Avoid costly network downtime--use predictive cable diagnostics

The high date rate, scalability, and packet structure of Ethernet is leading to widespread adoption in applications well beyond original target markets. A primary attraction in industrial applications (e.g. factory automation, automotive, test and measurement, building automation, security and surveillance) is that Ethernet can replace multiple interfaces, combining data, signal and control paths. Since Cat5 is a cost-effective media, this dramatically reduces cabling costs, connector footprint, and recurring cost of maintenance. In these applications the Ethernet interface becomes part of the core functionality of the system and performance, ruggedness and reliability cannot be sacrificed for price. The environments in which these products operate also impose temperature constraints and harsh signal conditions requiring very high component quality and reliability. These fixed application needs and challenges must be addressed by the system integrator.

The sophistication of diagnostic capability required varies significantly with the application. Simple diagnostic tools adequately service the needs of home or office computer networking, with usually only a status message or LED to indicate that the link is active. Unfortunately for these users, the only maintenance option available is to swap a cable or re-boot the PC. For enterprise networking, installation, provisioning, and monitoring tools are more sophisticated. However, the focus is usually on responding quickly to a system failure. For safety critical applications, or where the cost of a connection failure is high, the emphasis has to be on providing enough diagnostic information and system monitoring to avoid any plant failures. For example, a single component failure in a single critical link can be responsible for shutting down an automobile assembly line where every second of lost production is measured in thousands of dollars.

The focus of this article is advanced cable diagnostics for reliable connectivity in an industrial environment. In particular, some novel link monitoring techniques implemented in an industrial Ethernet transceiver are described. It is shown how these features reduce the total cost of ownership and increase the quality and reliability of critical industrial networks.

Failure modes
Historically, diagnostic tools for networks focused on initial testing and isolating issues during installation of the equipment and cabling. They also typically help with fault isolation when an error occurs. This is appropriate for traditional networking installations where the equipment is isolated in server rooms or equipment racks and the cabling is protected in wiring closets or on cable trays. By far, the most common failure mechanism in this environment is an open or short circuit in a cable. For example, an installer may inadvertently disconnect one cable while installing another or the pairs in a cable may be shorted when a heavy load is placed on a cable bundle. Since the numbers of connections run into the hundreds in even the simplest of installations, it can be a nightmare isolating which link has been affected and where along the cable the fault has occurred. For this reason cable testers have been developed that utilize a mechanism called time-domain reflectometry (TDR).

Time-domain reflectometry
TDR relies on the electromagnetic properties of waves along a transmission line. In simple terms, a pulse of known amplitude is launched into the cable, and a reflection will occur unless the impedance of the load exactly matches the characteristic impedance of the cable. The type and location of the fault is determined by measuring the response. As is evident from the following equation, the biggest reflection coefficient will occur with either an open circuit or short circuit failure.

The polarity of the pulse and the time it takes to return to the sending node is all that is needed to isolate the type and location of the failure. Some examples of common wiring failures are shown in Figures 1-3.

Page 2: next page
Page 1 | 2