The following letter is in response to a question in Ask Donna (April 1996, page 57) submitted by Scott Conti, University of Massachusetts (Amherst, MA), regarding rewiring "organic networks."
Pyramid Design and its parent company Alpine Computer were commissioned by the University of Massachusetts Office of Information Technologies to verify that two 10Base-T network segments could operate without errors over the existing horizontal station wiring in the Academic/Administration and the Residence Halls. These additional tests were required by the University`s network staff to confirm our recommendation to continue reusing existing wiring for the campus-wide 10Base-T network project. (Financial constraints prevent rewiring the entire campus horizontal cabling to implement 10Base-T in the near future.)
We used three network general protocol analyzer/traffic generators, a local area network (LAN) analyzer and cable tester to perform these tests, which were witnessed by several representatives from the UMass network study committee. The tests were considered a success, because the only error recorded was one bad frame checksum (FCS). This occurred when we changed ports to transmit in the opposite direction; therefore, the error did not apply to the performance evaluation. We recorded the following error/collision parameters: jabbers, bad FCS, short frames, late collisions, remote collisions, local collisions and ghosts.
Because of the concern about the prior testing of cables for Category 3 versus 10Base-T parameters on the LAN cable tester, we also performed 10Base-T autotests on each cable segment. They passed all test criteria.
We have successfully designed and installed dual 10Base-T ports over shared-sheath wiring schemes for several large networks, and UMass also has active prototype installations working at Tobin, Whitmore, Goodell and Munson Halls. Based on the IEEE 802.3 standard and the positive test results, we stand by our recommendation and believe that there are no technical reasons to prevent using existing 6-pair and 8-pair horizontal unshielded-twisted-pair (UTP) cables at the UMass Amherst campus for implementation of 10Base-T (10-Mbit/sec) network connectivity.
Tom McNamara, RCDD
Pyramid Design Inc.
Donna Ballast, communications analyst at the University of Texas at Austin and author of Ask Donna, replies:
You have proven that the pair combinations you chose in the existing cabling can support two 10Base-T channels in the same sheath. However, given Conti`s description of splitting and terminating the 8-pair sheath on four outlet/connectors, and your test data "proving" that the cabling can support two 10Base-T channels in the same sheath, there are still three other outlet/connectors.
The TIA/EIA-568A standard, the National Electrical Manufacturers Association (NEMA) premises wiring standard and the Underwriters` Laboratories performance-verification program all require that Category 3, 4 and 5 UTP cable with more than four pairs be tested using the power-sum method. (Power sum is a method of testing crosstalk as if all pairs were active.) The total coupling onto one pair from all other pairs could be excessive and cause loss of data integrity. The tester you are using tests pair-to-pair at any given time for the four pairs you have connected.