Best audiophile switch

[mxk116]

Coming soon at long last! New info, price, and pics just posted here.

Terrific news, thanks for the update.

 

[Superdad]

I really like the JS-2 so have high hopes for the JS-4. The internal PCB design looks radically different. Lots of little caps instead of one-two big ones.

Actually, those banks of super-capacitors--which sit between the second and third stage of regulators--are 1Farad each, x 7 per channel, in series, so total of 143,000µF per channel.
Whereas in the JS-2 we have just a pair of 33,000µF caps in parallel--so 66,000µF total--feeding both channels.
So vastly more filter capacitance in the JS-4, though that's really not what makes it special.
[Keeping modest capacitance at the first stage--coming off the diodes--greatly reduces charge pluses and ringing. And highly optimizing capacitance networks at the input and output of the final LT3045s in the third stage (exceptional derating curves of banks of 22uF Murata X7R) are just some of the key ingredients to our recipe. ]

 

[Republicoftexas69]

Coming soon at long last! New info, price, and pics just posted here.

When is soon Superdad? I really want one

 

[Superdad]

When is soon Superdad? I really want one

We are assembling chassis now and will prep and mount the transformers next week. But the first 110 units of the big circuit boards still have 2.5 weeks in assembly--and then get shipped down to us from our PCB fab/assembly partner in Canada. Hopefully no slow down in customs and hopefully the tariffs don't wreck our modest pricing model...

Safe bet that first JS-4s will begin shipping to our clients at the end of June for sure.

 

[tony22]

We are assembling chassis now and will prep and mount the transformers next week. But the first 110 units of the big circuit boards still have 2.5 weeks in assembly--and then get shipped down to us from our PCB fab/assembly partner in Canada. Hopefully no slow down in customs and hopefully the tariffs don't wreck our modest pricing model...

Safe bet that first JS-4s will begin shipping to our clients at the end of June for sure.

What about the ER2?

 

[Superdad]

What about the ER2?

Thanks for asking!
As you may have read, we have been working hard in development of EtherREGEN Gen2 for almost 2 years—utilizing more readily available parts and incorporating a number of technical advancements. These include Gigabit capability on both the ‘A’ and ‘B’ sides of our unique active-differential-isolation moat, even lower jitter clock synthesizers, an exceptional new PHY chip, and the world’s best sine-to-square wave converter (benefiting those who chose to pair with an external 10MHz reference clock). The new version will even include a second SFP cage—on the ‘B’ side—to improve performance for users of optical endpoints.

Unfortunately, due to documentation and programming issues with the different main switch chip we have been developing with—and now forced to abandon and pivot back to the expensive one as used in the original EtherREGEN—we now expect first shipment of EtherREGEN Gen2 will not take place until about August/September 2025.
Price is not yet firmly set, but like the original, EtherREGEN Gen2 will be priced incredibly reasonably for the advanced tech and high cost parts utilized. I always planned on it being in the mid-$700 range, certainly less than $800.
But now with the pivot back to the expensive main switch chip (the 1/4 priced one we now abandoned would have helped offset some of the above costly feature/perfomance additions) and the uncertainty of tariffs (our aluminum cases come from Japan and the boards are assembled in Canada), I am not at all certain where the final price will land.

 

[tony22]

Thanks. It sounds like (hopefully) the price will fall in line with something like the recently released LHY AS-6. Those other guys haven’t been standing still, @Superdad!

 

[Republicoftexas69]

Thanks. It sounds like (hopefully) the price will fall in line with something like the recently released LHY AS-6. Those other guys haven’t been standing still, @Superdad!

The other guys are still not on par with the current ER. Every LHY I have tried has been sent back.

 

[tony22]

The other guys are still not on par with the current ER. Every LHY I have tried has been sent back.

Others have had opposite results, which is no surprise I suppose. But waiting for the ER2 has kept me from venturing further into the morass of audiocentric network switches.

 

[treitz3]

Patiently waiting, Alex. Looking forward to it, along with what it brings to the table.

Tom

 

[Republicoftexas69]

Others have had opposite results, which is no surprise I suppose. But waiting for the ER2 has kept me from venturing further into the morass of audiocentric network switches.

Agreed

 

[agisthos]

@Superdad I thought you had done an update a while back where you finally solved the issues with the switch chip that had taken over a year to figure out? Did more stuff come up and that was enough for you to ditch it?

 

[Superdad]

@Superdad I thought you had done an update a while back where you finally solved the issues with the switch chip that had taken over a year to figure out? Did more stuff come up and that was enough for you to ditch it?

You nailed it!
For a while back in September we had the new-main-switch-SGMII>"moat">flip-flop>new-PHY>UTP port working. Though it was going to be a doodle from there onwards to a "unified" design test board and then a small pre-production/beta run of the final.

But one more test of what should have been easy: Putting the switch chip SGMII port into fiber mode to feed across the "moat", into the flip-flop, and then directly to what will be the new 'B' side SFP cage. That's when everything seemed to go sideways, including "bricking" of several parts on our test board and months of John Swenson immersed in the poor documentation and register bowels of the main switch chip. Even on the backup test board (we always run two) things are off and even SGMII>PHY interface is no longer working right.

Here is a pic of about where we left off (actually this board omits the "moat"and ultra-low-jitter differential reclocking flip-flops referenced above; it's all just for functional Ethernet port tests):


So where are we now? Aside from our both having fewer hairs and less patience, we have made some decisions and John has designed a new smaller test PCB (the bare 4-layer board(s) for it are to arrive this coming Monday).
The decision is that if we can't get the new switch chip--which we have now wasted about 18 months fighting with--working in all required modes, I will write off the $6K of stock of that part and we will pivot back to the original Microchip switch chip as used in EtherREGEN "gen1." It is larger, 4x the price, and consumes more current, but it works fine and we know how to use it. (Still not in the best of stock quantities at the distributors but I can manage.) [I was looking forward to ER Gen2 running a LOT cooler, but the original chip does run hotter than the one we might abandon. Lots of other new parts in Gen2 do run a lot cooler, and the case is going to be 50mm longer to accommodate the 'B' side SFP cage, so we can spread out a bit more. So EtherREGEN Gen2 will certainly still run cooler than the original.]

It is not a big deal to use our original switch chip since all the performance improvements are coming from the really great new low-jitter PHY, the much better (and more expensive) clock synthesizers we chose, the awesome sine>square converter chip (which which both the internal and external clock references will run through) and a number of power network improvements we are making (Yes, there are now LDO regs better than LT3045, in that are equal or better on noise and PSRR and are 1A devices; plus they cost less and being 1A we need fewer--do you have any idea how much 13 pcs. of LT3045 cost theses days?!)

Next week's new small test board is interesting in that it really just has a couple of PHY chips, one RJ45 several SFP cages, and lots of header pins to both connect SGMII inputs and outputs and to monitor packets and raw bits. John has reprogrammed an original EtherREGEN so that its switch chip goes into SGMII mode and puts that (instead of the essentially identical fiber-mode) out through its SFP cage--as a means of testing SGMII from the Microchip switch to our new PHY.
There is a chance that some of the trouble he had lately is with the resetting of the PHY. So he revised some things. Depending upon what he discovers, there is even a chance that frustrating main switch chip can be used after all (I'd like to save that $6K and to not have all the time for nothing). That's why this new small board will have various means of getting SGMII in and out--so he can hook back up to the above big boards well.

So perhaps you will see photos in a few weeks of several boards all lashed together into really messy nest!

Sorry for the long post. Some people here enjoy peeking into how the sausage is made...
I promise we will get there! (I still have $80K worth of parts for the model sitting on our shelves so we can't give up.)

Thanks all for your patience and enthusiasm.
--Alex C.