[ProAudio] AES on the west coast, was Re: Ampex/ProAudio List Dinner

[proaudio at bach.pgm.com]

Jim Brown writes:
> On 11/6/2019 11:19 AM, Scott Dorsey wrote:
> > Ethernet is a special case because UTP ethernet is completely (well, up to 1KV
> > if the interface meets specs) isolated from ground on both ends.  So you can
> > run UTP between systems without worry about creating spurious ground paths,
> > and you have no worries about induced ground currents (as long as the voltages
> > stay under 1KV).
> 
> This illustrates the falsehood of thinking about bonding and grounding 
> ONLY from the perspective of noise at AF. I know of at least two 
> engineers, one of them Bruce Olson, where everything that was part of 
> his Ethernet network was fried by a lightning strike. This goes back to 
> the days before SurgeX series mode suppressors, so he was using MOV 
> suppressors, and the high potential between the equipment fried everything.

If this is the case, then they had induced ground currents greater than 1KV.
Which is the case in a lot of situations, and when that's the case we have
fibre available.

MOV suppressors work as well as whatever you're clamping them to.  The thing
about the ethernet interface is that it's transformer isolated on either end,
so the breakdown voltage of the transformer becomes your limiting factor.
Adding an MOV suppressor to clamp a spike to the wrong place is apt to do more
damage than good.

> There is no such thing as "signal ground" for racks -- more "sump 
> theory." 

There is always a signal ground for racks.  It's signal ground because I
say it is, and I put a sticker on it that says GROUND.  That makes it ground.
If your ground is at a different potential than my ground, that's just because
it's not signal ground.

Ground is whatever I call ground.  It is a convenient abstraction to make
measurements easier and more comprehensible.  It is not a zero impedance
dumping ground, it's just arbitrarily whatever I want it to be.  Maybe my
signal ground is your +5V rail.  It's still ground because I call it that.

>In addition to bonding to perimeter grounds for lightning 
> protection in that microwave installation, the racks must also be bonded 
> via the "green wire" from the power source. Shielding does not depend on 
> an earth connection, and it DOES depend on shielding being continuous 
> without breaks or openings having wavelength significant at the 
> frequencies of interfering signals. Signal wiring to racks is virtually 
> always coaxial for RF, and should be shielded twisted pair for analog 
> audio, both with shields bonded to the shielding enclosure at the point 
> of entry.

All of this stuff is carefully detailed in Motorola R56, as I cited earlier
in this thread.
--scott