Chord Ethernet cables

See, you shouldn't go away!

I did point out that the effect could go either way (and have other effects to boot).

I should have added that the network traffic can have an influence.  I must try to remember this and include it in my answers, thanks for the reminder (again).

And this flame-inducing report on the effect of mechanical constraints on the network cables, Huge, don't forget !

Simon, when my Airport Extreme died, I switched back to my ISP router for a while, then gave up completely and separated all the functions (bridge/firewall/switch/access point). I use a small Cisco Wi-Fi access point, powered through PoE in my case but you can alternatively power them with a separate power supply. They are not very expensive and you can define VLANs to your liking, and there's even a new model with 802.11ac. The only thing I'm missing is a manual on/off switch when it's powered through PoE, but that's because I didn't bother define schedules yet.

Aagh! you're all trying to trigger my ADHD!

And just because I'm paranoid it doesn't mean you're not out to get me!

Maurice with managed switches and APs you can define VLANs easily. However unless you have a device, usually a router, that supports 802.1Q trunking and routing between VLANs each VLAN will remain isolated and totally separate from each other as if they were physically separate LANs

I previously used a mini Cisco 870 access router to route between my VLANs and my switches and router were interconnected using a trunked 802.1Q connection. Most consumer grade stuff, such as from Apple, won't let you do this and don't even understand trunked connections.

Simon

Ingenious, excellent topology. All the benefits of isolation, but bridging available when required.

On my own, I'd never have thought of doing that (but then I'm no networking expert). 

Yes, I'm not familiar with VLANs (a long time since my sysadmin days), but I was wondering last night if you were supposed to route at some point. Thank you for the explanation. And I guess the switch can only assign VLANs to ports, the reason why you would need trunking ?

One obvious need for the use of VLANs is to define a guest network on the access point, which can only access the Internet. I don't really need this and haven't taken the time to study it yet, but intuitively I was wondering what would happen at the switch.

I think my needs are too limited at the moment to work on such a network topology, but it is always good to know.

Huge, interVLAN bridging can be a bit of an ordeal in my experience - wasted many hours debugging. If going between VLANS I always like to try and route between them and use helpers if necessary.

Maurice - indeed a guest wifi access using something like an Apple Airport Timecapsule  uses two VLANs via separate SSIDs  - there is only very limited capability and it is rather crude but both are routable from and to the internet, but not to each other, and both use different subnets.

Simon

I was not happy with this statement, so dug a bit deeper & opened up the case to use Mk-II eyeballs & trusty Fluke 79. I found the case IS connected to the PCB at the two PCB mounting screws with 2x capacitors on each (marked C85,86,87 & 88)  I did not measure for capacitance, only resistance & this see's capacitance as infinity,  hence my previous post that the case is not connected to the switch circuits.

The value between case & port shrouds is 550pF, but that might not be just the caps. 

With the network connected & running & meter set to AC amps connected between case & port shrouds it’s a steady 0.03a with a flickering volt reading in the mV range.  It’s the same between case & nearby radiator water pipe.   With the switch powered off, its zero.  

I’m not sure what the amp & volt numbers are indicating & TBH don’t make sense (to me anyhow)

But as the NAS, NDX & Wireless Hub ports do not have shield connections its not coming from them so it kinda indicates there is some kind of potential coming from the switch. It’s very low level stuff as the amp number can be changed by holding each end of the meter probes, but it’s not RFI (IMO).

So I’ve concluded that I’m not grounding my switch until I know more about this & for sure am not grounding it to the mains power earth as that can carry voltage.

Leaving the ammeter connected between case & water pipe & listening to some familiar music from NAS, I really could not detect any change in SQ if connected or disconnected.

However, not saying anything other than report what I’ve found & correct my previous misleading post.   

OMG this just gets more complicated...

I have an ND5 XS, the Ethernet shield IS connected to protective ground Irrespective of the position of the 'Signal Ground' switch (the shells of the RCA Phono sockets, and presumably the 0V pin of the DIN connector are controlled by the switch, but I haven't tested this).

I have a Synology NAS, and again the Ethernet shield IS connected to protective ground.

I have a D-Link switch (unmanaged) and here, the case and Ethernet shield are not connected to protective ground, but there is a ground connection on the back of the case.

Incidentally I use 2 lengths of Cat6a STP to connect the NAS to the switch, and have a UTP coupler between them to break the earth loop.

So it seems that with different manufacturers (or even models), you can have almost any (random?) combination of earthing of the shields!

It's bad enough for we few who understand this, for those who don't want to start understanding electrical theory and just want to plug it in and for it to work; this is a nightmare.

I think that's a bit OTT Huge. Ethernet is just plug it in and it works - no need to understand the theory. With networked audio, it seems a bit more work can gain you a bit extra but you don't need to do this to get a working system. 

As you say though with so many different combinations, it's no wonder the results people get are so variable. 

Sorry, yes James, accepted.

I'm a little fragile for a few days (side effect of local anaesthetic), I can overreact a bit.

No worries Huge - hope you're on the mend OK after the anaesthetic 

Huge,  sorry I need to back off again,  the Synology port is shield connected,  but as its supplied only by the SMPS 12vDC,  it cannot be protective ground.  

If you say the ND5 port shield is connected to protective ground (power supply earth),  OK I believe you.  

So I have disconnected the ethernet from the NAS & turned off the switch & found I still get an 0.03a potential between the disconnected cable shield & water pipe & that means it can only be coming from the NDX & an earth to ground voltage potential. 

Like I said we need to better understand this.

It's connected TO protective ground, but doesn't function AS a protective ground (as you say it's connected vis the SMPS so it can't be considered A protective ground); this is relevant for ground loops, rather than safety.

Do you know if the mains protective ground in your house is referenced to the mains substation (e.g. via underground cable), or locally via a ground 'spike', or both.  These differences can result in small potential differences between the mains earth pin and the local ground potential.

All points accepted

My house earth is TN-C-S  (Protective Multiple Earthing (PME))

The incoming neutral is a combined neutral/earth & is connected to the house main earthing terminal which connects consumer unit earth (& all house wiring earths)  

This "neutral" supply connects with earth at the local area transformer & at intervals along the distribution run.  (hence "protective multiple earthing") 

Hmm, unless you have high current demand in the house that should result in very low differentials.  Now I'm as confused as you are!

30mA though is a surprisingly high current reading.  On the Fluke 79, I believe it corresponds to a PD of 33mv across 1K1Ω.  If you change range to voltage then put say a 10k resistor and then a 1K resistor in parallel with the meter, do you get the same readings?  These three reading will give a good indication as to whether it's a potential from a low impedance source, a high impedance source or from a current limited source.

I just ran a continuity check on the ethernet shields

It does connect to NDX (ND5 also I assume) protective gnd,  I get 1.2 ohms which considering the ethernet connections, materials & length & the NDX internals, is pretty near to zero.  

The STP shields cross connect thru the switch, & the wireless hub has insulated ports & the NAS port although it has shrouded ports, cannot connect to protective gnd as its only supply is SMPS supplied DC 12v/0v - therefore this (my) STP network is grounded at the NDX.

So I am not going for any other grounding of switch case or ethernet shield locations as that gives two ground points & a potential ground loop.   

I'm not confused in that regard Huge,  an earth to ground potential of 4v is considered OK in PME testing.  This is due to various reasons,  one is imbalance between 3 phases in the local supply.  

Mike,

On the SMPS for my Synology NAS the 0V of the +12/0V power connector is connected to the protective ground, and the Ethernet port shroud is connected to the 0V of the power connector.

If yours has a 3 pin mains plug (rather than 2 + a plastic pin) it's worth checking, it may be the same.

I will do that Huge,  but not today,  I have to go out for the evening.  

None of the various SMPS diagrams I have on file have an earth connection thats connected to DC-0v.  The active parts of the circuit are all across L&N; in some of the simple (& I guess noisy) designs N runs straight thru to DC-0v,  most however are connected via diodes and/or caps or via an isolation transformer.  

Anyhow - until tomorrow

PME differences between protective earth and true ground at your house can be even greater - and one of the main drivers will be the voltage drop across the neutral back to the multiple earthing point or substation.

The PME drop in my environment (all over head rural wiring) was so bad I sometimes got tingles when touching the tap in the bathroom if the solid bathroom floor was wet. (the safety earth is bonded to the water pipes).

I paid to have an electrician migrate us to a TT earth system (locally safety earthed with local spike / electrode) and all was well - and nuisance tripping reduced as well.

Simon

Both my Synology DS213j switching PSU, and the Naim Unitiserve switching PSU, have  IEC inlets and measure zero ohms between mains inlet ground and -ve DC outputs.    Conversely, TP PSUs do not connect through.

Good thing you live in Suffolk rather than Norfolk, or I'd have said that whoever designed your previous mains grounding set-up was N-f-N!