Ethernet cable effects overview - a summary

Good stuff Huge. Appreciate you taking the time to do this. 

James

Huge

Very nice and informative write-up.

Could you disclose what means of RFI protection you applied, did it go further than the ferrite cores we normally talk about?

cheers

Aleg

Huge, interesting stuff.

Now repeating your tests in the presence of an EMI/RFI meter on the cable could be further insightful

Simon

Yes...  Um, OK, here goes

1  No protection

No ferrites on mains leads, no ferrites on signal leads, no ferrites on SMPS, no mains conditioning units.

2   ferrites on mains leads,

no ferrites on signal leads, no ferrites on SMPS, no mains conditioning units.

3   ferrites on mains leads and signal leads,

no ferrites on SMPS, no mains conditioning units (audio or SMPS)

4   ferrites on mains leads and signal leads + mains conditioning unit on audio kit

no ferrites or mains conditioning unit on SMPS,

5   ferrites on mains leads and signal leads, ferrites and mains conditioning unit on all SMPS

No mains conditioning unit on audio kit

6   ferrites and mains conditioning units on everything (even the freezer!).

I'm currently at step 6

Next steps (not necessarily in this order)

7   Progressively remove the ferrites from the audio kit to see it an improvement to dynamics can be obtained without loosing resolution.

8   Build a capacitative DC blocking unit and test in place of the mains conditioner

9   Add a three way MOV network to the DC blocking unit

10 Repeat 7 with the DC blocking unit in place

11 Grow old disgrecefully

Simon,

Would love to, but currently I don't have access to one!

Huge

thank you for the details, so all with ferrites it is.

Cheers

I love ferrets, they tend to pop up everywhere.

S

Best tie string around you trouser cuffs.

What kind of ferrite should be installed? Where to get them ?

Wanna make a try ;-)

thx

If you Google 'ferrite core' you will find plenty of options.

Aleg,

It's not just the ferrites, as I seem to have a mains supply with a lot of muck (RFI) coming in, I use a mains conditioner for the audio system.  I've tested the system without the filter and the amount of audio information that gets masked is quite considerable, so I use the conditioner.

I also use another mains conditioner with all the SMPS plugged into it, to prevent them putting even more muck onto the mains.  This is in addition to the ferrites at both ends of their low voltage leads.

It's just occurred to me that a worthwhile test would be to remove the ferrites from the Ethernet cable and from the LV leads of the SMPS to see how much of the effect is related to the increased level of radiated RFI (as opposed to the mains born stuff).

But then you won't be able to walk, and ferrets love being taken for a walk.

Huge

Thanks for the additional information.

I have this nagging curiosity about power filters/conditioners.

On the one hand I would like to try one sometime (will certainly not buy to try), on the other hand I suspect with most of them there will be a negative effect on the Naim gear.

I think here in the Netherlands we have clean power on average (I have no hum on any of the Naim or Hynes toroids), but have only limited separation of linears and switched modes, as it is just not feasible to create the situation everywhere. Our wiring system is based on central junction boxes in a star arrangement:

A central junction box system for electrical wiring involves running wiring for each circuit (group) from the fuse/breaker panel to a 'central' junction box, from which branch wiring is connected to other junction boxes for outlets and light fixtures in a 'star' arrangement. The junction box for the ceiling light fixture is often used as the central junction box.

which might behave differently, electrically speaking, when trying to avoid mixing linears and switched modes.

But I still remain curious about any possible benefits.

I will follow your discoveries with interest.

Cheers

Aleg

Aleg,

I suspect this means that you can't separate the supply to the audio equipment from the supply to any dimmer switches for lights.  If you have dimmer switches wired from the same central junction box (i.e. on the same 'star') this could be more of a problem as they are even more noisy than most SMPS.

I'll have to have a think...

I also have little or no hum, but I found that when I installed a Tacima CS929 (€40) on the mains supply to my Naim kit, although there was a small loss of dynamics, there was also a considerable increase in the audible detail and the same time the treble became smoother and less fatiguing.  So, despite the loss, there were even more significant gains from using a conditioner on my mains supply and it has stayed in place.  You may get different results though!

At some time I do intend to try a capacitative blocker, as the capacitors do filter out higher frequencies, and there may be less dynamic loss than with a C-L-C filter.

Huge

Fortunately we don't have dimmers, but on the circuit 'star' that feeds my hifi, there are a few low voltage and mains voltage halogen lightning fixtures and also some smps'.

i have put all my hifi and lpsu on one end-branch and there is no smps on that branch, but still the returning neutral is connected to the neutral from other branches which do carry the smps and the trafos and lightning switches for the halogen fixtures.

the only clean solution would be to have a separate group in the CU, create a dedicated circuit with  new wiring in new ducts and use that solely for the hifi. That's my 'dream' for 'one day ...'.

Cheers

aleg

Aleg,

Mains voltage halogen won't cause a problem (electrically, it's just a piece of coiled tungsten wire), other than that you clearly understand the situation very well.

Good luck with the 'dream', it is the best solution for independence between different devices. It has the additional advantage that, if your mains quality deteriorates, you can drive that 'star' via an isolation transformer.

I'll post again when I have the capacitative filter / blocker up and running (it'll take me a little time to build it).

Anyone know of a US equivalent to the Tacima's?

Very nice investigation Huge, thanks !

How did you establish the condition of no detected errors in the data ? I have used iperf to obtain bandwidth measures in my local network but never tried to get specific data on error detection / correction and jitter / latency variations (there is a specific iperf option to get those measures, I think). I have no idea whether ethernet cables can have an impact on latency. If so, differences in sound quality should be expected even under ideal, zero error conditions, I guess.

To determine a suspect link in terms network data loss in connection to operation of a network, I would recommend two methods to be used in parallel. Step 2 ) confirms the impact of evidence found in step 1)

1) look at the frame drops on the port of the lead in question using a managed switch. Ideally one should look at the other end as well, but might not be possible.

This should show the counters of frames discarded matching certain criteria. Faulty / damaged  cables  tend to show issues with checksums etc.

2) the next tool is Wireshark. This tool usually runs on a PC and  needs to be fed by a spanned port on a switch or on the PC itself or fed by using a network hub instead of switch. The latter will change the network flow at layer 2 compared to using a switch so one should be cautious if looking at timings.

Wireshark will show the traces and decoding of the packets and higher level data encapsulations Here one can see the media,  UPnP control and discovery etc , and specifically one can see how the media is being transferred, any retries, and the windowing parameters used. Drops and retries on TCP windowed transfer within a LAN can suggest there is an issue somewhere in processing the data. If this correlates to step 1), its probably the impact of data loss from the cable / cable connected ports.

Iperf provides a view of layer 3 (TCP / UDP) dynamic and static  parameters of it's connected TCP/IP stack but doesn't show, in my experience at least, any protocol decoding so you can't see the specific impact of data loss , and it won't show discards at lower levels.

PS I have been running my Naim network streamer since 2011, and use Cat5e with a little bit of screened Cat5e using full duplex 100BaseT, and I have had not one single frame dropped through corruption/checksum error on my connected Cisco managed switch.

Simon

I too see no CRC Error Packets reported in my Netgear GS108ev2 (main switch for the NAS (main and backup volumes), Wireless AP, Internet Modem) and GS105E (for the Raspberry Pi running Asset UPnP & the ND5XS streamer). These two switches are connected by HomePlugs (Devolo dLAN 500 duo) as there is no physical connection/cabling possible.

These switches have a basic level of network monitoring, so show Packets Received/Sent and CRC Error Packets.

In the 'Show Errors' for the bonded Ethernet ports on the ReadyNAS Ultra 2+ running Adaptive Load Balancing over the 2 ports, there are just some RX packets dropped (across both ports) in the table.

Network Errors  [Ethernet 1+2]

If I look on my ReadyNAS Duo (much older, circa 5 years old, and was originally used with SB3/LMS over WiFi, and then Twonky to the ND5XS before the superior Asset UPnP server on the RPi), there are only 3 Bad Packets recorded in the table below, and there has been many TBytes of data carried over my network, as I have built, played and maintain my music library, which is currently 1.3TB in size with 22k tracks (36% in 24 bit too).

Network Errors  [Ethernet 1]

So bad packets, and Ethernet packet transmission reliability is not an issue nor influence in how different Ethernet cables are able to affect the sound reproduced.

Simon_in_Dublin