|
|
Shields Up!
Imagine water flowing through a thin walled rubber tube. A small part
of the water volume is "lost" as the pressure in the tube increases
because the tube "inflates" slightly with an increase in pressure. This
water is "recovered" when the pressure drops and the tube "deflates".
Likewise, when a current flows through a
wire a magnetic field "inflates" or expands around that wire. The field
is proportional to the current flowing through the wire. Unlike the
tube analogy, if the magnetic field crosses another conductor it
generates a voltage in that conductor and if the other conductor is
part of a circuit then a current starts to flow and some of the energy
from the first conductor is lost to the second conductor. Transformers
depend on, and are optimized for, this energy transfer. The ratio of
the induced voltage to rate of current change is called inductance. If
the current is changing rapidly then a significant portion of the
energy flowing through the wire is lost to this inductive resistance,
called impedance, and impedance is proportional to frequency.
When you measure the resistance of a wire with an ohm meter you are
measuring the resistance of that wire to a direct current flow that is
stable with respect to time. If you pass an alternating current through
that wire you will also have impedance to that alternating current.
That impedance is low at low frequencies and higher at high
frequencies.
Now imagine a wire conductor as a bundle tiny strands. The strands in
the center of the conductor have a lot of neighbors that want to steal
some energy but the strands on the surface only have less than half as
many thieving neighbors. The strands in the center therefore have a
higher AC impedance than the strands at the surface. This effect
becomes more pronounced the higher the frequency so that at high
frequencies the impedance at the center of the wire is much higher on
the surface. This "skin effect" means that at high frequencies almost
all of the energy flowing is following the path of least resistance,
which is on the surface, or skin, of the wire.
Cable shields act as a protective skin that intercepts the high
frequency energy and drains it to ground before it can get into, or out
of, the cable it surrounds. The shield's location and geometry make the
preferred path for high frequency energy. What is essential is that
there is a low resistance AND low impedance path from shield to ground.
A length of 22 gauge wire may have a low DC resistance but it may have
a high AC impedance such that, at high frequencies, if the shield is
connected to ground through a small wire then the shield loses most of
its effectiveness because its connection to ground has a big resistor
(the wire) in series.
The shield will work with only one end tied to ground because, like an
antenna, the circuit is partially copper and partially an oscillating
magnetic field. However, energy intercepted at the far end of the
shield must transverse the length of the cable, and all its DC
resistance, before finding ground. If there is a ground at both ends
then the energy only has to travel, at most, half as far, and there are
two parallel paths to take.
Now imagine you have a device that draws 3 amps at 14 volts and is
connected by a 24 gauge 25 foot power cable. The DC resistance of 25'
of 24 gauge wire is 0.66 ohms. Multiply this times 3 amps and the
voltage drop in the wire is 2 volts. 2 volts in the power wire AND 2
volts in the ground wire. This means that if you put a volt meter
between the aircraft ground and the case of the device you would see a
2 volt potential. Now imagine you need to pass a digital signal from
the device and that signal is referenced to ground. When the device
sets the signal to "low" it may be ground at the device but it will be
at 2 volts at the other end of the cable so the receiver may not
interpret the signal as "low". Remember that the signal path does not
have significant current flow so there is not the voltage losses that
are present in the power cable.
This difference in ground reference is the source of "ground loops"
which are unwanted currents flowing between devices that are not tied
to ground with a sufficiently low resistance connection. As Hamid said,
if DC current is flowing through the shield of a cable it is a symptom
of a problem elsewhere, not the source of that problem.
Twenty years ago the only high frequency devices in an airplane were
the radios. Now, with the introduction of computers with their high
frequency signals there are new challenges to keeping things
electrically "quiet". It does not surprise me that engineers who design
audio panels recommend terminating a shield at a pin. They are
concerned with audio frequencies only, so if you can't hear it it is
not important to them. The bottom line is that the better you can
connect all the grounds of all the chassis and all the shields the
better "damped" and electrically quieter the airplane will be.
Many engineers do not sufficiently understand or appreciate the issues.
For example, Hamid and I designed a tablet computer for Canadian
Marconi Electronics (CMA-1100 Pilot View) and as part of that effort,
took the system to a test facility to verify the electromagnetic
radiation of the system was within limits. The closest the system got
to busting the radiation limit was 3db (1/2 the allowed power) at about
150 MHz. When CMC repeated the test with a CMC built system they
reported their system was over the limit by more than 3 db, 4 times the
emissions we measured. Many phone calls and thousands of dollars in
travel and chamber time later we discover that CMC did not follow our
recommendation for terminating the power and serial communication
shields directly to the connector body at both ends. Rather, they
terminated using the CMC standard method of soldering the shield to a
22 gauge wire and then grounding the other end of the two inch long
wire. Replacing the cable with one that had properly terminated shields
fixed the problem. CMC did not believe this to be true until we did
side by side tests at an independent facility with a CMC engineer
present.
The above is a very brief review. I have omitted and simplified much
for the sake of brevity.
Regards
Brent Regan
|
|