Re: Control Box Isolation

Hello, gang--

(I read from a digest, and am often a bit behind. So if this problem is
already resolved, just ignore this...)

I once worked on a plasma cutting system and we had some similar
problems.
We found several sources to the problem, perhaps one (or more) of these
are what you're running into:

1) Make sure you minimize loop area in your circuits. When you get a
large
loop in an input circuit, you can couple to a *lot* of magnetic flux
generated
by high-current loops. (Such as welders, plasma cutters, stepper
controllers,
etc.) A loop with enough flux coupling to it will definitely get you.

For example, if you have a few series switches, wiring like this would
be
BAD:


+------o->o--------o->o---------o->o--------o->o---+
| |
| LOOP AREA --- BAD! |
| |
+-------------------+ +--------------------------+
| |
input

Using twisted pair is good because while you must still have some loop
area,
some if it is in one direction, and some is in the other direction,
causing
magnetic interference to cancel itself out:

A B A B A B A B A
+-\/-\/-\/-o->o-\/----\/-o->o-\/-o->o-\/-\/-o->o-+
+-/\-/\-/\------/\ /\------/\------/\-/\------+
| |
input

So the voltage induced is amount induced in all the A loops minus all
the B
loops.

2) Make sure that you have only a single, large, LOW IMPEDANCE ground.
If you
have multiple grounding points, then when one circuit activates or
deactivates,
the current flowing through the grounding system can cause voltage
shifts at
multiple other grounding points. Multiple paths in your ground circuit
can
often cause "voltage dividers", so as currents through one part of your
ground
change, the voltage at your ground point in other parts of the circuit
can
change. (For example, if you switch a 10 amp current and your ground
impedance is even a half an ohm, you can easily cause a 5V difference in
your
ground potentials.)

Making things worse is that wire has a natural inductance in it, so it
resists
instantaneous current changes across it, causing brief voltage
variations. So
don't use wires any longer than you need to. (Especially don't coil
them up
in a corder if they carry variable current loads. This causes
fluctuating
magnetic fields which can couple to other circuits, as in (1)).

3) Parallel conductors are effectively capacitors, and capacitors can be
very
effective conductors of high-frequencies. So a stepper-motor control
line in
parallel with any logic signals is a great way to cause yourself
headaches.
When possible, run your logic signals at right angles to any lines
containing
high-frequency transients. Failing that, maintain as much separation as
possible.
It's also a good idea to shield any logic lines. (See rule 2, though.
Don't
ground both ends of your shield, because providing multiple ground paths
can
be BAD!)

Sorry for the verbose message. (You'll be happy to know that I *did*
edit for
length. Normally, I run at the keyboard for much longer stretches...)

--marco