.

There are 5 clocks total (at 3 different frequencies), so we use the excellent (and again expensive and hard to get) Silicon Labs clock synthesizer (actually 2 of them).  Since those each have 4 outputs and 2 inputs, we will use one input for the CCHD-575 and program the other input to accept an external reference 10MHz clock if people wish to get extreme. 

 

All that said, excellent clocking is but one aspect of what will make this switch special.  A whole lot else will contribute to the performance.  As we get closer, we will likely publish a block diagram to explain a bit more about the topology of this unique switch.

[Superdad]

1 hour ago, tboooe said:

Hey Alex...how about accepting Pre-orders perhaps providing a little discount for early adopters!  I am afraid by Feb I will be out of money because of xmas, property tax, etc...in other words take me money now!!!!!!!

 

Sorry Tommy, we can't do that.  

a) It goes completely against my business ethics to accept money in advance for a product that does not yet have production PCBs and cases in production with firm delivery dates;

 

b) We would have to pay taxes this year for a product that will ship in 2019. 

 

c) We have not added up the final BoM (Bill of Materials) yet, so no final pricing has been set yet.  

We will do our best to keep the price very competitive (and I know it will be well under a rather pricey competitor), but there are a LOT of costly chips in this thing.  At last count I think were were up to 10 LT3042/45 voltage regulators, and those account for just half the power supply stuff.   There are quite a number of individual parts that themselves cost $10-15 each.  It is adding up quick! 

[Superdad]

2 minutes ago, auricgoldfinger said:

Will you be offering the option of a 50-ohm master clock connector for those of us who are locked into that impedance?

 

Well I might start a poll because it seems like half of you want 50 ohms and the other half 75 ohms.  We can do either--but not both.  It is simply a single tiny surface mount resistor change--and either the 50 or 75 ohm BNC.  But it is not cost effective to run both versions of the board--and rear panel labels (then get stuck with more of one versus the other), especially since 90%+ of buyers will never even be using it with an external clock.