Problems with Cable Shields
When chasing down an EMI problem, we find the I/O cables
are often at the root, whether we are fighting radiated emissions,
RF immunity or electrostatic discharge. When called in on
a problem the first thing we look at is the cable is
it shielded and how well is the cable terminated to the housing?
All too often we find the termination less than satisfactory.
Some of these problems can be identified by inspection, others
are less readily identified. The following paragraphs discuss
some of the problems we have seen.
Lets assume the cable is shielded (if not shielded,
then you will need filtering, but thats another story).
Check both the cable connector and the mating bulkhead connector
(see Figure 1). The bulkhead connector should conductively
mate to the enclosure bulkhead around the entire perimeter.
Be alert for non-conductive coatings. In many cases, the connector
will only mate to the enclosure at the two ends, at the mounting
screws. This may be adequate for moderate shielding needs,
but a better solution is to use connector gaskets so that
a connection is made around the entire perimeter.
Then turn to the cable end and make a similar inspection.
The connector should conductively mate around the perimeter
of the cable shield. Pigtail terminations are unacceptable
for any frequencies above audio.
Thats it for the visual inspection. Unfortunately, that
may be inadequate.
If your connector can be disassembled, you can take it apart
and do an internal inspection. What may appear to be a circumferential
wrap, may turn out to be something else. A circumferential
wrap can only be achieved using a compression fitting in the
connector, to clamp around the cable shield, and these are
not common outside the military arena. Usually, the termination
will be to a screw post inside the connector shell, hopefully
with a short lead. This is definitely not optimum, but if
your shielding needs are minimal, it may be good enough.
Worse, the shield may be connected to a connector pin, via
the drain wire. This is almost always unacceptable
the wire length from the cable shield through the connector
to wherever the pin is grounded inside is too inductive to
serve any purpose above audio frequencies. Worse, it will
form the basis for an LC circuit termination, turning the
cable into an antenna at resonance.
Unfortunately, visual inspection has some decided limitations.
Here are some problems you wont see:
First and foremost, the shield may be single-point grounded,
meaning it is grounded at one end only. This is almost always
the case with cables bought from your nearby computer supply
store. Cables that are grounded at both ends are so labeled
on the package. If not labeled, assume it is a single-point
You can check for a single-point ground with a DVM
look for continuity, or lack thereof, from connector shell
to connector shell. If there is no continuity, then you have
a single point ground or, worse yet, no ground. Yes, we have
found shielded cables without a termination at either end
but you cant distinguish these from a single-point
ground with a DVM.
If you do find the shield terminated at both ends, you still
wont know how the termination is effected it
will almost assuredly have a drain wire termination somewhere
in the connector shell.
Well, unless you are willing to destroy the cable by cutting
it open, you will need to resort to some fancy instruments
that most people dont have (the one we liked best was
an X-ray it clearly showed the construction). Or, lacking
that, you can resort to building an overshield to try it out.
If you want to find out if a cable shield is really working,
the fastest way is to build up a shield around the existing
cable. Start with a length of aluminum foil, perhaps three
feet long. Fold it double for strength, then place the cable
in the middle of the foil strip and fold over again, then
roll up the foil and tape it closed. Then take another segment
and do the same, overlapping the first segment. Using relatively
short lengths simplifies the fabrication process and allows
some additional flexibility. But the final product will not
be strong, so handle the cable gently.
Use copper tape to close up the adjacent foil segments. Extend
the foil right up to the connector shell and copper tape it
around the perimeter at both ends. This will be difficult
if the shell is plastic you will probably have to mount
the cable, then tape it shut.
If you have a round connector, you can substitute hose clamps
for the copper tape to get a more robust mount.
Done carefully, the makeshift shield should be good enough
to prove a point if there is no improvement, then the
shield is not the problem or you have not built the cable
For troubleshooting on the test floor, this should help identify
the problem cables. Obviously, they will need to be replaced
with a production shielded cable.
This approach can hardly be considered permanent, but it can
last quite a while if it is not exposed to potential damage.
If you have a problem in the field, you can either replace
the suspect cable with a better cable, or you can permanently
shield the existing cable. Zippertubing® is one good way
of shielding existing cables you place the cable inside,
then zip the shield shut. It is available in various widths
to accommodate different cable diameters or even several cables.
It also has provision for making a reasonable ground termination
at the cable ends.
If your cable is small enough, you may be able to get an overbraid
over the connector and onto the cable.
One more point if, after you have made sure you have
a good shield and the cable still seems to leak, look for
coupling directly to the cable. These are especially common
from seams near the cable at the bulkhead. Such coupling will
be a problem no matter how good the cable shield is.
Finally, once you have established that the cable is actually
built like you want it to be, you still arent out of
the woods. Our experience is that cable shields are very vulnerable
to undetectable damage. Here are two conditions we have encountered:
First, the cable termination breaks inside at or near the
connector shell. This is especially true when the drain wire
is terminated inside. Second, the shield ruptures somewhere
along the cable. An example, most commercial shields use Mylar®
foil, which ruptures quite easily. A keyboard cable with a
coiled up segment is especially vulnerable just pull
the coils out to their full extension and you have ruptured
the cable, seriously degrading the shielding effectiveness.
When chasing down an EMI problem, look first at the cable
shield, especially the termination. If you are satisfied with
the termination, there are still several other possibilities
for cable shield failure, including single point grounds,
hidden poor terminations and shield damage.
Temporary shields can be quickly fabricated to establish the
effectiveness of an existing cable shield, or to establish
the need for a cable shield.
The bottom line is, dont assume the cable shield is
doing what you want it to check it out and make sure.