Mike Rivers wrote:
amble wrote:

-i've read that grounds should be connected at mixer and lifted at gear
but i'm not sure how this applies when a patchbay enters the situation.

That's old school. Leave the shield (which you mistakenly keep calling
"ground") connected all the way.

It may be "old school" but it's still the best way to avoid ground
loops. Sure, you can just wire everything up with the shields connected
at both ends and it will work most of the time. Until it doesn't and
then you can spend hours chasing stubborn ground loops.

The basic idea is a "star" ground system where each piece of equipment
has exactly one path to ground*. If you connect two pieces of equipment
that are both connected to ground via their AC plug and you attach the
shield at both ends you will have a ground loop. If it's a small one,
it'll have no audible effect, although external things can change to
make it become a problem, often at the worst possible moment. So the
"professional" thing to do is to lift the shield at one end, usually at
the output. Lift the shield at the output, connect it at the input.

Note that this is different than always connecting the shield at the
mixer. Connecting the shields at the mixer only will indeed give you a
star ground system, but imagine trying to apply this rule to a
multi-studio complex with 4 mixers, each of which can be routed to any
of the others - where does it tell you to lift and where does it tell
you to connect the ground? Better to think in terms of inputs and
outputs - that's always clear. Plus there's a slight advantage of
tieing the shield near the higher impedence input than at the lower
impedance output.

Now, as far as the patchbay goes, you have two choices: carry the
shields through the bay, or continue with the approach of lifting the
shields on output and tieing them on the input (i.e. think of the bay as
a device). I've seen both approaches work, the main thing is to be
consistent.

//Walt


* If you've studied topology, this is equivalent to saying that the
ground system should have a trivial fundamental group. If you haven't,
nevermind.