[ale] Should I ground a ethernet switch?

[Alex Carver]

On 8/11/2013 04:05, Mike Harrison wrote:
>> Indeed DO use the ground rod as Jim suggests (make it as long as
>> possible) but don't use simple wire for lightning protection.  You
>> should instead use copper strap
>> (http://www.dxengineering.com/parts/dxe-cs2-25 plus the ground
>
> The strap, with a large flatish surface, is used because lightning level
> strikes travel on/near the surface of such straps. I've seen it
> described as 'plasma', but more importantly, I've seen it with my own
> eyes. It's impressive.

That's true, it's called the skin effect and happens with any high 
frequency current.  Lightning is a near-perfect impulse in the time 
domain (also called a delta function in signal analysis) and therefore 
it contains all frequencies from DC onwards (the Fourier transform of a 
delta function in the time domain becomes a constant in the frequency 
domain).  The self inductance of a round wire will allow some of the low 
frequency current to pass but the high frequency current will get 
blocked by the wire causing it to flash over through the air (which has 
less self inductance than the wire) and go somewhere unintended.  (And 
plasma is the right word since HF currents can light plasmas with less 
effort than DC currents.)

Inductance is frequency dependent, too, so it takes a lot of care to 
ensure that the cut-off frequency of what amounts to a low-pass filter 
is very high, beyond the range of the majority of the lightning energy, 
thus lowering the inductance across the range of interest.  The strap 
lowers this inductance by its geometry.  Very wide straps have lower 
self-inducance than round wire or narrow straps.  Large antenna towers 
also frequently have several straps (one per leg) to provide parallel 
paths but they are spaced far enough apart to prevent mutual inductance 
which would reduce the effectiveness of the straps.


>
> Now a lesson in grounding: At the normaal residential level, you want a
> SINGLE point of ground, typically tied to the ground/nuetral buss at the
> breaker panel. Beware of having two (or more) physically separate
> grounding points, there can and often is a small voltage difference
> between them.
> Small, but the kind of things that can cause noise in audio gear.
>
> IF you are a ham radio operator, your tower and radio shack might have
> it's own ground, and while commonly not done, it should be isolated from
> the your house. Isolation transformers are used.
>
> If you are a wireless ISP (I was), my tower and all of the associated
> electronics (Motorola Canopy gear, GPS Clock.. etc.. ) was isolated with
> it's own big flat grounding strap and ground field (collection of copper
> rods in ground, meshed together), a proper isolation transformer that
> did not pass 'ground' and a fiber optic ethernet connection. A
> configuration which later owners of the business did not understand and
> it cost them replacemment gear several times...
>

There are two avenues here.  Either fully separate the grounds as you 
describe (full isolation, separate rods, etc.) or fully bond the 
grounds.  The biggest problem is that many people don't go one way or 
the other, they sit in the middle with separate ground rods but other 
ground paths established leading to the ground loop noises and other 
problems.  A fully and properly bonded system will not have voltage 
differences from one ground rod to the other (just like your ground 
field does not with respect to itself).  By example, a couple commercial 
antenna installations that I've visited used the common ground method 
for the antenna tower, transmitter, and shack electronics.  The antenna 
was bonded to a ground field and wire mesh field (both rods and wire 
mesh for a significant distance around the tower).  The transmitter, 
shack power (and electronics), and feed line protectors were also 
grounded to this same network of ground rods by another copper strap. 
Any signal lines (phone, ethernet, etc.) leaving the area were isolated 
in some form because they were headed to something that was at another 
ground point.  But anything local stayed confined to the same common 
ground.  There were no noise issues caused by ground loops (radiated 
noise from HF electronics were a different story) and all points did not 
register potential differences in ground.

There was one exception which was the marker beacons for the tower. 
They ran through an isolation transformer in order to keep RF from being 
picked up by the very long power feed up to the lights and injecting it 
into the main power.  The conduit wasn't sufficient to keep any bleed 
from the hundreds of kilowatts of RF from running back down the wire.