On Fri, 14 May 2004 20:31:08 -0500, Robert & Linda McGraw K4TAX wrote:
>Over the years, I've dealt with a lot of complex analog and digital systems
>that must reside together and that must realize ground at some point.
Not ground, but signal reference. BIG difference. Sit out by your favorite
airport someday and tell me when you see a 747 taking off trailing a ground
wire.
More to the point, audio systems are pretty complex and interconnected too,
they have lots of gain, land they must reject some VERY strong signals.
Virtually every pro audio system involving a mic has voltage gain on the order
of 130 dB, and must achieve a broadband signal to noise ratio on the order
of 100 dB. Some systems stretch each of those numbers by 10-20 dB. Few
digital systems have much gain, and have at least 80 dB less sensitivity to
noise. Things you routinely get away with in digital systems would blow you
out if the water by 60-80 dB in an analog audio system.
I recently solved a severe RF susceptibility problem in my own ham shack
between the RS232 ports on my radio and my computer by using the chassis
screw on the DB9 as the connection point for signal return on the connection
between them, with NO connection at either end to pin 5 (the so-called "signal
ground"). The ports talk just fine, and RF stays outside the box. [The RF
field is strong in the shack because I load a dipole as a long wire against
ground on 80 and 160, which places a current node at the operating position.]
>Each
>system is different thus requiring often radically different means of
>achieving a 0 voltage point for a circuit. I do not subscribe to the
>statement that the "pin 1 problem" as being a design problem. It is an
>application problem. If one believes that the only place to connect a
>shield is the chassis, this is wrong because the chassis may not be at
>analog or digital ground. In RF applications, most likely the chassis is
>not at RF ground.
A pin 1 problem IS a design error, because it is a path into equipment for
interfering signals, especially at RF.
A shield does not need an earth connection to be effective as a shield -- it
simply needs to be an extension of the shielding enclosure. When it goes to
circuit common via the circuit board it does not do that. But the major factor
determining where to connect it is CURRENT -- specifically, the RF current
that can flow on it, and that we want to keep out of the equipment. If we
connect the shield to the chassis it doesn't enter the equipment. By contrast,
the audio signal involves very little current -- while the source is a low
impedance, the receiving circuit is nearly always a high impedance. The
signal circuitry needs to get its signal return from outside the box in a
manner
that does not allow noise (RF) to be added to it. That signal circuit's
connection to common usually needs a star topology to the chassis. When we
do that, there's no IZ drop due to shield current, so it sees the input signal
coupled in from the outside world just fine.
>Then of course comes the discussion of having the shield connected at both
>ends. Again, shield being correctly used as a shield, it should not be a
>current carrying conductor. If it doesn't carry current, regardless of the
>resistance there is no difference in potential thus it can be at 0
>potential. Run a microampere of current through it and it is no longer at 0
>potential. If ground is needed for the circuit, then a separate conductor
>must be used. Shield is connected at one end only thus to prevent current
>on the shield. Current flowing results in thus nullifying the effect of the
>shield.
True. But we insist on using unbalanced connections for ham gear and for
consumer gear. Ohm's law having not been repealed, we're stuck with having
that shield connected at both ends, with having some drop along it, and with
having that drop added to the signal.
If this were an audio list, I would also point out that the good reasons for
not
connecting the shield at the receiving end in balanced circuits are 1) that it
prevents shield current from exciting pin 1 problems; 2) it prevents the cable
from generating shield-current-induced noise; and 3) it prevents the balanced
input circuit from seeing the imbalance in the capacitance of the balanced
cable. :) #2 and #3 can be important with long cables. #1 can be a problem
any time. And obviously, we can't disconnect the shield when the connection
is unbalanced.
Jim
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|