Good planning and careful selection of transient voltage-
suppression devices can ensure adequate circuit protection
from electrostatic discharge (ESD) and cable discharge events.
Designing a system for both high-speed communication and
transient immunity requirements is nontrivial. Newer Ethernet
transceivers run faster, consume less power and use less PCB
space. But these advances have contributed to a reduction of
on-chip transient-voltage protection levels. Thus, designers need
advanced system-level circuit protection to ensure Ethernet
systems remain immune to ESD and cable discharge threats.

CDE is a real and frequent phenomenon in the Ethernet
environment. Moreover, while cable-discharge (CDE) can be
thought of as a type of electrostatic discharge (ESD), designers
should treat CDEs as a separate type of transient event. An
Ethernet cable -- generally unshielded, twisted-pair Cat-5 or Cat-6
can be simply modeled as a capacitive element that can store
a significant charge. That cable, which can run as long as 100m,
can accumulate charge via triboelectric or induction effects. Simply
dragging a cable along a carpet or removing it from a package will
lead to a stored charge. Inductive transfer from a user also can
charge a cable. Because Cat-5 and Cat-6 twisted pair cables have
low-leakage properties, the charge may remain stored on a twisted
pair for several hours and it can discharge into an Ethernet port
when a user connects it to equipment. The latter type of discharge
occurs directly into the communication interface and poses a
particularly dangerous threat to the communication interface such
as Ethernet ports. The high peak voltage and current during a CDE
can overstress an Ethernet transceiver and lead to intermittent
malfunctions or total failure.

The semiconductor industry has recognized the need for a standard
method for testing CDE and Working Group 14, ESD Simulators,
within the Electrostatic Discharge Association (ESDA) is currently
defining a standard method for CDE testing. This work will define a
testing method that uses an ESD waveform specified in IEC 61000-
4-2, but the new method will account for energy transfer through
a cable rather than a human body. Unlike a human-body-model
ESD, CDE has an initial current spike followed by a characteristic
plateau and then a ringing signal with rapid polarity changes. In
many cases, cable discharges can deliver more damaging energy
to CMOS structures that can human-body-model ESD. Thus it is
essential for Ethernet ports to add good system-level protection
circuits. Unfortunately, some protection circuits negatively affect
signal integrity and others offer inadequate protection. We
recommend engineers consider the following characteristics when
they review Ethernet-port protection needs:

· Fast response time
· Low clamping voltage
· Low leakage current
· Low capacitance
· High energy handling capacity
· Optimal PCB layout

First, an effective Ethernet-protection device must offer a response
time faster than the transient events the system will experience.
Thus to safely attenuate a fast discharge during ESDs and
CDEs, the protection device must respond within hundreds of
picoseconds, faster than the ESD rise time. Figure 1 shows an
example of a protection circuit scheme with a sub-nanosecond
response time. Placing the protection devices "behind" the
Ethernet transformer further reduces surges.

Second, a well devised protection circuit must provide a low
clamping voltage for a transient pulse. A transient voltage
suppressor (TVS) such as the RClamp2504N (Shown in Figure 1)
diode array offers a 4V clamping voltage (Vc) for a peak pulse
current (Ipp) of 1A. Its Vc increases linearly to about 10V for an Ipp
of 25A. This type of low-voltage clamping response provides a large
protection margin for an Ethernet transceiver.

Applications