Transmission Techniques: Wire and Cable 429frequencies, this means that even a correctly terminated
telescopic shield is less effective at RF frequencies.14.17 UTP and AudioOne other solution for ground loops is to have no ground
connection. For the seasoned audio and video profes-
sional, this solution may require a leap of faith. In can
clearly be seen that, with a cable that has no shield, no
drain wire, and no ground wire, so no ground loop can
develop. This is a common form of data cable called
UTP, unshielded twisted pairs.
With such a cable, having no shield means that you
are totally dependent on the balanced line to reject
noise. This is especially true, where you wish to use the
four pairs in a Cat 5e, 6, or 6a cable to run four unre-
lated audio channels. Tests were performed on
low-performance (stranded) Cat 5e patch cable (Belden
1752A) looking at crosstalk between the pairs. This test
shows the average of all possible pair combinations, the
worst possible case, and covered a bandwidth of 1 kHz
to 50 kHz. The results are shown in Fig. 14-16.You will note that the worst case is around 40 kHz
where the crosstalk is slightly better than 95 dB. In the
range of common audible frequencies (20 kHz) the
pair-to-pair crosstalk approaches 100 dB. Since a noise
floor of 90 dB is today considered wholly acceptable, a
measurement of 95 dB or 100 dB is even better still.
A number of data engineers questioned these
numbers based on the fact that these measurements were
FEXT, far-end crosstalk, where the signals are weakest
in such a cable. So measurements were also taken of
NEXT, near-end crosstalk, where the signals are stron-
gest. Those measurements are shown in Fig. 14-17.
The NEXT measurements are even better than the
previous FEXT measurements. In this case, the worst
case is exactly 95 dB at just under 50 kHz. At 20 kHz
and below, the numbers are even better than the
previous graph, around 100 dB or better.
There were attempts made to test a much better cable
(Belden 1872A MediaTwist). This unshielded
twisted-pair cable is now a Cat 6 bonded-pair design.
After weeks of effort, it was determined that the
pair-to-pair crosstalk could not be read on an Agilent
8714ES network analyzer. The crosstalk was some-
where below the noise floor of the test gear. The noise
floor of that instrument is 110 dB. With a good cable,
the crosstalk is somewhere below 110 dB.14.17.1 So Why Shields?These experiments with unshielded cable beg the ques-
tion, why have a shield? In fact, the answer is somewhat
reversed. The pairs in data cables are dramatically
improved over the historic audio pairs. The bandwidth
alone, 500 MHz for Cat 6a, for instance, indicates that
these are not the same old pairs but something different.
In fact, what has happened is that the wire and cable
(and data) industries have fixed the pairs.
Before, with a poorly manufactured pair, a shield
would help prevent signals from getting into, or leaking
out of, a pair. The fact that either effect, ingress or
egress, occurred indicated the poor balance, the poor
performance of the pair.
This does not mean shields are dead. There are data
cables with overall shields (FTP), even individually
shielded pairs (Cat 7) common in Europe. However,
these are subject to the same problems as all shielded,
grounded cables in terms of ground loops and wave-
length effects as shown in Sections 14.8.6.5 and
14.8.6.6.
The truth to the efficacy of unshielded twisted pairs
running audio, video, data and many other signals is
commonplace today. Many audio devices routinely useFigure 14-16. Crosstalk between Cat 5e patch cable.Frequency–Hz
85
90
95
100
105
110
115dB(^1000124414881732197621732365255827502942319434723750404344784913534857836294688275338125854289589900) 12,00013,12514,42915,87517,25018,20020,20023,25025,50028,12530,00034,00037,66741,40045,00050,500
Figure 14-17. NEXT crosstalk.
Frequency–Hz
dB
90
95
100
105
110
115
(^1000131716341951221224622712296233063667404346905174573964127200804285839300) 11,33313,12514,85716,83318,00020,60024,40027,66730,66735,40040,200 45 ,^000