Maybe someone else will chime in. Thanks 
3 Likes
I see what you did there ………
3 Likes
Like you, @badger1, my background is in software engineering, did a post graduate degee at Cambridge, got recruited to do some R&D works for some big guys in the Silicon Valley, including developing some back-end networking components for Netflix, Youtube, and even New York Stock Exchange realtime trading platform, etc.
I would be interested in your journey here, your would-be final destination? I bought the Linn NGKDSM when it first came out at the start of pandemic, and like many of you here, I did some experimental works with various network switches, copper vs optical, wifi, just out of curiosity, although my education kept me very skeptical at various claims (at improving the musicality of the system).
Now I’ve ended up completely wireless and cannot be happier.
3 Likes
A better clock, depends on what you measure. Like phase noise and jitter which can affect the streamer. HF phase noise can be reduced with a phase locked loop but if you try that with LF noise the lock will become very slow. So it is a harder problem.
I’m not very good at this but you get the idea. When looking at a clock both jitter and phase noise matter, and with phase noise the spectrum of the noise matters. And these can affect the audio clock in the streamer and the rest of the channel and the dac in particular.
But I may have got this wrong. I hope someone can correct me here! It was a while since I dealed with this. But basically you need to have clean clocks everywhere since the effect accumulates.
I recently read a French review on the Cybershaft clock, by Joel Chevassus.
He connected simultaneously the clock to the LH switch and to the Esoteric streamer,
Here his conclusion :
« Having recently been able to extensively test the combination of the Genesis clock and the Bonn NX switch from Silent Angel, my personal cascade remained competitive without synchronizing my devices with an external clock. And even if I recognize that the Silent Angel switch went a little further with the external Genesis clock than with its own internal clock, this association did not necessarily make me decide to take the plunge, the performance gap n 'being not enough to my ears to call into question my personal system.
With the LHY SW-10 switch connected to one of the four BNC ports of the OP21A-D, this performance gap now becomes much more marked. It’s hard to go back with my personal stunt, and I finally fell for this combination. The contribution of the OP21A-D to the LHY SW-10 switch covers all the criteria: better definition, better dynamics, more varied and realistic timbres, an even more structured soundstage with better relief. Let’s remember for the sake of good form that the LHY SW-10 switch already hosts the very good OCXO OCK-2 home clock. The significant contribution of the Cybershaft clock is therefore a real performance!«
Here is some writing by an engineer from from Afterdark, who design external clocks for switches. @simon.pepper has one. He can testify.
From the Afterdark site, on Mutec MC3 clock and merits on connecting it to a switch :
« First, did I correctly understand the function of MC3?
- Re-clock digital signal with native internal clock? Yes
- Can accept external 10Mhz clock for further enhancement of re-clocking? Yes it can accept 10M Master Clock
- Can distribute the connected 10Mhz clock to additional devices? However, it do not have 10M Master Clock output for additional devices.
- Can convert various digital inputs connections (USB to Coax, Coax to AES, etc.)? Yes, it have AES, Coaxial, Optical output, so they can output at the same time. it is pro machine, it works perfectly.
- Can also provide 44.1KHz, and 48KHz based clock signals? Yes, if you have a very good CD transport/DAT machine, this can all benefit with the world clock from Mutec, which was generated by 10M Master Clock.
Second is usage: based on Etherregen–>network player–>DAC
- Which will provide bigger improvement:
- High quality 10Mhz clock to ER (Cybershaft/Afterdark, Or:
- Mutec to re-clock before the DAC
The Mutec MC3-USB is different machine compared with switch, it can removes all the digital feelings, which make music more organic, more laid back. The mid-range is lush & airy, bass is more authority and controlled.
UpTone EtherREGEN is an excellent to bring the dynamic and details from the network, streaming like Roon, Qobuz, Tidal, Spotify can all benefit with it.
The Master Clock just bring all the factors and improvement to another levels, depends on the quality of OCXO Master Clock, of course we agreed with John, the building and material used in the clock cable and BNC connectors are all matters to ensure the optimisation for 10M signal to work perfectly too.
We have perform testing on different brand of BNC clock cables, they have their own sounding, the BNC plug is very important. We prefer to use the HUBER + SUHNER, this is the most natural sounding which frequency range is exceeding the requirement for 10M.«
Hi Sean. I’ve worked over in SF for a while and in Capital Markets in New York and London so sounds like we have had some similar experiences! And I graduated from Cambridge in the 80s! I will update with my findings when the NGKDSM arrives and I test the options. I am rather hoping that wifi works well for me!
I was triggered into investigating this aspect of sound when I realised a while back that my lax attention to my network mean I had accidentally wrapped the second to last ethernet cable - some 5m of unshielded cat 5 cable - around a power board. The distortion that generated in the voices was audible. It was instructive to hear what that common mode noise did to the sound in the Linn DS and with repeated A/B comparisons I taught myself what the distortion sounded like.
My goal has been to remove that distortion rather than rely on the network to shape or ‘EQ’ the sound as that is surely like building a house on shifting sands. Whether that goal is impossible or not and it’s all noise shaping remains to be seen.
So far I have separated power lines from ethernet right through the network, tried to decouple the ethernet hops between my switches and have dedicated two switches to the hifi - the last one a poe switch. I have heard the distortion effect that was at its extreme with the ethernet cable touching the power board drop completely away.
Part of me hopes that wifi works as least as equally well as my coper ethernet does now. If it does then Occam’s razor might suggest thats the simplest setup with less risk of change in the future! Will keep you posted.
Bfn.
2 Likes
I do not have the level of expertise of @sean86 and your good self, but like Sean I have gone wireless on my NDX2 and I could not be happier. I have 1Gb fibre broadband and so my WiFi is more than capable of dealing with Hi-Res files and using the wireless antennae on the NDX2 means one less cable and no additional box for a switch that also needs to be plugged in.
I used to have an expensive EE1/EE8/EE1 set-up (well it was expensive for me; but a lot cheaper than other options mentioned on this thread) with Chord Signature and Sarum T ethernet cables. I decided to simplify things and as I removed each component I actually found that the SQ of my system improved and got back to that musical Naim sound without any complications.
I also no longer sit there wondering if my system is optimized/I need more kit. I know it just works 
3 Likes
I am struggling a little to understand how the phase noise from the switch is transmitted into the DAC when there is a buffer in the digital streamer. The step that gets the data from the media server into the buffer in the streamer is an asynchronous step - there is some 10-30 seconds of audio buffered in my Linn DS depending on the resolution of the digital audio.
When segementa of the audio data are stored in the buffer they inherently don’t have phase noise at that point.
1 Like
Thanks Frenchrooster. I understand the esoteric has the option to synchronise to an external master clock?. So in that scenario I can understand how a better clock in the digital switch if that capability is switched on will make a difference to the sound.
In the case of the linn ds, it uses its own internal clock to help achieve the synchronous step of processing the buffered data and feeding that into the DAC.
1 Like
I feel, but only my intuition, that a good clock on a switch acts in the same way as daisy chaining switches. Something common with timing.
Don’t laugh if what I said is completely wrong.
2 Likes
I won’t laugh frenchrooster. The daisey chaining works to reduce electromagnetic /common mode noise, as was described to me earlier in the thread, when unshielded cable or shielded cable that is earthed only at one end connects the switches. This prevents most of the elctrical noise from being transmitted from switch to switch. The nosie likes to travel along the shieleded part of cables. It assumes the switch has a clean power supply and is grounded correctly to earth. Thats why I have put the last switch on the same electrical circuit as the hifi and the previous switch in the chain is on a totally different circuit. I have 6m of shielded cable connecting them that is grounded / attached at my hifi switch. At the other end it goes into a wall ethernet socket and that is connected to my next switch using a short run of unshielded cable. So it is earthed only at one end.
That mechanism I understand. Its reducing common mode noise.
A clock, if it does anything, affects phase noise/timing noise. Not common mode noise. Sometimes called jitter. I think therefore it all depends on how the clocks are setup in your system - the clock is key to feeding data into your dac. In my case a buffer in my linn ds and the use of its internal clock removes any value to me.
Waiting for anyone to pile in and contradict me!
Take care.
Edit. Just to add for the moment I have gone one step further and put in a ubiquiti poe powered switch as the last leg connected to my linn ds with utp cable. So that switch has no power supply. Not certain yet if that is the most optimal last leg. It definitely sounds better than going straight into my 8 port ubiquiti switch. Its unreassuringly inexpensive 
2 Likes
Just to point, all streamers has an internal clock and are buffering.
4 Likes
Yes, I can confirm that an external clock brings benefit to the EtherREGEN, and the Ethernet based stream to my NDS, both when using Roon with the SonoreUPnP bridge and also Asset UPnP server.
I now have a silver DC cable between the clock’s power supply and the clock unit, and a silver 75-ohm BNC cable feeding the 10MHz clock to the EtherREGEN. These cables are burning in.
New 20mm internal fuse for the Clock power supply is the last tweak to this configuration. The JPS Mains cable already has an upgraded plug fuse.
2 Likes
Thanks Simon. How is the clock helping - is it replacing / superseding the clock in the nds?
No, its an external clock feed into the EtherREGEN, as part of the isolation and reclocking of the Ethernet based stream into the NDS.
Benefits are an increased sharpness, lower noise floor, better transient response, improved sound stage.
But, isn’t the nds buffering the music data? Once the data is buffered, it has zero phase noise? I don’t understand I’m afraid.
1 Like
I think it is only the newer streamer boards (anything after the 272 I think). The older ones were memory constrained.
1 Like
It is, to some extent. It is relatively easy to A/B the difference with/without the external clock. And then as the OCXO stabilises, over time, it gets better.
Others have also reported good results with using an external OCXO clock on the EtherREGEN, it is also what probably makes the PheonixNET the price it is.
2 Likes
Is it possible that the changes introduced are reshaping / adding common mode noise rather than reducing phase noise? Even a small buffer would reset any phase artefacts to zero. The phase noise could and will re-occur later of course as the data is fed into the DAC stage by the streamer.
Changes to the common mode noise profile definitely change the sound characteristics in my streamer. Almost anything I do to change the configuration of my last leg - Ethernet cables, switch, second devices on the last switch, position of last cable etc seems to subtlety change the sound. They seem to be as sensitive as my naim power cables and the interconnects. Which was quite a surprise to me.
Just a thought.