Lumin U2 Streamer in 'da house

Just got the recently-announced Lumin U2 in for review. The U2 is solely a "streamer"; its function is to connect an upstream music server to a downstream DAC.

One of the nice things about the U2 is that it has two very useful new features that the U2 Mini and U1 don't have: an SFP-type optical transceiver port to allow connection via LC/LC optical fiber directly into the U2 from an upstream server (no need for potentially grungy and dirty copper Ethernet connections-gack! ), and a new "dedicated" USB port that is exceptionally quiet and...direct-coupled. This port's function is soley to connect to a USB DAC, rather than USB devices e.g. hard drives, flash media, etc.

Some pics:


Rear panel


You'll note the USB-A port at left that is specifically designed for audio; i.e., connecting to a downstream DAC. I'm told that this new DC-coupled USB port is quieter and cleaner than the S/PDIF, AES/EBU, and BNC connecting ports. The U2 also has dedicated chassis ground terminal so you can connect it to an Altaira.

The one I have in review is black; here it is on top of my Lumin P1, which for this application is functioning as the DAC and preamp.


I'm connecting to the U2 via LC/LC fiber from EtherREGEN in the remote server room, and the P1 with a USB cable.

It's only got about 70 hours or so on it, and it needs 500 hours to fully burn-in, but so far, I am very impressed. It's sounding very good, indeed.

More to come as I get more hours on it, stay tuned.

 

(Apologies I added to your quote lol)

[1] But as regards this part, can you explain what this means in layman's terms for the wrinkle-free, smooth brains (me) who know nothing ab "dirty" ethernet, and why ethernet is even bad, and what tf is SFP? How would someone who uses a pi2 design use this luxe-looking machine? As in, does this require a 10k internet setup, w mesh networks, switches, relays, etc? (i don't even know if any of that makes actual sense, pls forgive me.)

[2] have you actually tested this claim?

 

SFP is fiber Ethernet. As for the rest of the topic re: “dirty” Ethernet, I’ll just grandpa Simpson out of the thread, I think, so as to not contribute to sending @Thad E Ginathom to an early grave.

 

You just took years off my life by drawing my attention to this thread!



But in the end, it's all about ownership satisfaction. I'm sure Puma Cat will get lots. The thing is certainly pretty

 

SFP is an acronym for "small form-factor, pluggable". It is a specification for a digital networking device, e.g. a fiber media converter, aka "FMC", that accepts an optical transceiver and typically also has an RJ45 port to connect a copper Ethernet cable to a network. It's used to send and receive digital signals over a network via optical fiber, generally over very long runs as optical fiber is not sensitive to the problems that copper Ethernet can have when used over longer distances. There are two basic types of SFPs: SC/SC and LC/LC. The basic difference between these is the type of plug on the optical fiber cable that's used to connect the optical cable to the transceiver/SFP. The type that's most frequently used for audio is the LC/LC specification, and the optical transceiver for that looks like this:


It simply plugs into a fiber-media convertor, aka "FMC", e.g. a generic TP-LInk FMC shown here...


or it can be plugged into an optical-compatible Ethernet switch that has an SFP cage, as shown here:


There are now a number of audiophile-grade Ethernet switches, e.g those from UpTone Audio, SOtM, and Melco, etc. that have SFP cage(s) on them as well as RJ-45 copper Ethernet ports so you can connect to them via fiber instead of copper Ethernet. The SFP or Ethernet switch can then be used to connect a "remote" music server (ideally one in another room) to a network bridge or streamer in the main audio rack using a fiber connection, as shown.


Regarding "dirty Ethernet", there are some types of copper Ethernet cable designs, depending on their construction and shielding configrations, that can carry electrical noise components. There's a common misconception that "digital signals" cannot be susceptible to the impact of noise factors because the "signal" is simply 1's and 0's. This is simply not the case. The only thing that is "1's and 0's" is the encoding of the music "data file". The actual signal that is sent from a computer, CD player, streamer, network bridge, etc. to a DAC is an analog voltage "square wave", where, hypothetically (that is, in an ideal physics world, where there are no noise factors, which BTW, also does not exist) 0V represents a "0" and typically (but not always), +2V represents a "1" from the original digital file.

Because the actual signal is an analog voltage square wave, it is susceptible to various noise factors, including low- and high-source leakage impedance current, deterministic jitter, threshold jitter, and phase noise. You can read about the impact of these various noise factors, including threshold jitter, by professional Ethernet engineer and EtherREGEN designer, John Swenson. John knows WTF he is talking about because he was an Ethernet electrical engineeer for Broadcom and Cisco for over 40 years.
https://cdn.shopify.com/s/files/1/0...enson_EtherREGEN_white_paper.pdf?v=1583429386

Moreover, unlike most folk's understanding, who think that noise "rides down low", this is also inaccurate. Most noise components ride ON TOP of the signal. In the case of a signal from a digital source, the noise components may look like this. This figure shows a square wave from a digital source with 2 and 5 noise components.

Generally, the only principal noise component that rides "down low" is shot noise, but now we're getting into theoretical physics, and I should probably stop there, because my guess is most folks here would rather not be looking at differential equations.

The key point here is that there is another type of jitter, known as threshold jitter that is due to low-source and high-source impedance leakage current. The particularly bad one is high-source impedance leakage current, which is typically caused by SMPS. The high-source impedance leakage current causes threshold jitter and the threshold jitter results in....timing errors. And, our brains are exceptionally sensiitive to timing errors when listening to digitally encoded music; we can discern timing errors in the picosecond domain, which is why we need femtoclocks for our digital source components. And, guess what? Leakage current can be carried on, Yep! You guessed it: some classes (constructions) of copper Ethernet cables. So, if you've ever wondered why different copper Ethernet cables can sound different (and...they DO), this is a key reason, based on real physics, as to why. The other noise factor that copper Ethernet cables are susceptible to is common-mode noise, but for the sake of brevity, I won't go into that here.

Nice thing about optical fiber is that it circumvents the noise factors e.g. leakage current and concomitant threshold jitter that are attendant to copper Ethernet because the signal is now photons, not an electrical signal. Optical will not fix phase noise from crap upstream clocks, however, so bear that mind.

Not, yet because I don't presently have a S/PDIF or BNC digital cable to use to connect to the Lumin P1. If I can obtain one, I will.

I don't own the U2, it is just in for review. I am quite impressed with it, so far, though.

Cheers, gents!

 

You can actually put together an inexpensive optical "network" to try out in your own streaming set-up. All you need is a run of optical fiber, two generic FMCs, two optical transceivers, and ideally, a couple Jameco Reliapro linear wall-wart style power supplies.
Here's a BOM:
1) 10Gtek FMCs + Optical Transceivers; you'll need two of these, one at the source end, and one at the "destination" end. About $27/each
https://shorturl.at/fsB04
2) A run of LC/LC optical fiber. The length isn't critical, what's important is having a run of fiber to prevent passing of leakage current. A run of 1M would work, for example. The 10Gteks are "multi-mode" FMCs, so you'll need something like Tripp-Lite 62.5/125 multi-mode fiber. ~$20
https://shorturl.at/dzI17
3) Two Jameco Reliapro wall-wart LPS (double-check the voltage of the FMCs, generally they are 9V) $11/each
https://www.jameco.com/z/GPU4109005...Volt-500-mA-2-5mm-Plug_1953647.html?CID=MERCH

All in: 94 bucks!

Here's a diagram using generic FMCs. In your application, your Raspberry Pi would function as the "network bridge", connecting the music server to your DAC.

 

Good to hear! At 2.5k it’d better be quiet as hell (I, for example, rely on PiAES on a couple of rigs, and a CXNv2 - very different price points)

 

Just an update that the U2 continues to rack up hours on the clock (per above, it needs 500 to fully burn-in), but already it's getting better and better and I continue to be very impressed with it. It's sounding...wonderful.

For folks that already have a Lumin U2 Mini, and are considering an upgrade, or looking for a streamer to connect a "remote music server" to your system, the U2 would be my "go-to" recommendation for a "straight streamer" (i.e., no internal DAC) for folks looking for a streamer to connect to a DAC of their choice.

It also has a dedicated chassis ground terminal for connecting to a Shunyata Altaira ground-plane noise reduction hub. This takes its already impressive performance to yet another level. Shown here is the rear panel depicting the optical fiber input and the ground cable connecting to Altaira.

 

This not the U2 mini, it’s the $5k U2!

 

Yes, and compared to the U2 Mini, it has a very quiet, completely shielded internal linear power supply. Also an innovative "dedicated for audio only" DC-coupled USB-A port.

it's the LC/LC optical SFP input that's "kicker", as far as I am concened, though. Presently using a Planet Tech MGB-TLX LC/LC optical transceiver. I'm using this specific transceiver, quite simply, because it sounds better than the other optical transceivers I've tried (yes, they all sound dfiferent). This is hgh-end audio, after all, where even the barometric pressure makes a difference one can hear.

 

Curious if you have any thoughts on active optical cables? The idea in audiophilia is you won’t disconnect the fiber and risk debris when moving them as they are connected to the optical tx/rx

 

I've been using a run of LC/LC optical fiber to connect from the music server in another room to the streamer/DAC in the main audio room for four years now, and have never encountered any problems. I sometimes unplug the optical cable to move things around or swap between componentry, but have never had any problems.

 

Problems don’t arise very often, just in my experience on my microtik switch I can measure the amplitude of the signal attenuate with each subsequent plug/unplug

 

Everyone's "use-case" is different, and based on direct experience, some set-ups may be more complex and "problematic" than others. There are no hard and fast rules, here.

One of the reasons I bought the Lumin P1 was that I was fairly consistently having problems with "connectivity" of my SOtM SMS-200UltraNeo network bridge. It creates it's own "server" on your home network, called "Eunhasu". I should add that my home is a veritable frickin' BLACK HOLE for WiFi. And, there were times I simply could not connect to Eunhasu via the home network to connect the SMS-200 to the system in the listening room. Sometimes for a couple days.

Also, my previous system comprised of a separate preamp with dual mono external power supplies (each with their own power cord), power amp, power cord for the power amp, network bridge, power supply for the network bridge, power cord for the power supply for the network bridge, AfterDark master clock, LPS for the AD clock, power cord for the AfterDark LPS, EtherREGEN (ER), power supply for ER, power cord for the power supply for ER, Ethernet cable to connect ER to the SOtM network bridge, USB cable for connecting the SOtM to the Lampizator DAC, power cord for the Lampizator DAC....arrghhhh! It was all too...complicated. And frequently did not work because of Eunhasu's "wonkiness".

When I got the P1 and the integrated, I removed 15 separate bits of gear from the main rack, alone. Fifteen.

Now, I just run fiber into the P1 and connect to a nice integrated with one pair of ICs. Only require 2 power cords instead of...7.

"Job done", as the Brits say.

And....it always works.

Simpler really is better. Just ask Nelson...

 

Anytime you expose the fiber mating interfaces to the environment you should clean them. The pen type cleaners work well and can clean both the cable ends and the SFP sockets.

https://www.amazon.com/FiberShack-Fiber-Cleaning-Tested-Cleans/dp/B07CNX8FJ3?th=1

You can also buy inexpensive (under $100) fiber microscopes if you care to examine the surfaces visually.

 

Very useful information, thank you. I'll order some today.

Cheers.

 

Thanks a lot for this link! This is the first time I ever read an explanation about how noise from the input of a DAC should influence the digital to analog conversion if there’s a buffer in between, and it’s an explanation that makes sense, too: The receiver that is subject to the noise creates noise on the ground plane, which in turn influences the DAC chip.

This makes me no wiser regarding the question how relevant this actually is, because these sparks on the ground plane can’t be large, and maybe isolating the power sources of components like many audiophile devices do will also isolate their ground plane from each other etc.

But: Finally, I read a sensible explanation about how it might be possible at all that such influences exist. Being a rational person, this makes my life much easier when I hear differences that, up to now, I didn’t have an explanation for that made any sense at all.

Thanks again!

 

Glad you found it helpful and informative. I learned a lot about this subect matter from reading a lot of John Swenson's postings about the development of EtherREGEN at Audiophile Style. And...John would know.

Cheers!

 

Nice to see they went with a linear supply on this one.

 

Yes, I agree.