Applying Surge Protection Devices on High-Frequency Data Systems
With the rising importance and complexity of data networks in today's competitive business climate, the concern of losing communications is steadily increasing. Companies with critical transactions, such as those found in the financial and industrial sectors, have tremendous downtime costs. As a result, these firms depend more heavily on the reliability of their data networks, which run on higher data rates.
Fast Ethernet is a good example. A typical industrial application for such a network is a highly automated manufacturing plant, where data lines generally run parallel to power equipment, creating a lot of electrical noise. The most damaging type of noise is transient voltage, where a voltage spike can reach thousands of volts in microseconds.
Even if the noise is directly coupled only in the power lines, the data lines are still affected due to field coupling effects, such as inductive and capacitive coupling. By exposing the data lines to a noisy environment, the susceptibility of the connected devices, such as switches, routers, and network cards, rises because of the higher integrity needs of the IT equipment. All of these circumstances lead to high-risk disruptions of data flow or even the destruction of vulnerable devices.
Protecting equipment against transient voltages in data applications gets even more challenging due to high frequencies and the requirement of lower power consumption. Let's take a look at some of the potential sources of transient influence in an industrial plant.
Sources and coupling of transients to data systems. A lightning strike close by or to an industrial facility itself — directly to the power line or even to ground — is only one source of transients. Motors, relays, inductive loads, contactors, and capacitor banks create voltage spikes in different amplitudes and with different energy content. Independent of the source of the transient and even if the transient affects the power, electromagnetic interference (EMI) can cause problems within the data system.
In any environment with a disturbance source, transient voltage will couple to receiving equipment. Three types of coupling exist: galvanic, inductive, and capacitive.
Direct/galvanic coupling. An example of how direct/galvanic coupling affects data lines is when lightning strikes directly to or close by a building or facility (Fig. 1). Due to the resistance of the conductors, the sudden high current to the grounding system leads to a rising of the ground potential itself, which influences the shield of data cables.
The signals on shielded twisted pair (STP) or unshielded twisted pair (UTP) cables are differential signals, and the rising ground potential can damage the susceptible equipment at the other end of the line. The surge protection device (SPD) will protect the susceptible equipment by clamping the voltage under a critical value.
Inductive coupling. This occurs when two conductors run parallel to each other. For example, the fast-changing current in a power line creates a magnetic field around it. If there is an adjacent conductor, which could be a data cable, the magnetic field also surrounds this wire and induces a voltage. All this happens almost instantaneously, and the amplitude of the induced voltage can reach up to several thousand volts.
The amplitude of the induced voltage can be calculated by the equation Vi=-L×(di/dt). Basically, this equation says that the induced voltage is defined by the inductance value of the wire (L) and the change in current (di) over time (dt). The effect of inductive coupling is shown in Fig 2.
To reduce the inductive coupling effect, you should use twisted pair cables. Twisting effectively reduces the circuit area to zero. Therefore, the induced voltage over the entire cable is reduced. In the real world, a perfectly symmetric cable (perfect twisting pitch throughout cable length) is next to impossible to achieve, so there remains the possibility of induced voltage.
In addition to coupling of a transient voltage, coupling between two adjacent signal lines is a major concern. Fast data rates are a challenging design criterion. The termination of the wires can also cause bigger problems. For example, to terminate an RJ45 connector, you have to untwist the end of the cable. The influence of the signal from one conductor to another now becomes significant.
Capacitive coupling. This always occurs via the electric field between two points with a high potential difference. A high load is generated in a conductive part or in operating equipment. An example is the lightning rod of a lightning arrester as a result of a direct strike. An electric field is created between the lightning rod and other parts with a lower potential, such as a conductor from the power supply or signal transmission within the building (Fig. 3). The potential difference between the two points strives to equalize, resulting in a charge transfer. This leads either to an increase in voltage or to a surge voltage in the affected conductor as well as the equipment connected to it.
This effect happens in other situations as well. For example, if internally created transients occur due to switching actions of inductive loads or even your office equipment, and you have data lines running parallel to the power lines, the potential difference between the two conductors can cause transients in the data lines.
SPDs in high-frequency data systems. To protect susceptible data system equipment against transients, you have to insert an SPD in the communication path, without affecting the reliability of the data network. The validity of this can be seen in the testing of an SPD to the same standards and with the same requirements as the data cable itself. The difficulty in meeting the same requirements of a cable is obvious, if you consider the basic function of an SPD.
In the case of a transient voltage between two signal lines (differential mode protection) or between one signal line and ground (common mode protection), the protective circuit provides an additional path for signals. The voltage at the output side of an SPD that is connected to the susceptible electronic device (switch, router or network card) is clamped to a safe value. The typical component used to bypass the energy is a silicon avalanche diode (SAD), which is basically a Zener diode with a wider p-n junction to handle the energy of the surge.
Every diode has a parasitic capacitance value, which is proportional to the amount of energy the diode can handle. A capacitance is an open circuit for DC; however, for higher frequencies, its impedance gets lower.
For the design of an appropriate SPD, two requirements have to be fulfilled. The first requirement of the device is to protect the equipment against transients. This requirement does not permit the use of diodes with a lower capability to carry the energy. The second requirement is not to interrupt the data flow.
By inserting a surge protection device that is not specifically designed to work in these higher-frequency applications, data signals can be attenuated in an unacceptable amount, and the signals between adjacent wires can disturb each other due to high crosstalk and reflections. The reason for the limitations of SPDs designed for 10 Mbits/sec Ethernet in Fast Ethernet applications is caused by both the components themselves and the circuitry of the SPD.
To achieve sufficient protection without influencing the signals, diode networks replace the use of single components between two signals. The change from a single component to a rectifier circuit reduces the capacitance value enough to fulfill the requirements specified as Cat. 5. For enhanced Cat. 5 (Cat. 5e) or Cat. 6, the reflection will still be too high. The most difficult parameter to achieve is the return loss, which characterizes the reflection behavior of the device. To reach return losses below the required values, two additional diodes have to be connected in series (Fig. 4).
SPD connector limitations. Even if the components and the protective circuit of the SPD fulfill the previously mentioned requirements, the manufacturer still has to consider the connection that inserts the SPD into the network. Besides the importance of the technical realization of a design, the manufacturer must consider and address the flexibility in real-world applications.
The most commonly used connection type in networks, such as Ethernet, Token Ring, or CDDI, is an RJ45 type of connecter. To use the protection device inline and with any cable length needed for different applications, most manufacturers use female RJ45 connectors for the input and output side of the SPD.
This circumstance limits the use of a device to networks with data rates lower than 100Mbits/sec. To make the connection to the RJ45 plug, the twisted pair shield is removed and the pairs untwisted. The wire pairs then run 1 cm to 2 cm parallel to each other, which causes significant capacitive coupling between the signal lines. Also, the near end crosstalk causes problems in communication.
The newer generation of RJ45 connectors fulfills the requirements for Cat. 5e and connectors tested to frequencies up to 250MHz. The improvement of the termination itself opens new possibilities to protected data applications with frequencies even higher than 100Mbit/s.
Another consideration for the SPD is packaging. If the application's sensitive device consists of only one or a few network connections, usually every single port is protected with a separate SPD. However, in installations such as office buildings, a switch has multiple Ethernet ports. Typically the IT equipment fits in 19-inch racks. For these applications, SPDs in the same form factor make the wiring clear and easy.
Stolzenberg is a product specialist with Phoenix Contact, Inc., in Harrisburg, PA.
Sidebar: Data Communication Gets Faster and Faster
With 200 million installed nodes worldwide, Ethernet is the most popular method of linking computers on a network. With its transmission rates in the 10Mbits/sec range, the requirements of the components used in SPDs are similar to simple serial data communication systems. You can connect the signal wires to the connector plugs without additional shielding, and crosstalk is most likely not a concern.
But Ethernet is too slow for very data-intensive or real-time applications. So, in the early 1990s, 100Mbit Ethernet (Fast Ethernet) was introduced, changing the requirements. Return loss and insertion loss are now significant concerns due to the higher-frequency parameters like near end crosstalk (NEXT).
A functional communication at higher frequencies can be only guaranteed if every component and device inserted in the signal path is designed for these frequencies. This includes devices such as connectors (RJ45), cables, and SPDs.
To address the rising requirements and guarantee interruption-free communication, standards organizations like the International Electrotechnical Committee (IEC) defined maximum values for attenuation, crosstalk, and other parameters. The requirements are defined in IEC 11801 (EN 50173-1). Specifications such as Cat. 5, Cat. 5e, and Cat. 6 are also commonly used to describe the ability of twisted pair cables to be used in data applications with higher frequencies.
