UpTone Audio EtherREGEN Listening Impressions

[_JL_]

Received my ERs today and spent the day setting it up and did some evaluations. If you just want the conclusion then: don't walk, run to buy one for your system!

 

Okay, now all the details. First my setup (see my system for the audio/video config).

 

Network Setup:

(a) ISP Router (Huawei) > GBE > Netgear R7000 (router mode, DHCP server) > Long and lousy GBE cable > ER > Streacom Win10 PC w/ Roon > USB ...

(b) NUC ROCK also connects to Netgear R7000 via GBE

(c) UTP cables are mostly Bluejeans CAT-6 plus a few others

 

Power Setup:

(a) Netgear R7000 and NUC ROCK on Teradak LPS

(b) Roon PC on Teradak ATX LPS

(c) EtherRegen on UltraCap LPS1.2 on Teradak LPS, EtherRegen grounded to mains Earth ground via ext clock BNC shell.

 

First Impressions:

The ER replaced the common Cisco switch which the latter was modded to be powered by the UC LPS1.2.

First impressions are:

(a) Cleaner across the board with better sense of the ambiance cues. This results in deeper soundstage where instruments previously buried in the background can now be more easily heard. I kept hearing new things in familiar tracks.

(b) The ER gets rid of some slight digital artifacts which are most prominent in the treble range, such as female vocal and violin.

(c) The sound is much more organic and sounds very much like good vinyl - it's not just smooth sounding mind you, as too smooth could loose resolutions. Here with the ER resolution jumps upward a few notch and at the same time sounded natural and inviting. I kept raising the vol of the amplifier as I can now listen much louder without stress. The sound just blossoms and you keep wanting more of it.

 

Now this is clearly just first impressions as there was not run-in time at all but I am one happy camper already.

 

Cascaded EtherRegen:

Okay I bought two ERs as there were reports using other audiophile switches that cascading them further improves the sound. I tested that by installing the second ER upstream between the Netgear R7000 and the long and lousy GBE cable connecting to the downstream ER. I also moved the NUC ROCK to the upstream ER for obvious reasons.

 

Initially it didn't work - because the upstream ER's B side is 100Mbps only but in the downstream ER I connected the upstream link to a copper module plugged into the SFP port which is 1Gbps only. Switching the link to the downstream ER to another A port solved the problem.

 

Now what about the sound? Well, surprisingly it is not positive. In comparing to a single ER the cascaded setup sounded too smooth which rounded attacks and seemingly lowered resolution. I quickly reverted back to the single ER configure and more surprisingly it did not fully restore the sound - halfway maybe but definitely not as good as before.

 

Then I realized the ER's uplink was connecting to a normal A port so I swapped that back to the copper-module SFP port and voila the good sound is back! What this means is that the normal A ports sound different from the SFP port - the latter sounded better in fact! How this is possible is beyond me but that's what I heard. This also implies that the cascaded ER setup was perhaps handicapped by its inability to use the SFP port.

 

I originally installed a copper GBE module to the SFP port only because I needed the extra port. It is a pleasant surprise that using it as the uplink (to the ROCK/Qobuz) sounds even better. In case anyone wonder the module was from fs.com: Cisco GLC-T Compatible 1000BASE-T SFP Copper RJ-45 100m Transceiver Module #11773.

 

An Extra Surprise:

If you're still reading then I have another pleasant surprise for you. I have a home theater where I use the Apple TV 4K for Netflix. Naturally I connected the ATV-4K to the ER and when I fired up the projector I was stunned by the improvement in picture quality! The picture appeared to be more solid and the black is much deeper.

 

This really blew me away and frankly the ER is worth it for the improvement in picture quality alone. The sound was also improved, not as significant as in the 2-Ch setup but still clearly discernible. If you have a video streamer then the ER is a must try!

 

Future Work:

I now put the second ER in a branch network connecting PCs, NAS, etc., for run-in. I'll likely be getting a second copper GBE SFP for it and possibly give the cascaded ER setup another go. I also have a newclassd NS2 10Mhz clock to experiment with so plenty to try.

 

That's it for now. Thanks for reading!

[_JL_]

7 hours ago, dminches said:

 

Is 100Mbps enough bandwidth for Apple TV 4K streaming?  The minimum requirement is 25 Mbps so I guess 4x should be sufficient?

 

Very excited to try this.

 

 

Well my ATV-4K is plugged into one of the A ports, not the B port so it has full 1Gbps bandwidth. The surprising thing is that even that improves the sound and picture of the ATV-4K, compared to my Cisco switch with LPS.

 

I haven't tried plugging the ATV-4K to the B port as I need that for the streamer PC but I think even 100Mbps is more than sufficient for 4K streaming as the incoming link on the A side is 1Gbps.

 

Try it and see if you like it!

[_JL_]

On 11/17/2019 at 12:17 AM, _JL_ said:

Then I realized the ER's uplink was connecting to a normal A port so I swapped that back to the copper-module SFP port and voila the good sound is back! What this means is that the normal A ports sound different from the SFP port - the latter sounded better in fact! How this is possible is beyond me but that's what I heard. This also implies that the cascaded ER setup was perhaps handicapped by its inability to use the SFP port.

 

 

Updated Impressions After Run-in

I have my ERs running continuously for a couple days so it's time to reassess it. Previously I found connecting the uplink to the SFP port sounds better than the normal A port. This turns out to be due to (lack of) run-in. Now I realize that the ER fresh out of the box sounded rolled-off in the treble (which is consistent with some other members' reports) and was not as transparent as the SFP port. After run-in it changes significantly - it now sounds smooth and resolving. The difference between the A port and the SFP port is now very very slight, with the SFP port perhaps a bit brighter but the A port is smoother and quieter.

 

Now I prefer the normal A port over the SFP port running copper Ethernet for the uplink. I wonder if the SFP cage has the same isolation transformers as the normal A ports? Perhaps Alex/John can shed some light on this.

 

I'll revisit this once I receive the fiber SFP modules.

 

Internal Clock vs External NewClassD NeutronStar-2 10Mhz Clock

The NS2 clock is powered by a KECES DC-116 LPS at 12V. A short matching 50-ohm coaxial cable is used to connect it to the 50-ohm version ER. The NS2 clock has a few days' continuous run-in.

 

The two clocks do sound different but the difference is very small after run-in. The NS2 clock has a slightly more lit-up treble than the internal clock which can be appealing in darker sounding systems. In comparison, the internal clock is smoother, have a deeper soundstage, and slightly better resolution. This suggests that the internal clock probably has lower noise floor than the external NS2 clock.

 

I'm not sure if more run-in will change things but at this point I found the internal clock to be a better and simpler option.

 

Cascading ERs

So I tried this again, using two configs. First is with the Internet uplink, Roon ROCK, and link to the second ER all on the A side. Second is with the Roon ROCK moved to B side. In both cases Roon exhibited some connectivity issues. When it works I can't really say if there is any difference at all so I think the ER is working as intended, at least in these two configurations.

 

In retrospect I should have tested linking the upstream ER's B port to the downstream ER's A port. Maybe next time once I receive the fiber SFP modules.

 

Removing Other Devices from A Side

I tried removing all other devices (ATV-4K, AV Receiver) except for the uplink from the ER's A ports to see if that will improve the sound. I did some back-and-forth comparison and honestly the difference is so small I could have imagined it so I think the ER again is working as intended!

 

Summary

So the key findings here are: (a) run-in is a must before serious comparisons can be made; and (b) the ER performs exactly as predicted by Alex/John - practically immune to upstream and concurrent connections in the A ports, and superb internal clock performance.

 

Congratulations to Alex/John for a fine piece of well-engineered component!

 

[_JL_]

9 hours ago, stevebythebay said:

Nice to have confirmation on what many of us have experienced so far.

Have you or do you plan to test an LPS or two in lieu of the supplied power supply?

 

I have been using the UltraCap LPS 1.2 to power the ER. I will try using a more ordinary LPS (KECES) to power it later to see if it makes any difference. Hopefully not then I can repurpose the UC elsewhere. Otherwise I'll have to get another UC for my second ER. Bad for the wallet.