Subject: Topic: Video Transmission and Receiving
Topic: Driving Unshielded Twisted Pair (UTP) Cabling
Figure 1 Typical Video Surveillance System
Video surveillance systems generally have a basic architecture as shown in Figure 1. There is
camera that sends a video signal that must be individually transmitted and received (usually
over a long cable) into some type of receiver/hub that equalizes (reconditions the signals) and
switches an array of these video signals (from other cameras) into a variety of monitors and/or
recording devices. Of course, inherit in these types of systems, is the cable that must be driven.
The parasitic characteristics of the cable (and driver amplifiers) between the camera and the
receiver/hub, as well as the cables from the receiver/hub to the monitors/recording devices
(and drive amplifiers), must be well understood in order to design a system for the desired
performance requirements.
Although there are various types of cables used in today’s high performance video surveillance
systems, the most basic (and commonly used) involves a UTP (unshielded twisted pair). An
unshielded twisted pair can be combined with other sets of pairs for various types of cable
groupings. For instance, 4 pairs of UTP are usually grouped to create a CAT5e cable (but a
CAT5e cable is not limited to 4 pairs either). An unshielded twisted pair forces a balanced pair
operation in which the two wires carry equal and opposite signals and the receiver detects the
difference between the two. This is simply known as differential mode transmission. Noise
sources introduce signals into the wires by the coupling of electric or magnetic fields (and in the
case of a twisted pair), and these fields couple to both wires equally! Therefore, the noise
should theoretically only produce a common-mode signal which is then cancelled/rejected at
the receiver when the difference signal is taken. Of course this method begins to fail when the
noise source is closer to one wire than the other (or if there is asymmetrical ringing on the
amplifier outputs), in which case, a common mode voltage difference between the two
develops, and the receiver begins to detect the difference. This is especially true over longer
and longer cable lengths.
Figure 2
UTP Cable Data Skew
So let’s take a look at some basic parasitic cabling issues that must be taken into account when
designing a video surveillance system. The first and most basic issue in driving UTP is data skew
(as shown in Figure 2). Data skew is simply the time difference in “propagation delay” between
the fastest and slowest set of wire pairs (that are carrying signals that are related and need
timing alignment). Of course data skew fundamentally stems from the simple physics of
propagation delay in a wire. An approximate propagation delay for a signal transmitted through
a given wire is around 6.7 psec/mm (this number is dependent on multiple factors including the
surface effect of the conductor (related to surface area) and current density (related to the
value of current through a given wire area). In general, propagation delay is a good number to
always keep track of in driving any high speed cabling system. Remember, even if the speed of
electricity approaches the speed of light, the propagation delay only approaches 3.33 psec/mm
(which equals a frequency of 300GHz for a wavelength of 1mm, 30GHz for a wavelength of
10mm, 3GHz for a wavelength of 100mm, and 300MHz for a frequency of wavelength of
1000mm (given frequency=velocity (speed of light)/wavelength)).
Propagation delay is a fundamental limit that must be taken into account in the physical layout
of ANY high speed system. Let’s take the case of driving a high speed video signal with a high
definition signal data rate of 180 MHz. The UTP cable driver amplifier (driving a differential 180
MHz video data rate) would need at least a bandwidth of approximately 3X180MHz=540 MHz
and a slew rate of nearly 1400 V/µsec in order to maintain the differential signal integrity
driving the input port of the UTP cable (to maximize common mode rejection). Utilizing a
CADEKA CLC4601 (a quad 550Mhz bandwidth with a slew rate of 1500 V/µsec) shown in Figure
3, the device allows the user to differentially drive a 4 pairs CAT5 UTP cable with just 2 quad
CLC4601 devices.
Figure 3 Differential CAT5e Cable Driver (One of Four Pairs of Drivers)
Let’s consider first the PCB layout of a differential cable driver device like the CLC4601 and look
at the effects of simple propagation delay on the performance of the device (see Figure 3). If
the approximate propagation delay for a wire (and let’s assume a resistor length has
approximately the same propagation delay) is 6.7 psec/mm; combined with a simple surface
mount resistor, along with a PCB trace length from the output of the amplifier to amplifier
negative input (which forms the negative feedback gain setting loop of the amplifier), with a
combined length of only 15 mm gives propagation delay of nearly 100 psec. An amplifier like
the CLC4601 with a 550MHz BW has a maximum operating wavelength (according to
wavelength=velocity (speed of light)/frequency (BW)) of 545 mm. The ¼ wavelength of ¼ X
545mm=136 mm is the physical distance/propagation delay in which a 550 MHz signal will shift
in phase by 90o. Simplistically speaking, a trace/resistor PCB layout length of only 15mm would
be (15mm/136mm) or 11% of 90o which is 10o. This means that the simple propagation delay of
100 psec for a 15 mm signal trace length yields a 10o phase shift in the negative feedback
network and thereby reduces the phase margin and overall stability of the amplifier by 10 o (out
of the theoretically stable amplifier phase margin target of 90 o). Usually, artificially reducing the
phase margin of any high frequency UTP cable driver amplifier by 10 o, by not taking into
account basic PCB propagation delay issues, will increase the likelihood of amplifier instability
problems in which “ringing” will occur on the amplifier output and asymmetrically cause
differential signal errors at the UTP cable input port and therefore premature signal
degradation that corrupts the signal significantly over even short cable lengths.
When designing a high performance video surveillance system, it is important to break down
the system into the various functional blocks that make up the system. Depending on the
overall system specification, such as the video data transmission rate and resolution, these
numbers will determine many of the required analog performance specifications of the system
including simple layout geometries. The number one passive component within a high
performance video surveillance system is the cable and understanding the positive and adverse
effects of this single component will greatly enhance your ability to design the system. Factors
such as data skew, insertion loss, return loss, crosstalk (near-end, far-end and power
summation) and DC loop resistance all affect the system in different ways and therefore must
be accounted for in the overall design. We will go over more of these factors next week and
how to design with these in mind in order to maximize performance.
Kai ge from CADEKA
(www.cadeka.com)