EtherREGEN: Installation, Usage, Difficulty, Questions thread

[One and a half]

On 11/23/2019 at 11:29 AM, stevebythebay said:

Wouldn't switching from an STP to UTP also "fix" the problem?  Try a simple Cat6a for all your copper Ethernet to see if that makes things work best.

A UTP would work, the other proviso is the routing of the power supplies that noise isn’t  transported via the mains. Simplest way I find is to create a drawing and determine the noise makers and how to isolate the trouble. 

[One and a half]

Apart from this post, has anyone else used the JCAT net Femto card & connected the EtherRegen in this fashion, or an alternative method?

 

JCAT port A -> to Router via switch (really dirty side), fixed IP address

JCAT port B (DHCP) -> EtherRegen Port A

                                      EtherRegen Port B -> Network Audio device (this case, Lumin U1)

 

JCAT ports A & B are in bridged mode, so that DHCP works on the JCAT port B.

 

Just need to know if DHCP will transfer across the EtherRegen, should, just checking the boxes.

 

[One and a half]

9 hours ago, Ehsu said:

Dear One and a half, I would love to know how JCAT Net Femto card alone compared to ER alone. Maybe not possible in your case as its already in the PC? I am thinking about building a fanless server with JCAT Net card and HDPLEX PSU. I am wondering if the JCAT card is unnecessary if I have an ER before my end point. I am looking forward to your listening impression and comparison if you have an ER already. 

Hi there @Ehsu, at this stage only planning. The JCAT Net Femto is the 'source' or 'start point' of Ethernet transmission. Unfortunately, for the Lumin to work, it needs to be connected to a router, and that means picking up other noise from the devices that are connected to the network. 

In the Ethernet transmission before the Lumin is a Baaske Ethernet isolator, which I could place in front of the JACAT Net Femto to reduce some crud, however @skatbelt tried this (with a similar device) further upstream of the EtherRegen with mixed audible results. He has a similar AC network to mine, so need to be mindful of similar.

 

The EtherRegen in itself is 'affordable', however when teamed with a suitable power supply, the process is somewhat more involved and the price escalates. The JCAT Net Femto supply is 2A, 5V and with the EtherRegen, a JS2 should work fine, now the price is close to USD1565, plus cables on top. 

[One and a half]

2 hours ago, Superdad said:

 

Try it with the stock SMPS we include. I think you will be surprised. That's what is being used by about 90% of the people already sharing their fine experiences with the EtherREGEN. 

Oh, I see the JS2 isn't available until mid Jan 2020, there are only so many hands, huh? Pity you're not in the shoe business, can put you in contact with a crew of elves...

[One and a half]

Thanks to @Cable Monkey, @elan120, @Dutch, @Ehsu for inputs with the JCAT NET Femto. Certainly doable.

[One and a half]

16 minutes ago, jcn3 said:

 

Have you  considered improving your source first? The idea with the ER is to isolate a low noise source/endpoint from other noise on the network. A laptop is inherently noisy so I'm not sure the ER will help you out much. 

I would agree. EtherRegen is best suited to renderers directly connected to the network. 

There are other methods for USB, for that there’s scope for experimentation, isolators, regeneration, cabling and so on.

[One and a half]

21 minutes ago, jamesg11 said:

Now with Clipsal earth plug direct into wall - does power outlet actually need to be switched on to get an earth connect?

The earth is always connected regardless of the switch which 99% of cases switches the active or hot wire only. The other 1% is a double pole switch which  switches the neutral at the same time.

[One and a half]

22 minutes ago, Stereophonic said:

Thank you. 
Yes. I have a Jcat Optimo Duo powering my EtherRegen. 
I didn’t find a better picture. I was talking about grounding ER to the JCat chasis ground screw... 

The Optimo duo has an earth connection at the DC output by looks of it, to connect a shield. Use that shield instead, since the DC power is referenced to that in the Optima. that way, the earth connection is short, the earth wire shown looks like an antenna to me.

Means braking open the DC connector and looping the shield to the EtherRegen.

 

An alternative is to run another earth cable from the Ether Regen to the same AC power distribution box as the Optima Duo.That cable  can be a standard schuko connector with only the earth connected. In that way the same potential works for the duo and the EtherRegen, and no cutting the DC cable. Would recommend the earth cable to be AWG12 or 2.5mm flexible.

[One and a half]

Received the EtherRegen today (22-Mar), eager to try, let's see which install works 'best'. JCAT = JCAT NET Femto (the original).

 

1. JCAT -> 1m UTP -> A Side | B Side 35m UTP -> Lumin U1. Power : 120-0-120V Uptone PSU.

 

Result : Clear treble, but found to be over the top, like adjust the treble control up a notch. Stage height down some 10-30cm.

 

2. JCAT -> 35m UTP -> A Side | B Side 1m STP  -> Lumin U1. Power : 60-0-60V Uptone PSU.

 

Result : Treble under control now, stage height and separation still not right, image squashed. Why is one instrument coming out of the right speaker only, should be both [faint triangle "Stay with me till dawn", FAC15 with Kathi Ogden]

 

3. JCAT -> 35m UTP -> A Side | B Side 1m UTP  -> Lumin U1. Power : 60-0-60V Uptone PSU.

 

Result : Stage height and spread of the soundstage not working, the height is collapsing and instruments aren't placed like before ER. The frowns are starting to appear.

 

4. JCAT -> 35m UTP -> A Side | A Side 1m UTP  -> Lumin U1. Power : 60-0-60V Uptone PSU.

 

Result : EtherRegen effectively bypassed but now includes the noise reduction of the A side transformers. It doesn't make that much of a difference. Not that much of a difference with the STP or the UTP (the STP is CAT 6, the UTP Cat 5e). No name branded cables, very generic. 

 

With USB, the cable makes a big difference, but that's more fine tuning than getting fundamental connections correct. Was I expecting too much from the ER at switch on plus 3 hours? No mention of burn in published in the manual, it is a switch after all. It does get hot, ambient 25.6C, case 43.5C. That's within spec, so no problem as far as compliance is concerned, but 43C is a lot.

 

5. JCAT CAT -> 35m UTP  -> Lumin U1

 

Result : Original config. Spread of sound has returned, instruments where they should be, height returned.

 

Burn in, will it make that much a difference, talking about thermal stability here, which should be good from the get go. TBH, I expected a lot more from the EtherRegen, for now I'll ask for an RMA and send it back to Mariposa.

[One and a half]

28 minutes ago, JohnSwenson said:

Hi one and a half,

three hours is not nearly enough time for burn in on the ER, its not just thermal stability, there is real burn in going on, primarily on the capacitors, there are over a hundred caps in the ER, they need to burn in with real signal for at least several days before anything approaching real performance shows up. It usually takes a couple weeks (some people say a month) before it really "opens up".

 

John S.

The 30 days is the limit for a money back guarantee, by the time the burn in is due, the time has expired. I didn't read burn in needed in the manual, but understand caps need time to settle. I'll ask for the RMA.

[One and a half]

7 hours ago, Johnnydev said:

swapping that fast also doesn't work at all!

Well it must have worked, since I could note the changes as each stage was tested. It's no different to adding another component like a dac or interconnect. 

[One and a half]

4 hours ago, agladstone said:

What’s the best way to burn it the ER, I myself (about one week so far with the ER in my system) haven’t noticed that big of any improvements, but I was being patient because I assumed it must require some sort of burn in. 
if I was to remove the ER from my system and connect it to my router and then to a computer running 24/7, would that be an effective method of burn in or is it best to just try and let internet radio play on my system at low volume or with power amplifiers off for 24/7 for a few weeks instead? 

That's a logical approach, either one, would have thought. To try and burn in the ER after recommending to do so, it's now placed between the cable modem and main router. 

[One and a half]

19 minutes ago, PYP said:

As soon as I read your first sentence, I thought:  Oh, no.    The eR took a month to fully settle into my system, but I think two weeks (powered continuously, if with a signal that whole time even better) is a good timeframe after which experimenting could be practical.  

 

I've had lots of equipment that never mentioned burn-in, but literally all have required it.  In my system, a month seems to be the magic number regardless of the type of gear.  I've always theorized that this is the result of capacitors.  They all seem to take a very long time and go through cycles of sounding harsh/bad, then good, then bad -- until they have fully settled in.  

I hear what you're saying, especially of analog equipment. For the last 3 components, the sound is actually 'OK', then improves as time goes on. This ER in relative terms is 'bad' from the start which is an inversion of what I've experienced. Soundstage collapse is due to interference on the signal lines, Ethernet or USB, the end result is the same. I will give the ER a week between the main router and the cable modem, there's plenty of traffic, with the supplied PSU. Two weeks was about typical for power amps and SACD player, worked fine, but a month!

[One and a half]

7 minutes ago, R1200CL said:

How about doing a test in reverse B>A and fiber to Lumin ?

The Lumin U1 does not have a fibre input, the test is not possible. In any case, the agonising wait must pass first before anything else is tried. As reported, the best effort is when the source cable and the target cable are plugged into the A ports.

 

I nearly burnt my hand in moving the ER off the router, that thing also runs hot, so the two of them create a nasty situation. In measuring the case of the ER, still reads 43C with an ambient of 26C.

 

With regard to thermal stability, I have planned to switch the ER on with the rest of the system which is on demand. The thought of the ER running all the time is a waste. Doesn't move the socks.

[One and a half]

5 hours ago, mattjtaylor2809 said:

Wow...lots to take in.

 

I didn't get the opportunity to play with my set up yesturday but hopefully will later today.

 

I did like the fact that the Zen Mini had bridged Ethernet, the idea of connecting the LAN to one port & then the STREAMER port direct to a noisy switch before then heading back to your DAC made  sense.

 

However, maybe in this example, the ER is working as an even better "Bridge" & the signal isn't travelling back to the main switch but to the A side of the ER, then isolated to the B side feeding the cleanest signal into the DAC.

 

So, in theory, this should work.

 

 

 

PS- I've been told a few times now to remove the ROON Core from the Innous, if I were to go back to having the Core on my office PC, is it still ok, even beneficial to have the music stored on the internal memory of the Innous, it's just removing the processing heavy ROON software that can create noise?

 

Cheers

Matt

 

 

The Cat8 are most likely shielded. Please replace with ordinary UTP Cat 5, this will avoid ground loops and differing potentials.

Yes, also move the Roon Core to your office PC.  I'm wondering if the Innuous is becoming obsolete, perhaps use it as an endpoint for Roon, but since it cannot accept SFP, this just complicates.

[One and a half]

15 hours ago, mattjtaylor2809 said:

Is this a problem? I spent a bit of money on these specifically shielding Cat8 cables which have been highly recommended & have led to big improvements in SQ....I don't understand why after installing an ER I should then go backwards to un-shielded Cat 5 cables, seems counter intuitive?

When using the DX filters with shielded cables, the difference is notable. DX filtering attenuates picked up noise on Ethernet cables, so the common mode noise that's still present, is conducted by the shield. The cables are dirt cheap, I would advise a trial without much money spent.

Shielded cables are great BUT. There needs to be a parallel run of an earth cable that runs with it, star grounding schemes, for Ethernet, that's not so simple. Easier on the power AC side, but when long lengths are involved with many nodes, problems arise.

[One and a half]

2 hours ago, FIndingit said:

But the DX filters themselves have no ground. 

They do maintain the shield of the connector.

[One and a half]

12 minutes ago, mattjtaylor2809 said:

 

I've just looked at the GentAudio website & I'm happy to buy some new RJ45 cables but it's not obvious that either version they have for sale is ungrounded at one end or does that not matter, as long as the shielding isn't connected to the outer connector?

 

With this design, does Cat 6 or Cat8 matter, I like the look of these ET12 Cat8 cables?

 

https://www.ghentaudio.com/pc/et12.html

 

I started to read John's DIY DC Power cable thread but at 51 pages long...😬

 

Cheers

Matt

No need to look too hard, the local Jaycar can fix you up, UTP Cat5e is more than adequate for speed.. Just to prove a point.

[One and a half]

5 hours ago, FIndingit said:

No. Every DX filter breaks the shield. 
 

Once again I’m going to recommend that people buy a cheap multimeter. Too much bandwidth is wasted and too many useless emails sent to cable manufacturers after these “Omg omg, I don’t know if my grounds are connected” discussions. 

Here are the components that make up the DX filter. It's a type E design.

 

 

The schematic diagram for one side of the filter. The pair are wired in series.

 

If you measure across P1 and P10 with a Multimeter (DC) across the DX it will read an open circuit.  Noise is not DC, but AC, even if the waveform is a bit weird. The filter creates an artificial 0V which is tied to the shields of the cable. 

 

There is electrical AC continuity across the shields when the circuit is made, not totally isolated, since the shields are capacitively coupled, and probably where the nuisance frequencies exist, the caps offer a reasonable opposition to those frequencies.

The 2kV cap will offer some expectation of a high voltage spike to occur.

 

Experiments this morning revealed that 50Hz leakage is prevalent when using shielded cables to a (non powered up) laptop's with the DX Filters. My Fluke 289 doesn't reach much higher, but it's good to see a stable reading. 

 

This could explain why shielded cables for the DX filters and other connections, transfer 50Hz leakage currents to devices, much in the same way as USB does and causing SQ to become rough and lose image height and depth. Using a UTP between the DX pair dropped the voltage on the laptop input from 2V 50Hz to around 50mV at 0.00Hz.  This is measured to a normal earth (receptacle ground). When using the UTP cable, connections P1 and P10 are not used, so they float as far as the cable is concerned, but internally create a balanced network., keeps the various pairs from drifting too far. 

 

That 2V could be eliminated by connecting the screen to a house ground, but it's impractical, so use UTP and avoid the headache.

 

 

 

[One and a half]

3 hours ago, Superdad said:

 

 

By the way, all of the EtherREGEN's RJ45 mag-jacks (Bel and Pulse brands) also "ground" the center-taps of a core on each pair through a capacitor (we add the caps, they are not part of the jacks we use).  Only we use a 0.1uF and not 0.01uF.  

In addition each of the RJ45 ports of the EtherREGEN are expensive (and often hard to get) 12-core-per-port variety. For example, here is the Pulse jack of the 'B' port:

 

 

Hi @Superdad, thanks for the correction on the type of circuit.

 

You'd think with all the filtering on the Ethernet Jacks that should be the end of noise, with the DX filters, the noise at my place must be very bad since they work well on the ER ('A' side only). Nice to see a choke on the Tx/B I wonder if this works as some kind of variable saturation effect of the cores between the upper and lower chokes or keep the two in balance. Just guessing here, each Ethernet Pair could receive packets at different times, if there wasn't a bridge between each section, suppressing the noise would be very ineffective. All this is very, very small stuff, but it makes such a huge difference to what we hear.

[One and a half]

18 hours ago, zoltan said:

I apologize if this has been shared and discussed. There is a short paper on the the effectiveness (or the lack there of) of the JSSG 360 type cables on Audio Science Review. It seems to me that it only makes sense if the connecting cable is very short and doesn't run along the cable itself. In fact that makes sense as explained and measured. I don't argue for or against these measurements but found it interesting. I'm just trying to figure out how I should terminate my LAN cables, shield connected to RJ45 plugs, not connected or just connected one end. Or strip the shield braid and foil altogether. Or the non of these will make any difference if the cable and plugs are the same. Btw, I'm not a fan of ASR as they some of the measurements seem to be pointless. Maybe this one is also flawed??
 

 

Let's pick some holes in this test setup from Audio Science Review.

 

The test setup measures a voltage across a wire by a very nice waveform that's repetitive via the induction process from 1MHz down to 1kHz.

This in real life doesn't happen. The test is measuring across milliohms from a 'nice' source. Induced noise is a spike with high dv/dt with varying decays and can contain multiple frequencies. doesn't take into account the capacitance of the conductor to the shield and the insulation properties.

 

What would be better:

Apply a DC voltage to the cable and a load that draws some spikes.

Then apply a nasty induction pulse like from a showering arc, transformer, fluorescent lamp or a new type of 'efficient' LED which draws really ugly waveforms.

Then measure across the braid and conductor with and without the extra drain wire. 

 

Once again the 'science' can be ridiculed as with most of the garbage at Audio Science Review which proports that measurements can frame the behavior of SQ, which is a nonsense.

 

I would dearly like to buy some 'better' cabling than the CAT5e UTP for 20c/m. Something with Teflon insulation and unshielded, haven't quite come across any yet, maybe some high temp designs.....

 

As for Ethernet shielding at one end only, the shield is still or should be at earth (ground) potential and shouldn't interfere with the transmission of the packets, since Ethernet can 'work' without the shield. I am not prepared to hack the shield out of a $400 cable, since the resale plummets to zero. Once again from experience with the supplied 6 inch shielded cables from DX-engineering, their influence does have an effect on the sound and the shields aren't connected to anything other than the chassis at the source and receiver end (router and PC), so very much on the fence here and leaning towards UTP as the saviour.

 

[One and a half]

@Superdad, yes, common mode noise if it does have a level needs a frequency specified. 100mV is very high for (digital) audio, you'd hear this as splashy hi hat and would quite easily affect the ground plane of the source and receiver. The network topology plays an important role as noise can come from anywhere, channeling results from just the source and receiver is a bit meaningless of a measurement.

[One and a half]

If I need to power down the network that feeds the ER and ending up at the Lumin, there's a ritual of what to boot up first as far as networking is concerned. Eventually, there's music again, takes about 30 mins to figure out what component is not happy, it's usually the ER though.

Once power is on the ER, it just works until today.

 

5 days continuously on after a power shutdown, music was playing, then stopped dead.

Lumin U1 advised to 'check network connection'.

Noted the A side network indicators were still working, so the network is OK.

Moved the B cable to the A side and the Lumin is connected again.

Moved the B cable back to the B side, and  'check network connection'.

 

Replaced ER with a Netgear GS108 with the same PSU for the ER 12V 1.5A  ex Lumin U1.

Works fine (actually the SQ is not so bad)

Setup a Lenovo PC wired connection to B side on the workbench with a 5A 12V SMPS and went through several reboots and updates, the B side is fine, network indicator is OK on the B side.

Disconnected the test setup, and reinstalled ER to system.

Works again as if nothing happened.

 

Is the ER on the fritz, it was shipped March 16, 2021, getting old?