Given the frequency of moves and changes in today's workplace, building owners require flexible wire and cable management systems that provide easy access to power, voice, and data lines at every workstation.

A popular solution is to locate power and data cabling in multichannel raceway systems. However, in a typical commercial building, erratic fluctuations of load conditions subject power lines to electrical fast transient (EFT) disturbances that are generated when inductive-capacitive circuits are interrupted. These disturbances can affect the integrity of the data cabling located in the same raceway.

Of particular concern is EFT disturbances most often caused by transient currents (commonly called arcing) during a make or break of contact. Another culprit is sudden changes in the magnitude and direction of currents in everyday office equipment, such as copiers, pencil sharpeners, and power switches. What makes this phenomenon of particular significance to data transmission is that the frequency spectrum of the transient is in the vicinity of the frequency band within which information-carrying signals of high-speed local area network (LAN) protocols (100 MHz and higher) are transmitted on copper media. In such a "noisy" environment disturbances have the potential to degrade network performance resulting in increased network errors and response time.

Previous research conducted by The Wiremold Co. and others concluded that 10BASE-T (10 Mbps) Ethernet performance was not affected by locating power wiring and unshielded twisted pair (UTP) data cabling in the same raceway. However, researchers expressed concern that EFT disturbances could have a greater impact on the performance of higher speed data transmission protocols, such as 100 Mbps Fast Ethernet.

A new series of tests evaluated the effect of noise, induced from power cables to UTP data cabling within the same raceway, on Fast Ethernet performance. These tests determine that the degree of physical separation of power and data cabling mattered in high-speed network performance.

The test bed was constructed of 90 m of nonmetallic (plastic) three-compartment raceway. Within the two data compartments, Category 5 UTP cable was affixed to dividers or compartment walls to achieve separations of 0, 1, 2, and 3 inches. The presence of a raceway divider between the power compartment and the adjacent data compartment resulted in an actual separation of approximately 1/8 inch (referred to in the test results as 0 inch).

An EFT generator was chosen as a controlled means for introducing a defined electrical disturbance into the power line. The EFT generator output was applied at the near end to a single-phase, three-loose conductor, 20-A-rated unshielded active power line. The far end of the power line was terminated with a 7-A resistive load.

A Fast Ethernet LAN was assembled consisting of one hub and two PC-based workstations, each with a Network Interface Card (NIC). A hub is a device that serves as the center of a star-topology network. It takes any incoming signal and repeats it out to all ports.

Two methods were used to evaluate Fast Ethernet performance. The first was to measure the number of cyclic redundancy check (CRC) errors. The CRC is a method of error detection used when transmitting packets of data from station to station.

For this test, both workstations were connected to the hub via a short patch cord. Furthermore, a network protocol analyzer was attached to the hub through 90 m of Cat 5 data cable located in the raceway.

To generate traffic, large program files (66MB) were transferred between the workstations. Since the workstations were located outside the "noisy" environment, the exchange of data was not affected by EFT disturbances.

At 0 inch separation, the number of errors stays at a constant level (essentially zero) until the EFT generator voltage reaches about 700 V. This indicates that any EFT disturbances up to this level will not reduce Fast Ethernet performance.

The second method used to check Fast Ethernet performance was measuring the file transfer time. In this test, one of the workstations was connected directly to the hub, thus simulating a server. The other workstation was connected to the hub via the "noisy" link, thus simulating a client in the work area. The file transfer time test results follow the same trend as the CRC error test data. Regardless of the separation between power and data lines, the file transfer time remains constant at about 50 seconds up to the 700 V level.

The CRC error test shows that in a nonmetallic raceway system EFT noise from typical office disturbers induced from power to data cables have no detectable impact on the performance of Fast Ethernet running on unshielded copper media.

The file transfer time test provided the most valuable information for the LAN end user. It indicates that in a typical office environment, multichannel plastic raceway systems can be used to conveniently route power and data cables to the point of use without affecting the speed or quality of data transmissions over high-speed UTP LANs.

The test results illustrate that the physical separation provided by the 1/8-inch raceway divider is sufficient to achieve satisfactory LAN performance in the presence of EFT amplitudes from typical office disturbers like pencil sharpeners, copiers, and light switches. There is no evidence that separation between power lines and Fast Ethernet Category 5 data cables is required.