[Soekris] Strange problem with net4801
czauner at onlineloop.com
Thu Jan 20 01:17:26 UTC 2005
On Wed, 19 Jan 2005, Horst Laschinsky wrote:
> > Do not use cables with shielding on both sides. either unshielded,
> > or shield connected to ground on only one end.
> What the f...? That worked! Can anyone explain me why? Up to now,
> I always thought the sole reason for the existance of common grounds
> was to _avoid_ disturbances???
Well, see - thats complicated. Ignoring the fact that Ethernet-Spec
ist for *U*TP (with "U" meaning: unshielded) you can build up funny
effects in AC-electrical Environment.
1) define ground. Look at your DC-Power-supply. Are you sure that
ground == ground for two of them (The anser is usually "NO")?
2) electrical installions ofter suffer from poor grounding. In many
3) What you have / build when connecting two devices with an R >0 Ohm
against pyhsical ground, by some rather massive Cable (=Capacity)
is just a plain Oscilator (R-C Chain).
4) Its easy to construct a "Spannungsteiler" building this way.
Sorry, I don't know the english verb for that. e.g. here in
.at Power-Companies run AC with 230V. in about 10% of all instalations
one can find 115V AC on patchpanels (on poorly grounded equipmet)
And so on ... additionaly rx/tx on ethernet is usually galvanic seperated
on the in/out leads.
When planting ethernet, and you feel like grounding, do in *one*
place. for instance at the patch-rack. Never connect the shield on
Besides avoiding nasty HF-Problems, this can be a real life-safer
for your core-equipement if one of this cheap office-computers
does something very stupid like providing 230V on shield.
Besides: Ethernet-cables ar not for providing "common ground".
For providing common ground there is this green/yellow stuff
with copper in it. Usually with a thickness >= 1.5 mm².
PS: Avoid *any* cabling company doing incompetent shielding.
Throw them out.
More information about the Soekris-tech