Superdad

@Iving surely does not live in Idaho; In fact I think he lives in Europe, so odds are slim that he will get the American references (to the terrific Gus Van Sant film, to the City of Night book it was based on, or to the B-52's song). But I did...

Where please?

No way. Probably not going to happen.

Darn, I was too slow!🤣

Not sure why you would want to have one of those capacitor banks between your battery supply (with 200mA LT3042 and pass transistor to make it a 2A supply) and your FMC. Compared to a LT3042 (though real world effect of the output transistor LHY uses is unknown), the Unisonic UZ1085 3A voltage regulator at the output of those supercap boxes has horrible PSRR and 15,000 times more noise. Not to mention that the output impedance is probably very high. Just sayin'...

Life happened. Plus I was slow in doing the chassis CAD work (my longtime industrial designer/architect friend who usually does this stuff for us had health issues and needed to bow out). And we decided to give the JS-4 a ton of longterm torture testing. It is a sophisticated and costly design and we want to be 100% confident of reliability. CAD work is now done. Laying out all the back panel letting for silkscreen this week. Then a couple months wait for chassis production once we submit all the CAD milling and silkscreen artwork files. All other parts are here. Yes, JS-4 uses the 4-pole Neutrik SpeakON jacks, so NL4FX (as well as other 4-pole SpeakON models) are compatible. 4-pole is most convenient for building star-quad DC cables; Plus, if you look you'll see that Neutrik has been phasing out their 2-pole SpeakON parts--it's mostly all 4-pole and 8-pole now.

Yes, that will be the input voltage range for EtherREGEN Gen2. We expect the current draw to be about the same as it for the original model (approx. 9~10 watts regardless of input voltage).

FYI: Spoke to him and problem is solved. (Likely just an issue with the network feed cable into the 'A' side.) He is happily playing music now.

If you do not get an amber LED flashing at above-right of the 'A' side EtherREGEN port you have your router fed into (amber for Gigabit or green at top-left LED for 100Mbps), then the EtherREGEN is not receiving a good network feed to pass to your Grimm on the 'B' side. So please focus on getting that upstream side working. Be sure that the Clock switch was set to INT prior to applying power to the EtherREGEN. We also received your direct inquiry and I'll phone you in a little while. --Alex C.

Please understand that a direct attach cable is simply two 100-Ohm shielded twin axial cables terminated to a nearly bare PCB which is fitted into an SFP shell. Other than whatever maximum data rate bandwidth it is tested and certified to—and of course the gear it is connected to—there is nothing about such DACs that dictates or limits connection speed. So yes, a “25G” cable can pass 1G just fine.

Received the following e-mail from a new REF10 Nano owner this morning. And he does own a popular clock from another firm. (Reprinted with his permission.)

There are not any optical fiber elements contained in the type of RJ45 SFP transceiver module you linked to. Those simply take the 1000Base-X (fiber mode) SGMII differential pairs—at the back of the SFP cage coming from the switch chip—and run them through a PHY chip to produce the UTP for the transformers and RJ45 jack. But since the signal from the cage does not include a clock and all PHYs require one, those modules also have to have a 25MHz clock—as well as multiple voltage regulators to take the 3.3V feed and supply other voltages to the PHY and clock (usually just a cheap crystal). It is all far from ideal to have all that stuff crammed into an SFP form factor. And while Finisar is a quality brand, there are still a lot of compromises to build at low cost.

FYI: UpTone is now stocking the technically excellent MUTEC Prime Select Cables (PSC) which are optimized for square wave clocks. We have them in both 75-Ohm and 50-Ohm versions--present stock just in the 0.75 meter length, we can quickly bring in any of their other longer (1.0m, 1.5m, 2.0m) or shorter (0.5m) upon request. $155 for the 75-Ohm, $205 for the 50-Ohm. https://uptoneaudio.com/products/mutec-psc-clock-cable-bnc Please don't ask me how they compare sonically to the various crazy-expensive voodoo cables many of you are playing with. I have no idea what you will hear but I do know that these are properly engineered and precision fabricated cables for reference clock usage.

We have never required coded modules.

.

Well that’s simply the standards for SFP+ interfaces. So yes. The signals at the connector in the back of the ‘A’ side SFP cage of EtherREGEN Gen2 will come from an exceptional new PHY chip which is interfaced to an Intel switch chip via RGMII. The ‘B’ side SFP cage will driven directly from a special high-speed, ultra-low-jitter differential reclocking flip-flop, whose signals are coming from high speed differential digital isolators sitting across our PCB’s “moat”; those signals are coming from the SGMII port of the Intel switch chip situated in the ‘A’ side domain. When the ‘B’ side SFP cage sees an active connection at power-on, the SGMII>UTP PHY on the ‘B’ side will be turned off, the signals will be switched to route to the SFP cage, and the main switch chip will have its SGMII port go into 1000Base-X “fiber mode.” (Fiber mode and SGMII are essentially the same signals with some small differences, just relating to speed negotiation if I recall correctly. It all will work just fine. One can install SFP or SFP+ modules, but our SFP ports are 1Gbe-only. So whatever is connected at the other end must also be 1Gbe (there is no speed intelligence in SFP/SFP+ modules; there’s just a pull-down resistor that tells a 10Gbe host device that the module can handle that speed). Hope that addresses your question. —Alex C.

GREAT NEWS!! As of this past Monday, the new main switch chip on the latest EtherREGEN Gen2 subset test board is--at long last--fully talking to our specially chosen new PHY chip via SGMII. (That is the mode of the data we are putting over our differential isolators with Gen2--so that we can offer Gigabit RJ45 and SFP connections on the 'B' side.) This was a very hard won battle on @JohnSwenson's part, and due to weaknesses in the documentation of the switch chip, almost a whole year was spent (wasted) getting to this moment. This version of our test board even included as well a different PHY chip (the T.I. one used on the original EtherREGEN)--so that in case the issue was with the new PHY in this mode we could pivot. It turns out that, despite our being able to program the hundreds of registers of the switch chip via several methods, the only method that is able to get its SGMII port properly negotiating with a PHY transceiver in that mode is to use an EEPROM loaded with the exact same code as the chip vendor's reference board. This despite the manufacturer insisting that such was not necessary. It is a lot more complicated than I am even able to convey, but suffice it to say that this is a watershed week for us. [I think my blood pressure must have dropped 10 points this week. John is also quite relieved--and exhausted. I was getting worried that he was running out of strands of his hair to pull out.] On the other end of the board, the RGMII to PHY connection--used for the 'A' side SFP cage is also working properly. Issues with that were solved long ago several test boards earlier, but it was comforting to see that continuing to work perfectly with this latest board--and via the EEPROM programming of the switch chip (which did require some editing of the code). This present board has some other sections with new power network parts under test, but those are separate, easy matters. So now we can finally move onward towards the first "unified" EtherREGEN Gen2 board. It will be a 6-layer board and will include all the elements of our sophisticated new clocking scheme--consisting of two new clock synthesizers, the extraordinary sine>square wave converter seeing its first use, and ultra-low-jitter buffers for the various SE>LVDS and LVDS>SE conversions required. And of course now our magic ultra-low-jitter 10GHz capable differential flip-flop, running at 1.25GHz--to alternately feed the 'B' side PHY for RJ45 and 100% directly the new 'B' side SFP cage (that's going to be incredible for endpoints with SFP cages--especially via use of a Direct Attach Cable). The high speed isolator chips will also be in place on this next board, as will our first pass at the final power networks for all the various voltages needed. The building blocks of our power networks (mostly with lots of LT3045s as before, though we are experimenting with an equal-performing new 1A reg that may generate less heat and we would need fewer of them than the 500mA LT3045) are all known, but balancing their use everywhere will take some measurement to finalize. So the board will have these very cool NVE magnetic current sensors sitting above power input traces next to a bunch of chips. Those get connected back to some sort of Arduino and display--so he can see exactly how much each chip draws under various operating conditions. John and I spent several hours this week discussing the time and risk tradeoffs of how far to take this next board--which still must have lots of test points and programming interfaces, along with the aforementioned current sensors. Because we really want the board that comes after this next one to be the 100% true final, sure thing, pre-production beta test board (have to run a dozen final pieces before I risk $100K+ on salable production), we have decided to take the time for him to lay out this next test board with all I/O (including 4-port RJ45 module) and circuit elements to be as they will be in production. Even to the point where I can do the case end-panel CAD drawings--with known positioning of all ports--and order cases to be milled. The main difference is that the board will be wider and include the various programming and test elements which will then get stripped away to leave us with the production layout. That's where we are. A real path forward with no more stupid Ethernet chip programming obstacles in site. All just stuff that John has full control and expertise in. Now I know everyone just wants to know when will begin accepting orders and delivering EtherREGEN Gen2. Me too! (And my wife, who was starting to question my sanity. ). Truth is, I'm only just now able to start time-lining this out. Seems like the design, layout, and board-house fabrication/assembly/delivery of this next test board will take us to mid-November. A couple of weeks with that board and then moving on to the final pre-production/beta design/fabrication/assembly/delivery is likely another 6 weeks (though there are some holiday periods). So truthfully we'll be thrilled to be shipping beta test units (to a list mostly finalized, and USA only) by mid-January. [Assuming all goes well--and that I can get production cases milled in time.] So I guess that really means a full launch in March (though if beta goes fine and I bust a hump with everything else, maybe we would open up to orders in February with a promised March delivery week). Launch will be two batches, a few weeks apart, of 250 units each. They'll probably all pre-sell on the first day! Oh, here is a photo of the current test board being probed on John's bench. This is the one passing data from end to end at last...

Well I can report that we several of the many Mutec REF10 Nanos we have sold have gone to owners of Rose Audio RS130 units. They all seem to extremely pleased the combination. [They also are using one of the other REF10 Nano outputs with an EtherREGEN that is feeding their RS130.]

I hope the private rooms in Austria have better acoustics than those glass atrium echo chambers that I remember from my one and only High-End Munich trip about 15 years ago. 🤣

Oh good grief! This from the fellow who also wrote this gem yesterday: ”Most of us have come to the conclusion that almost, meaning almost every DAC uses the same chip set and they all sound the same.”

I have no idea. That big old Merging network board might have been sold as an option at one time, or it might only be available as a spare part. Either way, directly from Merging as a service item. I’d take a wild guess that they would quote some where between $1,100 to $1,600 for one.

Those two are the exact same home studio processor, just in different colored case. It's a neat product though. Interesting. So Merging is divesting of their home audio line. And NADAC went from $10K now to $25K. Wow. By the way, the spec page for the NADAC D on the Master Fidelity site shows the optional Ravenna input module as "Temporarily Unavailable." Well the price list Merging sent me back in October 2017 had the OEM price for 100+ ZMAN modules at $132 each. That's certainly more than the $45 for a ConversDigital mConnect CDM4140, but about on par with the very nice StreamUnlimited Stream1955 module.

It baffles me that Merging is still using that giant old Horus network board in the NADAC instead of their own ZMAN OEM module: https://www.merging.com/products/zman-oem AFAIK, Happy Mk.II is the only Merging product utilizing ZMAN. The board has been out for many years now. Back when it was introduced, some were claiming that with this module Ravenna/AES67 was going to quickly achieve hegemony in the Ethernet-input-equipped DAC market, but aside from maybe 3-5 pro-sound processors and cards, 3rd-party ZMAN usage has not happened much at all. Oh well.

Looks--and sounds--like it was a fabulous show Chris! Thanks for sharing your experience with us. I had a laugh at your cable TV story at the beginning. But I don't suppose your cable box in 1984 was as primitive as what we had in our house in the late 70s. I remember--when 1981 rolled around--discovering MTV on button #24! 🤯

Glad to know you got working stable at last. As much as I enjoy interacting with our clients and helping them, easily 75% of our support emails are about things that are fully covered in the EtherREGEN User Guide. 😐 [What amazes me though is that I wrote that whole manual before anyone in the field ever used an EtherREGEN, yet it still seems to anticipate and answer the vast majority of issues that arise.]